Home > TKK Dissertations 238

TKK Dissertations 238

TKK Dissertations 238
Espoo 2010
MICROFABRICATION OF HEATED NEBULIZER CHIPS
FOR MASS SPECTROMETRY
Doctoral Dissertation
Ville Saarela
Aalto University
School of Science and Technology
Faculty of Electronics, Communications and Automation
Department of Micro and Nanosciences

TKK Dissertations 238
Espoo 2010
MICROFABRICATION OF HEATED NEBULIZER CHIPS
FOR MASS SPECTROMETRY
Doctoral Dissertation
Ville Saarela
Doctoral dissertation for the degree of Doctor of Science in Technology to be presented with due
permission of the Faculty of Electronics, Communications and Automation for public examination
and debate in Large Seminar Hall of Micronova at the Aalto University School of Science and
Technology (Espoo, Finland) on the 15th of October 2010 at 12 noon.
Aalto University
School of Science and Technology
Faculty of Electronics, Communications and Automation
Department of Micro and Nanosciences
Aalto-yliopisto
Teknillinen korkeakoulu
Elektroniikan, tietoliikenteen ja automaation tiedekunta
Mikro- ja nanotekniikan laitos
Distribution:
Aalto University
School of Science and Technology
Faculty of Electronics, Communications and Automation
Department of Micro and Nanosciences
P.O. Box 13500 (Tietotie 3)
FI - 00076 Aalto
FINLAND
URL: http://nano.tkk.fi/
Tel. +358-50-304 9085
Fax +358-9-470 26080
E-mail: ville.saarela@tkk.fi
? 2010 Ville Saarela
ISBN 978-952-60-3339-6
ISBN 978-952-60-3340-2 (PDF)
ISSN 1795-2239
ISSN 1795-4584 (PDF)
URL: http://lib.tkk.fi/Diss/2010/isbn9789526033402/
TKK-DISS-2802
Aalto-Print
Helsinki 2010
ABSTRACT OF DOCTORAL DISSERTATION AALTO UNIVERSITY
SCHOOL OF SCIENCE AND TECHNOLOGY
P.O. BOX 11000, FI-00076 AALTO
http://www.aalto.fi
Author Ville Saarela
Name of the dissertation
Microfabrication of heated nebulizer chips for mass spectrometry
Manuscript submitted 18.6.2010
Date of the defence 15.10.2010
Monograph Article dissertation (summary + original articles)
Faculty Faculty of Electronics, Communication and Automation
Department Department of Micro and Nanosciences
Field of research Semiconductor Technology
Opponent Prof. Klaus Bo Mogensen
Supervisor Prof. Pekka Kuivalainen
Instructor Prof. Sami Franssila
Abstract
Microfabrication technologies originating from the semiconductor industry were applied to the instrumentation
of analytical chemistry. Heated nebulizer (HN) chips made of silicon and glass were developed. The HN chips
are used to vaporize a sample prior to detection by a mass spectrometer. The chips can be used with both
liquid and gaseous samples and they are compatible with multiple atmospheric pressure ionization techniques,
which enables wide applicability with different separation methods and various types of analytes. Better
sensitivity, flexibility and operation with a lower sample and nebulizer gas flow rates was achieved by the
miniaturization of the heated nebulizer. The chips can operate with 50 nL min?1 to 5 ?L min?1 sample flow rates
typical of microfluidic separation systems.
Silicon and glass microfabrication methods ? etching, wafer bonding and thin film technology ? were
developed and applied to the fabrication of the HN chips in 40 different layout and process variations. The
thermal behaviour of the chips and the shape of the gaseous jet produced by the chips was studied. A method
was developed for measuring the temperature distribution of a gaseous jet using a miniature thermocouple
attached to a computer controlled xyz stage.
Different methods for making capillary tube and electrical interconnections to the chips were also studied.
Liquid chromatography (LC) column chips were developed resulting in an integrated chip having both an LC
column and a heated nebulizer on a single chip. At the end of the LC column there is a high aspect ratio
micropillar frit which enables packing the column with particles.
The novel chips developed in this work extend the available ionization methods and the range of suitable
analytes compared to the previously presented chips for mass spectrometry.
Keywords chip; heater; liquid chromatography; mass spectrometry; microfabrication; nebulizer
ISBN (printed) 978-952-60-3339-6 ISSN (printed) 1795-2239
ISBN (pdf) 978-952-60-3340-2 ISSN (pdf) 1795-4584
Language English Number of pages 88 + 49
Publisher Department of Micro and Nanosciences
Print distribution Department of Micro and Nanosciences
The dissertation can be read at http://lib.tkk.fi/Diss/2010/isbn9789526033402/

V?IT?SKIRJAN TIIVISTELM? AALTO-YLIOPISTO
TEKNILLINEN KORKEAKOULU
PL 11000, 00076 AALTO
http://www.aalto.fi
Tekij? Ville Saarela
V?it?skirjan nimi
Mikrovalmistettuja kuumasumutussiruja massaspektrometriaan
K?sikirjoituksen p?iv?m??r? 18.6.2010
V?it?stilaisuuden ajankohta 15.10.2010
Monografia Yhdistelm?v?it?skirja (yhteenveto + erillisartikkelit)
Tiedekunta Elektroniikan, tietoliikenteen ja automaation tiedekunta
Laitos Mikro- ja nanotekniikan laitos
Tutkimusala Puolijohdeteknologia
Vastav?itt?j? Prof. Klaus Bo Mogensen
Ty?n valvoja Prof. Pekka Kuivalainen
Ty?n ohjaaja Prof. Sami Franssila
Tiivistelm?
Puolijohdeteollisuudesta l?ht?isin olevia mikrovalmistustekniikoita sovellettiin analyyttisen kemian
laitetekniikkaan. V?it?skirja k?sittelee lasista ja piist? valmistettuja kuumasumutussiruja, joita k?ytet??n
n?ytteen h?yryst?miseen ennen massaspektrometrista analyysi?. Kuumasumutussiruja voidaan k?ytt?? sek?
neste- ett? kaasumaisten n?ytteiden kanssa ja ne soveltuvat useille eri normaalissa ilmanpaineessa teht?ville
ionisaatiotekniikoille, mik? mahdollistaa sirujen laajan sovellettavuuden erilaisten erotusmenetelmien ja
analyyttien kanssa. Kuumasumutuksen miniatyrisoinnilla saavutetaan aiempaa parempi herkkyys,
monik?ytt?isyys ja tarvittavat n?ytteen ja sumutuskaasun virtausnopeudet ovat pienemm?t. Sirut soveltuvat
mikrofluidistiikassa tyypilliselle n?ytevirtauksille v?lill? 50 nL/min ? 5 ?L/min.
Ty?ss? kehitettiin ja sovellettiin piin ja lasin etsausmenetelmi?, kiekkosubstraattien liitosmenetelmi? ja
ohutkalvotekniikkaa 40 erilaisen kuumasumutussirun valmistamiseen. Lis?ksi tutkittiin valmistettujen sirujen
l?mp?ilmi?it? ja siruista l?htev?n kaasusuihkun muotoa. Kaasusuihkun l?mp?jakauman mittaamiseen
kehitettiin menetelm?, joka perustuu tietokoneohjattuun xyz-p?yt??n kiinnitettyyn pienoistermoparianturiin.
Lis?ksi tutkittiin erilaisia tapoja tehd? siruille tarvittavat kapillaariputkiliit?nn?t ja integroidun l?mmittimen
s?hk?iset kontaktit.
Ty?ss? kehitettiin nestekromatografiakolonnisiruja ja yhdistettiin ensimm?ist? kertaa nestekromatografiakolonni
ja kuumasumutus samalle sirulle. Nestekromatografiasiruissa on kolonnikanavan loppup??ss? suodattimena
toimiva mikropilaririvist?, mik? mahdollistaa kolonnin pakkaamisen kiinte?faasimateriaalipartikkeleilla.
Ty?ss? kehitetyt uudet sirut lis??v?t k?ytett?viss? olevia ionisaatiotekniikoita ja tutkittavissa olevia analyyttej?
verrattuna aiemmin julkaistuihin massaspektrometriasiruihin.
Asiasanat l?mmitin; massaspektrometria; siru; mikrovalmistus; nestekromatografia; sumutin
ISBN (painettu) 978-952-60-3339-6 ISSN (painettu) 1795-2239
ISBN (pdf) 978-952-60-3340-2 ISSN (pdf) 1795-4584
Kieli Englanti Sivum??r? 88 + 49
Julkaisija Mikro ja nanotekniikan laitos
Painetun v?it?skirjan jakelu Mikro- ja nanotekniikan laitos
Luettavissa verkossa osoitteessa http://lib.tkk.fi/Diss/2010/isbn9789526033402/

vii
Preface
In May 2003 after two years of studies in the Degree Programme in Electronics and
Electrical Engineering at Helsinki University of Technology I was hired as a summer
trainee at the Microelectronics Centre that was in charge of the university's newly built
cleanroom facilities at Micronova. During that summer I got fascinated by the world of
microfabrication and focused my subsequent studies towards the field. I also continued as
an employee in the research group lead by Prof. Sami Franssila where I still reside after
several organizational changes and the university's transformation into Aalto University. I
have enjoyed becoming familiar with the multidisciplinary world of microfluidics and
putting the theory into practice. I have learned a lot, but at the same time I have realized
that there is so much more to explore. Well, I suppose that is the very nature of science ?
the work is never truly completed!
I wish to express my warmest thanks to all my co-workers here at Micronova ?
especially to Prof. Sami Franssila and Prof. Pekka Kuivalainen for instructing and
supervising my dissertation, respectively. Thanks to the pre-examiners: Dr. Ciprian Iliescu
and Prof. Levent Yobas.
This work would not have been possible without the collaboration with the groups of
Prof. Risto Kostiainen and Prof. Tapio Kotiaho at the University of Helsinki where the chips
I fabricated were put into use with mass spectrometers. Big thanks to all the co-authors of
my publications ? two of which are shared in the thesis of Dr. Markus Haapala who
provided me with plenty of valuable feedback and discussion about improving the chips.
Thanks to Dr. Kai Kolari at VTT for the DRIE of silicon and glass. Thanks to William Martin
for proofreading the final version of the manuscript.
In summer 2007 I had the opportunity to get two months of hands on experience with
micro powder blasting at EPFL, Switzerland. Thanks to Dr. Abdeljalil Sayah for supervising
my work during my stay and all the others in Prof. Martin Gijs's group for the hospitality.
I am grateful for the funding and grants provided by CHEMSEM graduate school,
TEKES ? the Finnish Funding Agency for Technology and Innovation, the Academy of
Finland, TES ? the Finnish Foundation for Technology Promotion, the Walter Alhstr?m
Foundation and the Emil Aaltonen Foundation.
Last, but certainly not least, I am deeply indebted to my family and friends for the love,
support and joy you have given me. I love you!
In Espoo, September 2010,
Ville Saarela

ix
Contents
Abstract of doctoral dissertation iii
V?it?skirjan tiivistelm? v
Preface vii
List of publications xii
Author's contribution xiii
Abbreviations and symbols xiv
1 Introduction 1
1.1 Background...........................................................................................................................1
1.1.1 Microtechnology.....................................................................................................1
1.1.2 Analytical chemistry and mass spectrometry....................................................2
1.1.3 Microfluidics, ?TAS, BioMEMS and Lab-on-a-Chip.........................................3
1.2 Microchips for mass spectrometry....................................................................................5
1.3 Aim of research....................................................................................................................6
1.4 Summary of publications....................................................................................................6
1.4.1 Publication I: Re-usable multi-inlet PDMS fluidic connector..........................6
1.4.2 Publication II: Glass microfabricated nebulizer chip for mass spectrometry
....................................................................................................................................7
1.4.3 Publication III: Deep plasma etching of glass for fluidic devices with
different mask materials........................................................................................8
1.4.4 Publication IV: Microfluidic heated gas jet shape analysis by temperature
scanning....................................................................................................................8
1.4.5 Publication V: On-chip liquid chromatography ? atmospheric pressure
ionization ? mass spectrometry............................................................................9
1.4.6 Publication VI: Integrated liquid chromatography ? heated nebulizer
microchip for mass spectrometry.........................................................................9
2 Microfabrication for fluidics 13
2.1 Materials..............................................................................................................................13
2.1.1 Substrate materials...............................................................................................13
x
2.1.2 Thin-film materials and deposition...................................................................15
2.1.3 Diffusion and high temperature processing....................................................16
2.2 Patterning and micromachining......................................................................................17
2.2.1 Lithography...........................................................................................................17
2.2.2 Etching....................................................................................................................18
2.2.3 Lift-off patterning and shadow masks..............................................................19
2.2.4 Polymer micromachining....................................................................................20
2.2.5 Other micromachining methods........................................................................21
2.3 Silicon micromachining....................................................................................................21
2.3.1 Wet etching of silicon...........................................................................................21
2.3.2 Dry etching of silicon...........................................................................................22
2.4 Glass micromachining.......................................................................................................24
2.4.1 Wet etching of glass..............................................................................................24
2.4.2 Dry etching of glass..............................................................................................25
2.4.3 Powder blasting....................................................................................................26
2.4.4 Discussion about glass machining.....................................................................27
2.5 Wafer bonding....................................................................................................................28
2.5.1 Direct bonding.......................................................................................................29
2.5.2 Anodic bonding....................................................................................................30
2.6 Packaging and chip-to-world interfacing......................................................................30
2.6.1 Fluidic interconnections.......................................................................................31
3 Heated nebulizer chips 35
3.1 Different chip designs and fabrication processes.........................................................35
3.1.1 Substrate materials...............................................................................................36
3.1.2 Channel fabrication processes............................................................................37
3.1.3 Cover wafer bonding............................................................................................40
3.1.4 On-chip heater.......................................................................................................41
3.2 Lifetime and failure mechanism of the platinum heater ............................................43
3.3 Fluidic and electrical interconnections...........................................................................45
3.4 Jet shape and temperature................................................................................................46
3.5 Applications of the heated nebulizer chips...................................................................50
4 On-chip liquid chromatography 53
4.1 Review of LC chips............................................................................................................53
4.1.1 Materials.................................................................................................................53
xi
4.1.2 Columns.................................................................................................................53
4.1.3 Interconnections, sample injection and pumping...........................................55
4.1.4 Detection................................................................................................................56
4.2 All-glass LC-nebulizer chip..............................................................................................56
5 Conclusions and outlook 59
References 61
Appendices: Publications I ? VI
xii
List of publications
This thesis consists of an overview and of the following publications which are referred to
in the text by their Roman numerals.
I Ville Saarela, Sami Franssila, Santeri Tuomikoski, Seppo Marttila,
Pekka ?stman, Tiina Sikanen, Tapio Kotiaho, and Risto Kostiainen, Re-usable
multi-inlet PDMS fluidic connector, Sensors and Actuators B: Chemical 114 (2006)
552-557 doi:10.1016/j.snb.2005.06.009
II Ville Saarela, Markus Haapala, Risto Kostiainen, Tapio Kotiaho, and Sami
Franssila, Glass microfabricated nebulizer chip for mass spectrometry, Lab on a
Chip 7 (2007) 644-646 doi:10.1039/b700101k
III Kai Kolari, Ville Saarela, and Sami Franssila, Deep plasma etching of glass for
fluidic devices with different mask materials, Journal of Micromechanics and
Microengineering 18 (2008) 064010 (6pp) doi:10.1088/0960-1317/18/6/064010
IV Ville Saarela, Markus Haapala, Risto Kostiainen, Tapio Kotiaho, and Sami
Franssila, Microfluidic heated gas jet shape analysis by temperature scanning,
Journal of Micromechanics and Microengineering 19 (2009) 055001 (10pp)
doi:10.1088/0960-1317/19/5/055001
V Ville Saarela, Markus Haapala, Jaroslav P?l, Nisse Kalkkinen, Marko Hukka,
Kai Kolari, Raimo A. Ketola, Risto Kostiainen, Tapio Kotiaho, and Sami Franssila,
On-chip liquid chromatography - atmospheric pressure ionization - mass
spectrometry, The Proceedings of ?TAS 2007 Conference 2 (Paris, France, 7?11
October 2007) 1435-1437
VI Markus Haapala, Ville Saarela, Jaroslav P?l, Kai Kolari, Tapio Kotiaho,
Sami Franssila, and Risto Kostiainen, Integrated liquid chromatography?heated
nebulizer microchip for mass spectrometry, Analytica Chimica Acta 662 (2010) 163-
169 doi:10.1016/j.aca.2010.01.005
xiii
Author's contribution
I The PDMS connector was invented by the author. Experimental work except for
the microfabrication of the nebulizer chips and mass spectrometry measurements
was performed by the author. The article was written by the author with
contribution from the others.
II All experimental work except for the mass spectrometry was conducted by the
author. The article was written by the author with contribution from the others.
III Design of the APCI chip and partial microfabrication of the samples were carried
out by the author. The article was written by Dr. Kai Kolari with contribution from
the author and Prof. Sami Franssila.
IV Microfabrication of the chips and computer simulations were carried out by the
author. Thermocouple measurements were done together with Dr. Markus
Haapala. The article was written equally by the author and Dr. Markus Haapala
with contribution from the others.
V Layout and process design, microfabrication of the chips and writing of the paper
were performed by the author.
VI Microfabrication and chip layout design were carried out by the author. The
publication was written equally by the author and Dr. Markus Haapala.
In addition to the publications above, the author fabricated the heated nebulizer chips and
contributed to the writing of other journal articles [1-12] that show different applications of
the chips.
xiv
Abbreviations and symbols
AFM atomic force microscope
APCI atmospheric pressure chemical ionization
API atmospheric pressure ionization
APPI atmospheric pressure photoionization
APTSI atmospheric pressure thermospray ionization
ARDE aspect ratio dependent etching
CE capillary electrophoresis
CEC capillary electrochromatography
CMOS complementary metal-oxide-semiconductor
COC cyclic polyolefin (also known as cyclo olefin polymer, COP)
DAPPI desorption atmospheric pressure photoionization
DRIE deep reactive ion etching
DSP double side polished (wafer)
EBL electron beam lithography
ECDM electrochemical discharge machining
EDM electric discharge machining
EDP ethylenediamine pyrocatechol
EDS energy dispersive spectroscopy
ESI electrospray ionization
FIB focused ion beam
FT-ICR MS Fourier transform ion cyclotron resonance mass spectrometry
GC gas chromatography
HN heated nebulizer
IC integrated circuit
IPA isopropyl alcohol, 2-propanol
IS ionspray
KOH potassium hydroxide
LC liquid chromatography
LIF laser-induced fluorescence
M molecular weight in unified atomic mass units or daltons (Da)
m/z mass to charge ratio in mass spectrometry
MEMS micro electro mechanical systems
MS mass spectrometry
MS/MS tandem mass spectrometry
MST micro system technology
?TAS micro total analysis system
NIL nanoimprint lithography
xv
PDMS polydimethylsiloxane
PLIF planar laser-induced fluorescence
RIE reactive ion etching
SEM scanning electron microscope
SOI silicon-on-insulator
SSI sonic spray ionization
SSP single side polished (wafer)
T temperature
TMAH tetramethylammonium hydroxide
UV ultraviolet
xyz three dimensional coordinate system with orthogonal axes

1
1 Introduction
1.1 Background
This work is about applying microfabrication technology to create instrumentation for
analytical chemistry. This section introduces these two fields of science and their
multidisciplinary combination.
1.1.1 Microtechnology
Applications of microtechnology have revolutionized the world since the first integrated
circuits (IC) were made in the 1950's. Originally, most microfabrication processes have been
developed with semiconductor ICs in mind, but the same methods can be applied in
countless applications. Our everyday life is full of appliances that have microfabricated
parts inside; flat-panel displays, ink-jet printers, acceleration sensors for car airbag systems
and game consoles ? just to name a few examples ? could not exist without the
development of microfabrication processes. Even mechanical wrist watches may have small
gearwheels made with microfabrication techniques. The abbreviations MST (micro system
technology) and MEMS (micro electro mechanical systems) are used with microfabricated
non-IC devices that include various sensor and actuator devices and mechanical
microstructures.
Microfabrication is based on the addition, patterning and selective removal of layers of
various materials on flat substrates. Layer thicknesses range from sub-nanometre to
hundreds of micrometers. Microfabrication is ideally suited for mass production because
multiple copies of the same component can be made with the same amount of work that is
required to make a single component on the substrate. For example, with the fabrication
steps that are needed to make a single transistor on a silicon wafer there can be hundreds of
individual IC chips on the wafer each consisting of billions of transistors. The famous
Moore's law formulated by Gordon E. Moore, the co-founder of the Intel Corporation, in
1965 predicted an exponential growth in the number of components on a single IC chip [13]
and the trend has continued already for half a century. This has been enabled by the
tremendous advances in several aspects of microfabrication technology. Most importantly
the patterning resolution has steadily improved. The minimum linewidth has decreased
from 10 ?m of the early 1970's to 45 nm used for production today with 32 nm and 22 nm
processes in the development phase [14]. For comparison, the diameter of a human hair is
over a thousandfold. In this respect it is not difficult to understand that dust-free
cleanrooms and protective gowning on the human operators therein are needed for
successful microfabrication. Cleanrooms have precise temperature and moisture control in
2 1 Introduction
order to achieve repeatable results with the sensitive process equipment. The
microfabrication methods will be further discussed in Chapter 2.
The infrastructure needed for microfabrication is expensive to set up and maintain.
Setting up a newest generation IC factory (fab) with a cleanroom and a comprehensive
equipment base costs billions of dollars [15] and only a few major companies can afford
them. Therefore, the industry relies on mass production in order to keep the business
profitable. However, thanks to IC foundries that sell their wafer processing capabilities, it is
possible for small fab-less design companies to work in the industry as well. Of course, not
all microdevices require a multi-billion dollar manufacturing infrastructure. For example,
many MST devices (e.g. fluidics) require minimal processing equipment and the cleanroom
requirements may be less strict. Sometimes older generation equipment is sufficient and
can be bought refurbished at a fraction of the cost of the latest generation automated
production tools.
1.1.2 Analytical chemistry and mass spectrometry
Analytical chemistry is the study of the composition of materials, identification of the
molecular components in a sample and their concentrations. Analytical information and
methods are widely applied in biomedical applications, environmental monitoring,
industrial quality control and forensic science.
A plethora of analytical methods exist and, depending on the specific application,
different combinations of them are used for optimal results. Typically, an analytical process
can be divided in the following three sub-processes: sample pretreatment (e.g. purification
and concentration), separation of the sample components (analytes) and detection.
Naturally, for any meaningful results the raw signal of the detector has to be interpreted by
a skilled chemist. The flowchart of an analytical method is shown in Figure 1.1. Liquid
chromatography?mass spectrometry (LC?MS) is a widely used analytical technique which
has also been used in this work.
In liquid chromatography (LC) a liquid sample plug is introduced to an eluent flowing
through a porous column. With proper eluent composition the analytes may be adsorbed
and thus concentrated at the beginning of the column. By changing the eluent composition
the analytes will then start passing through the column where different components of the
Figure 1.1: Flowchart of a generic analytical method with schematic details of liquid
chromatography and mass spectrometry as possible choices as the separation and detection methods.
This work deals with the highlighted parts.
SAMPLE
- analy tes
- matrix
SAMPLE
PREPARATION SEPARATION DETECTION
INTER-
PRETATION
RESULTS
ION SOURCE MASSANALYZER DETECTOR
MASS SPECTROMETER
COLUMN
SOLVENT A
SOLVENT B
GRADIENT
PUMP
INJECTION
VALVE
LIQUID CHROMATOGRAPHY
1.1 Background 3
sample adsorb and desorb between the stationary (particles packed in the column) and the
mobile (eluent) phases at different rates leading to spatial separation of the sample
components. A detector placed at the end of the column provides a chromatogram
containing the detector signal as a function of time. LC will be further discussed in Chapter
4.
Mass spectrometry (MS) is a very powerful analytical method offering both high
sensitivity and specificity. Mass spectrometers measure the mass-to-charge (m/z) ratios of
ions and from the mass spectra it is possible to identify the sample compounds by their
molecular masses. Large molecules can be fragmented into several ions producing
characteristic peaks in the mass spectrum. A liquid sample needs to be vaporized and
ionized prior to analysis and detection in a mass spectrometer. For higher analytical
performance, a sample separation step, such as LC, is typically done before the ionization
and detection. In this way it is easier to identify different sample components from the
mass spectra. There are several ionization methods and the optimal choice depends on the
sample type. Atmospheric pressure ionization (API) methods are widely applied for
introducing liquid samples into mass spectrometers [16]. A common API method is
atmospheric pressure chemical ionization (APCI) where the sample is sprayed by a heated
nebulizer which vaporizes and mixes the sample with nebulizer gas. The mixture is
directed towards the ion inlet of a mass spectrometer. A sharp needle with high electric
potential is used for ionization. A corona discharge takes place at the tip of the needle
leading to ionization of the analytes via complex gas phase charge transfer processes before
the inlet. Another ionization method, atmospheric pressure photoionization (APPI), is
similar to APCI but a 10 eV photon energy photoionization lamp is used to initiate the
ionization. Figure 1.2 shows a mass spectrum of protonated analytes with APPI?MS.
Ionization methods utilizing microchip technology will be further discussed in Sections 1.2
and 3.5.
1.1.3 Microfluidics, ?TAS, BioMEMS and Lab-on-a-Chip
Microfabricated separation and ionization chips are examples of microfluidic devices. The
best known microfluidic devices are ink-jet printer heads that have the highest commercial
value with estimated 28% share of the $6,000 million total value of the MEMS market [17].
Ink-jet heads contain an array of tiny nozzles that shoot out picolitre-size droplets (~1 ?m
diameter spheres) of ink. Other applications for micronozzles can be found, for example, in
micropropulsion engines for miniaturized satellites [18], liquid dispensers [19], steam
(a) (b) (c)
Figure 1.2: (a) Mass spectrum of testosterone (M = 288) and progesterone (M = 314) using APPI.
Concentration of both analytes was 1 ?mol L?1 and sample flow rate was 5 ?L min?1. [II] Molecule
structure of testosterone (b) and progesterone (c).
4 1 Introduction
locomotive power generator [20], micro flame ionization detector [21] and the heated
nebulizer (HN) chips of this dissertation. A seminal example of a microfluidic chip for
analytical chemistry is the gas chromatograph made on a two inch silicon wafer by Terry et
al. in 1979 [22]. In 1990, Andreas Manz conceived a concept called the miniaturized total
chemical analysis system [23] that has since then been established as ?TAS (micro total
analysis system) [24]. The number of yearly publications containing the concept
'microfluidics' has been increasing for two decades as shown by the statistics in Figure 1.3.
BioMEMS is sometimes used as an alternative term for microfluidics, but the applications
are focussed for biological analytes, such as viruses, bacteria, cells and DNA. Term ?lab-on-
a-chip? is also frequently used in conjunction with chips for analytical chemistry.
?TAS holds a great promise from scientific, commercial and health care point of views.
For example, rapid point-of-care or even home diagnosis tools can help in treatment of
diseases [25]. In micro and nano scale different phenomena can become dominant. For
example, fluid flow will be laminar, instead of turbulent and mixing will only take place
through diffusion. Miniaturization of analytical tools has several potential advantages:
? Suitability for smaller sample volumes because of smaller dimensions and smaller
dead volumes.
? Higher sensitivity because analytes can be more precisely guided to the detector.
? Faster analysis because of faster transportation and diffusion over short distances.
? Integration of several analytical steps on a single device, for example, drop
metering, mixing, thermal reactions, electrophoresis and detection [26].
? A plethora of suitable detection methods based on various electrochemical [27]
and optical [28] principles or mass spectrometers [29,30].
? Improving a key part in a larger system (this work) or
? Portable instruments after complete miniaturization of the system.
? Easy mass production and lower manufacturing cost of batch fabrication
compared to conventional machining.
19
70
19
72
19
73
19
75
19
76
19
77
19
78
19
79
19
80
19
81
19
82
19
83
19
84
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
19
99
20
00
20
01
20
02
20
03
20
04
20
05
20
06
20
07
20
08
20
09
20
10
0
500
1000
1500
2000
2500
3000
3500
4000
4500
2 1 1 1 1 1 1 2 1 3 1 3 2 6 4 5 11 9 10 18 19 29 27 35 43 53117178
317
558
855
1230
1838
2253
2629
3046
3594
3943
1781
Figure 1.3: Number of references found by SciFinder Scholar containing the concept 'microfluidics'
between years 1970?2010. (Status of June 17th 2010.)
1.1 Background 5
To date the examples of lab-on-a-chip devices that do not require any additional
instrumentation are few. Most publications concentrate on the miniaturization of a single or
a couple of functions and rely on external equipment for sample pretreatment, pumping,
power, detection or output of the results. In many cases it might be more appropriate to
talk about chip-in-a-lab instead of lab-on-a-chip. Also, the benefits of miniaturization are
sometimes exaggerated. For example, the fabrication cost of microfluidic devices may be
very high if only a few devices are produced in a cleanroom with high capital and running
costs. Also, some microfabrication processes are still experimental and unsuitable for mass
production because of repeatability problems and low yield. The economics are analogous
to the semiconductor industry where high utilization rate of production equipment is
crucial for a profitable business. Nevertheless, ?TAS technology and applications are
constantly evolving and in the near future plenty of success stories should arise from the
field. The acceptance and validation of revolutionizing new technology always takes time.
1.2 Microchips for mass spectrometry
Mass spectrometers enable accurate detection of the smallest amounts of sample. In
principle a single ion could be detected. Microfluidic devices operate with small sample
flows and, therefore, mass spectrometric detection is well suited for use with microfluidic
devices. During the past decade, microchip technology has been applied in ion sources,
mass analysers and even in complete miniature mass spectrometers [29].
Electrospray ionization (ESI) has gained the widest interest from the microfabrication
community [29,31,30]. In principle, the implementation of an ESI source is straightforward
as it only requires a high voltage preferably applied to a sharp tip. This dissertation work
has been a part of larger research project where also different ESI microchips have been
developed made of SU-8 [32] and silicon [33]. On-chip capillary electrophoresis (CE)
separation has also been integrated to the SU-8 ESI source [34]. Others have reported
microchips that combine LC and ESI with either monolithically integrated [35-37] or hybrid
integrated [38] emitters. Agilent Technologies Inc. has a commercial LC-ESI chip platform
based on laser ablated polyimide films [35,39-44]. LC chips will be further discussed in
Section 4.1.
Although ESI is convenient with microchips it is not capable of ionizing all kinds of
sample molecules. Nonpolar molecules are not efficiently ionized by ESI [45]. In contrast,
other atmospheric pressure ionization methods, such as APCI and APPI are applicable to
less polar and nonpolar samples. The HN chip is an essential part of the miniaturized APCI
and APPI methods. The fabrication, properties and applications of the HN chips will be
covered in Chapter 3.
Desorption ionization methods are used in combination with microchips as well.
Matrix-assisted laser desorption/ionization (MALDI) plates can be prepared with
microfluidic dispensers [31]. Porous silicon desorption plates have been demonstrated [46].
In desorption ESI (DESI) both microfabricated nebulizers [47] and microfabricated
desorption surfaces [48] can be used. Also the HN chips are suitable for desorption
ionization as will be discussed in Section 3.5.
Most of the work on ionization chips has been done with ordinary mass spectrometers
that are of the size of a refrigerator. Nevertheless, miniaturized mass spectrometers have
also been developed [49] and a complete mass spectrometer has been enclosed in the size of
6 1 Introduction
a shoebox allowing portability [50]. Mass analysers are based on ion trap, quadrupole or
time-of-flight principles, or on electrostatic and magnetic filters [29,51]. So far, the
analytical performance ? in terms of mass range, resolution and accuracy ? of
miniaturized mass spectrometers has been inferior compared to ordinary mass
spectrometers. For example, a typical miniature mass spectrometer offers a mass range
(m/z) of a few hundred with unit resolution whereas a typical laboratory mass
spectrometer has a few orders of magnitude better performance values. Therefore, the
current miniature mass spectrometers can only be applied in some special applications,
where small size or low price outweight the performance, such as space exploration or
process monitoring, respectively. The miniaturization of mass spectrometers is mostly
limited by the size and power consumption of the required vacuum pumps.
1.3 Aim of research
This work deals with the microfabrication, materials and integration of the HN chips. The
first prototypes of the chips were published in [52-54] and they have a patent pending [55].
The HN chips are used to vaporize a sample, mix it with a nebulizer gas and spray the
mixture out for analysis in a mass spectrometer. Various aspects of the HN chips were
further developed and studied during this work. These include:
? Layout of the fluidics and heater (II and Section 3.1)
? Different fabrication processes (II, III and Section 3.1)
? Lifetime of the chip (Section 3.2)
? Fluidic and electrical interconnection schemes (I and Section 3.3)
? Characterization of jet shape (IV and Section 3.4)
? Development of new applications for the chip (Section 3.5)
? Optimization of the chip for different applications (Section 3.5)
? Development on a pillar frit LC column chip (V and Chapter 4)
? Integration of an LC column to the HN chip (VI and Chapter 4)
1.4 Summary of publications
1.4.1 Publication I: Re-usable multi-inlet PDMS fluidic connector
Publication I addresses an important practical issue with microfluidic chips by presenting a
method for connecting small tubes to microfluidic chips. A connector block with holes for
the tubes is cast from polydimethylsiloxane (PDMS). The mould has a matching surface
topography as the fluidic chip for which the connector is intended for enabling easy
alignment of the connector and the chip. When the connector block is contacted with the
chip a leak tight connection is formed due to the self sealing nature of PDMS. It reversibly
adheres to various polished surfaces, such as glass and silicon. The pressure tolerance of
the connector can be increased with an additional compressing fixture or by making the
PDMS bonding permanent by oxygen plasma activation prior to contacting of the surfaces.
Various leakage tests were conducted. Figure 1.4 shows a comparison of the PDMS
connector with commercial NanoPort connectors. The chip size in an early HN chip design
was limited by the area required by the three NanoPort connectors. Significant chip size
1.4 Summary of publications 7
reduction was possible when using a PDMS connector. Fluidic interconnection schemes
will be further discussed in Sections 2.6.1 and 3.3.
Although the PDMS connector principle is mechanically working, other connection
methods have been used with the HN chips after the publication of the paper. The
operating temperature of the HN chips is too high for PDMS and the outgassing from
PDMS increases background noise in mass spectra. Therefore, the use of PDMS connectors
is limited to room temperature applications or in use with less sensitive detection methods
than mass spectrometry.
1.4.2 Publication II: Glass microfabricated nebulizer chip for mass
spectrometry
The first generations of HN chips were fabricated from silicon and glass substrates [52,56].
Publication II describes how these chips can be made from two glass wafers. In addition, a
fluidic connection method that uses a custom made holder and a commercial flat-bottom
fitting is presented (see Figure 1.5). Through-wafer glass wet etching is done with an
LPCVD silicon hard mask. Alternative approaches to glass-to-glass bonding and platinum
patterning are evaluated. Compared with silicon?glass nebulizer chips the all-glass design
enables higher on chip thermal gradients ? thanks to the lower thermal conductivity of
glass. This enables higher operating temperature at the vaporizer end of the chip as the
interconnections end can remain at a lower temperature. Different chip materials and
fabrication processes will be further discussed in Section 3.1.
Figure 1.5: All-glass nebulizer chips with a NanoPort connector and a custom made holder. [II]
Figure 1.4: Nebulizer chips with three NanoPort connectors (left) and a PDMS connector (right).
[I].
8 1 Introduction
Similar all-glass HN chips are still being fabricated in the follow-up research project
MISIMA but there has been modifications to the chip layout and the glass-to-glass bonding
process has been fine-tuned to give better yield.
1.4.3 Publication III: Deep plasma etching of glass for fluidic devices with
different mask materials
Deep reactive ion etching (DRIE) of Pyrex and silica glass substrates is demonstrated in
publication III with the use of four different masking materials: silicon shadow mask,
electroplated nickel, LPCVD amorphous silicon and SU-8. Over 100 ?m etch depth was
achieved with nearly vertical side wall profiles. The challenge in glass DRIE etching is
limited selectivity between the etch rates of glass and the mask material. Fairly good
selectivity (up to 35:1) is obtained with silica glass. However, Pyrex glass contains metal
oxides that do not produce volatile etch products and hence requires more physical
etching. Therefore, mask selectivity is reduced by one decade compared to silica glass.
Figure 1.6 shows a nebulizer chip nozzle structure etched with a silicon shadow mask.
Glass micromachining methods including plasma etching will be further discussed in
Section 2.4.
The presented glass DRIE methods have been applied in the fabrication of both HN
chips (see Section 3.1.2) and integrated LC?HN chips (Section 4.2). However, because of
added process complexity and lack of a high selectivity Pyrex glass etch process these
experiments are currently on hold.
1.4.4 Publication IV: Microfluidic heated gas jet shape analysis by
temperature scanning
Publication IV presents a new measurement technique for determining the temperature
distribution of a gaseous jet. A miniature thermocouple is attached to a computer
controlled xyz stage which scans over the jet. Temperature data with spatial resolution
below 100 ?m is easily achieved. The jet shapes from two different nebulizer chip nozzles
are compared. A computer simulation was done to study the relationship between
temperature and velocity distributions. A previously unpublished 3D temperature
measurement of a jet produced by a HN chip is shown in Figure 1.7.
The thermocouple scanning method has been used to study differences in jet shapes
with different nozzles. The shape of the jet from a HN chip has an influence on both
ionization and ion collection efficiencies. HN chip jet shape measurements will be further
discussed in Section 3.4.
Figure 1.6: Nebulizer chip nozzle structure DRIE etched in Pyrex glass. [III]
1.4 Summary of publications 9
Figure 1.7: 3 ? 3 ? 12 mm? 3D temperature scan of a gas jet coming from the rear left. The darker
the colour, the higher the temperature. Measured with the technique presented in publication IV.
1.4.5 Publication V: On-chip liquid chromatography ? atmospheric pressure
ionization ? mass spectrometry
A microfabricated LC column with a micropillar frit is presented in publication V. The chip
consists of bonded silicon and glass wafers with DRIE etched channels in the silicon part.
The LC column is a 40 mm long, 200 ?m wide and 170 ?m deep meandering channel and it
ends in an array of 5 ?m diameter pillars with 5 ?m gaps. Layout of the LC chip is shown
in Figure 1.8 (a). Either optical detection, such as laser-induced fluorescence (LIF), through
the glass cover can be used or the LC outlet can be connected to an external detector via
tubing. The tubes are inserted from the side of the chip and sealed with glue. The LC
column is packed with particles prior to use. The inlet capillary used during packaging will
be filled with particles too, but it can be easily removed by snapping a 5 mm part of the
chip away and inserting a fresh capillary to the column. The break up point is determined
by a shallow cut made with a dicing saw during the dicing of the wafer.
Mass spectrometric detection was demonstrated by connecting the LC chip to a HN
chip for APCI (Figure 1.8 (b)). This LC chip design was an intermediate step on the way
towards an integrated chip with both the LC column and the heated nebulizer on the same
chip. The integrated chip is presented in publication VI.
(a) (b)
Figure 1.8: (a) Layout of the liquid chromatography chip. (b) Extracted ion chromatograms of
corticosterone (solid line) and 5?-pregnan-3?-ol-20-one (dashed line) measured with LC?APCI?
MS. [V]
1.4.6 Publication VI: Integrated liquid chromatography ? heated nebulizer
microchip for mass spectrometry
Publication VI describes a silicon?glass HN chip with a integrated LC column (LC?HN
chip). The LC column of the integrated chip has a similar micropillar frit as the LC chip in
packing capillary
input capillary
40 mm LC column
output capillary
3 mm
26 mm
pillar frit and
optical detection channel
10 1 Introduction
publication V. However, the column is straight and the interconnections are done with
through-holes in the silicon part. The construction of the chip is illustrated in Figure 1.9. In
addition to the LC inlet, the chip has inlets for nebulizer gas which is mixed with the LC
eluent flow at the end of the LC column. A vaporization channel ends to a nozzle at the
edge of the chip. The glass cover has separate platinum heater elements for the LC column
and HN part of the chip. However, the LC column heating was not yet used in the
experiments presented in the publication. Figure 1.10 shows the LC?HN chip with its
fluidic connector and water cooler blocks. Cooling was necessary to prevent excess heating
of the LC column part because a lot of heat would otherwise be conducted the vaporizer
part of the chip. Silicon is an excellent thermal conductor.
The chips were successfully tested using both optical and MS detection. Figure 1.11
shows the separation of two fluorescent compounds detected by LIF after the micropillar
frit. Results from a LC?APPI?MS/MS experiment are shown in Figure 1.12. MS/MS denotes
tandem mass spectrometry. The results show clear separation of the analytes but the peaks
in the chromatograms are tailing. Ideally, the peaks should have a sharp Gaussian profile.
LC chips will be further discussed in Chapters 4 and 5.
Figure 1.9: Layout of the 55 mm ? 5 mm LC?HN chip. A silicon chip (a) with etched structures is
anodically bonded with a glass cover (b) with heater elements. (c) Chip cross-section at the column
section. [VI]
Figure 1.10: The integrated LC?HN chip with its fluidic connections, and the water cooler block
with electrical connections. [VI]
1.4 Summary of publications 11
Figure 1.11: Chromatogram of fluorescein and BODIPY? 493/503 separated and detected by
microchip LC?LIF. Concentrations were 1 ?M and injection volume was 0.1 ?L. [VI]
Figure 1.12: Selected reaction monitoring chromatograms of three different compounds (selective
androgen receptor modulators, SARM) measured by chip LC?APPI?MS/MS. Concentrations were
10 ?g mL?1 and injection volume 0.5 ?L. [VI]

13
2 Microfabrication for fluidics
Microfabrication methods can be applied in countless ways to produce different
components. Figure 2.1 lists the basic microfabrication processes steps that each have
numerous alternative practical implementations. A fabrication process starts with choosing
a suitable substrate, e.g. a silicon wafer, and then the process continues with thin-film
deposition, diffusion or etching into the wafer. In principle, all processing affects the whole
surface of the substrate or at least one side of the substrate. Patterns are generated through
lithography. Typically, a substrate contains multiple copies of a component and at the end
of the processing the substrate is diced and the chips are individually packaged. This
chapter reviews the basic microfabrication process steps from the viewpoint of fluidic
devices ? the HN chips in particular. For additional information and other applications
there are plenty of good books [57-61].
2.1 Materials
Materials for microfabrication can be divided into two categories: substrate and thin-film
materials. Early CMOS ICs were composed of only five elements: Si, O, B, P and Al. Silicon
substrates were doped with boron and phosphorous. Thin-films of silicon dioxide and
aluminium were used as electrical insulators and conductors, respectively. In principle, all
elements, compounds and alloys can be applied in microfabrication. However, most
cleanrooms are made for IC processing which limits the allowed materials therein. For
example, even the smallest traces of alkali, noble and transition metals have detrimental
effects on the electrical properties of semiconductors. IC compatible materials and
chemicals can be obtained in very pure and well controlled form with extremely low levels
of contaminating agents. For microfluidic components the IC compatibility is usually not
an issue and therefore the range of materials used for fluidics is wider ? although
sometimes limited by restrictions in shared cleanrooms.
2.1.1 Substrate materials
The most common substrates in microfabrication are single crystal silicon wafers. The
diameter of the wafers has increased along with the development of manufacturing
technology from 1 inch of the early days of ICs to 450 mm that are expected to come in
industrial use in a few years [15]. Usually, all the process equipment in a cleanroom is
dedicated to a single wafer size, for example, at TKK Micronova 100 mm wafers are used.
Smaller substrates can in some cases be processed using the same equipment with
additional fixtures or by temporary gluing to normal size carrier wafer. However, when
moving into larger substrates the equipment base needs to be renewed which is very
expensive.
14 2 Microfabrication for fluidics
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(i)
Figure 2.1: Basic microfabrication processes steps: (a) thin-film deposition, (b) diffusion, (c)
lithography, (d) etching, (e) lift-off, (f) master replication, (g) wafer bonding, (h) wafer dicing and (i)
packaging.
X123
die attachment wire bonding encapsulation
1 2
3 4
1 2
3 4 1
2 3
4dicing
separate
chips
annealingbondingwafer alignment
casting curing releasing of the replica
lithography metal deposition lift-off
lithography etching of a thin film removal of photoresist bulk etching with the thin film as a mask
application of
photoresist
UV -exposure through
a photomask
development of
exposed photoresist
solid phase diffusion
gas phase diffusion
ion implantation
substrate single sided deposition
double sided deposition
2.1 Materials 15
In fluidic applications, glass wafers are often used as substrates in addition to silicon
wafers. The advantages of glass are optical transparency, electrical insulation and good
chemical resistance. In addition, it is possible to tailor a lot of the material properties by
adjusting the composition of the glass. A good book about microstructuring of glasses was
recently published [57]. Two most common glass types that are available in wafer format
are fused silica and borosilicate glasses. Fused silica is pure SiO2 in amorphous form.
Borosilicate glass brand names include Corning Pyrex 7740 and Schott Borofloat 33. They
contain approximately 81% SiO2, 13% B2O3, 4% Na2O, 2% Al2O3 and traces of other
compounds. The coefficient of thermal expansion of borosilicate glasses is close to that of
silicon, which is important if they are bonded with silicon wafers. Different bonding
methods will be discussed in Section 2.5.
The typical thickness range of 100 mm diameter wafer is 300?1,000 ?m and the wafer
has a 32.5 mm flat for aiding both physical and crystal orientation alignment. Single crystal
silicon wafers can be polished to atomically smooth surface with RMS roughness below
0.5 nm. Glass polishing is more challenging but typical surface roughness is below 2 nm.
There are both single side polished (SSP) and double side polished (DSP) wafers. SSP
wafers are used in IC processing, but MEMS often requires DSP wafers.
Several fluidic devices are made on polymer substrates [62]. They have potential for
cheap and disposable mass production. Polymers are composed of long organic molecules
which offer wide selection of different physical and chemical properties. Polymers have
limited temperature tolerance and they are usually intended for room temperature
operation. Nevertheless, some polymers can tolerate temperatures slightly above 300 ?C
and in the case of polyimide thin-films even up to 450 ?C [63].
Table 2.1 lists some key properties of substrate materials. Depending on the
composition of glasses and polymers they offer a wide range of different material
properties. Naturally, most material properties are temperature dependent.
Table 2.1: Selected properties of different materials at room temperature.
Density
(g cm?3)
Resistivity
(? cm)
Thermal
conductivity
(W m?1 ?C?1)
Coefficient of
thermal expansion
(10?6 ?C?1)
Melting
/softening
point (?C)
Young's
modulus
(GPa)
Silicon 2.33 10?5 ? 104 (a) 150 (a) 2.6 1,400 180 (b)
Fused silica 2 1016 1.3 0.55 1,650 70
Pyrex 7740 2.24 106 1.13 3.25 821 63
Platinum 21.45 10.6 ? 10?6 71.6 8.9 1770 160
Polymers (c) 0.9 ? 2 10 ? 1018 0.1 ? 0.4 10 ? 100 90 ? 410 0.01 ? 4
(a) Doping dependent
(b) Crystal orientation dependent
(c) Typical property ranges for the material category
2.1.2 Thin-film materials and deposition
In microfabrication, thin layers of different materials are deposited on the substrate (Figure
2.1 (a)). Thin-films with thickness ranging from atomic layers to several micrometers can be
deposited on a substrate with various methods. Physical vapour deposition (PVD) methods
such as vacuum evaporation and sputtering are mostly used to deposit metal films.
Dielectric and semiconductor thin-films are usually deposited using chemical vapour
deposition (CVD) methods. PVD films are deposited only on one side of the substrate at a
16 2 Microfabrication for fluidics
time, but some CVD processes can deposit on both sides simultaneously. Low pressure
CVD (LPCVD) and plasma enhanced CVD (PECVD) are the two dominant CVD methods.
Thin-films can also be deposited in liquid form by applying a drop on the wafer and then
spinning the wafer for the drop to spread out throughout the wafer surface. This is called
spin-on coating. Electroplating is another method where the deposited material comes
from liquid solution.
The properties of thin-films are very diverse and they usually differ from the properties
of the corresponding bulk material. For example, the resistivity of metal thin-films are
frequently two times higher than the table value for the bulk metal. Relevant film
properties include elemental composition (stoichiometry), microstructure (amorphous vs.
polycrystalline vs. single crystal), film stress (tensile vs. compressive), thickness uniformity
(across wafer), step coverage (over topography) and surface roughness. The exact
properties depend from the deposition method, equipment and process variables such as
deposition chamber pressure, gas composition, temperature, applied electric field, plasma
power and frequency. Sometimes repeatable results are not easy to obtain, especially in a
research environment where the same equipment is used for multiple processes.
Some materials and deposition methods are incompatible. Thermal budget and
contamination issues have to be considered beforehand, because they can limit the
applicable deposition methods. In general, no high temperature processes above 450 ?C are
allowed after deposition of metals to ensure contact stability between silicon and metal or
to avoid softening and deformation of glass.
2.1.3 Diffusion and high temperature processing
Diffusion is concentration gradient driven movement of material. Concentration
differences will gradually smooth out with a rate exponentially proportional to
temperature. In microfabrication, various high temperature process are used to modify the
surface layer of the substrate or a thin-film or to promote reactions that would not occur at
room temperature where diffusion is usually negligible. Different diffusion methods are
presented in Figure 2.1 (b). Group III or V elements are diffused into silicon to change the
conductivity at the subsurface region of the substrate. For example, boron doping can be
made either by applying a boron-rich spin-on-glass film on the substrate and then
annealing at high temperature (solid phase diffusion) or by adding BCl3 gas to the
annealing atmosphere (gas phase diffusion) or by exposing the wafer with accelerated B+
ions (ion implantation) followed by activation annealing.
Thermal oxide is frequently used in the fabrication of microfluidic components.
Thermal oxidation of silicon is a thin-film deposition process where oxygen diffuses
through the surface oxide layer to react with underlying silicon. Oxidation furnace
temperatures are from 900 ?C to 1,200 ?C with either oxygen (dry oxidation) or water
vapour (wet oxidation) in the furnace atmosphere. Compared to CVD deposited oxides,
thermal oxidation provides superior quality SiO2 layers. Oxide layers are used for several
purposes, for example, as etch masks, electrical insulators, adhesion or etch-stop layers.
Semiconductor pn-junctions can be used as detectors but, in general, microfluidic
components rarely utilize the electronic semiconductivity of silicon. However, the doping
dependent material properties have other applications, too. For example, diffused silicon
resistors can be used as heaters and optical waveguides can be made by doping oxide films.
Doping and crystal orientation determines silicon piezoelectricity which can be utilized for
2.1 Materials 17
measuring strain of cantilever structures. Membranes with precise thickness control can be
achieved with the help of doped etch-stop layers (further discussed in Section 2.3.1).
2.2 Patterning and micromachining
For most practical applications, it is not enough to have layers of thin-films covering the
whole substrate. Instead, the thin-films are patterned and recesses are made to the bulk of
the substrate. The pattern making mostly relies on lithography and etching. These
processes are discussed in this section along with some less conventional ways for
micromachining.
2.2.1 Lithography
Lithography is the basis of making patterns on the substrate. Historically, different kind of
lithography and etching have been used by artists and printers for centuries before they
were adapted by the microfabrication industry. Also in microfabrication there are various
lithography processes (optical lithography, electron beam lithography, X-ray lithography
etc.) but they all follow the same basic procedure. The basic steps of photolithography are
illustrated in Figure 2.1 (c). First, a thin layer of photoresist is applied on a substrate. Then,
the photoresist is exposed with a mask pattern followed by development of the photoresist
leaving a patterned photoresist layer on the substrate. This layer can then be used as a
mask in subsequent processing, for example, in etching.
Optical lithography is the dominant technique for microfluidics. The photoresist is
sensitive to ultraviolet (UV) light and the exposure is done through a transparent
photomask with opaque patterns. Typically, a glass plate with patterned chromium thin-
film is used as the photomask. A five inch square photomask with 1 ?m minimum feature
size costs about ?500. For rapid prototyping of many microfluidic devices lower resolution
photomasks are sufficient. Around 100 ?m patterns can be printed on transparency films
with ordinary office laser printers. Laser photoplotter films that are used in the
manufacture of printed circuit boards have a minimum feature size of about 20 ?m and
they cost only a few Euros. However, the pattern edges are not sharp but have several
micrometres of ripple. In addition to the resolution of the photomask, diffraction limits the
maximum resolution of optical lithography. Diffraction depends on the wavelength of the
light source, the photoresist thickness and distance between the photomask and the
substrate. In contact lithography, the photomask and wafer are contacted during exposure
and the whole wafer is exposed simultaneously. Step-and-repeat aligners can have
reducing optics between the photomask and substrate. Step-and-repeat aligners are more
expensive than simple contact aligners but on the other hand they can reduce photomask
costs as smaller mask area or pattern resolution is sufficient for reproducing similar
patterns in photoresist.
Photoresists are divided into two categories: positive and negative tone photoresists. A
positive tone photoresist is rendered soluble upon exposure whereas an exposed negative
tone photoresist will cross-link further preventing it from dissolving in the developer.
AZ5214E photoresist from Clariant is a special photoresist that can be used both as positive
and negative tone photoresist. With normal process sequence (spinning, pre-bake,
exposure, development, post-bake) it is a positive tone photoresist. However, two
additional process steps after the exposure (baking followed with a flood exposure without
18 2 Microfabrication for fluidics
a photomask) will result in a photoresist layer with the negative image of the mask pattern.
Various photoresist chemistries exist with wide range of properties such as thickness,
resolution, exposure sensitivity, sidewall profile, temperature and chemical tolerances.
There are also other means to pattern a photoresist in addition to optical lithography.
These methods are also used for fabricating photomasks. Down to 0.6 ?m feature size can
be patterned using laser lithography systems [64]. Ultimate resolution is achieved with
electron beam lithography (EBL) which uses a focused electron beam to expose the
photoresist. The beam itself can be only a few nanometres in diameter but the exposed
resist area will be larger due to electron scattering. There is plenty of activity in the
development of EBL resists [65]. Sub-100 nm features are patterned on a routine basis but
sub-20 nm resolution is still in the development phase. The major problem of EBL is limited
throughput. Scanning over a large area takes long time, even several days, which renders
EBL unsuited for mass production. Another type of lithography is nanoimprint lithography
(NIL) where a master with the desired pattern as 3D surface topography is pressed against
the resist. Either UV curable or thermosetting resist types can be used. NIL has the
potential for high throughput nanometre scale lithography [66].
In microfluidics, a negative photoresist SU-8 is frequently used as a structural material
[62]. Different variations of SU-8 allow spin coating in wide thickness range from a few
hundred nanometres up to a few millimetres [67]. Polymer micromachining is further
discussed in Section 2.2.4.
2.2.2 Etching
In an etch process material is being removed starting from the surface of the sample.
Etching is done either with a corrosive solution (wet etching) or with aggressive gasses (dry
etching). Typically, the etch process follows lithography and the patterned photoresist is
used to protect some areas from being etched (Figure 2.1 (d)). This requires that the
photoresist tolerates the etchant. However, most etchants will attack the polymer
photoresist as well ? either gradually during a long etch process or sometimes very fast. In
such a case, the photoresist mask is first used to pattern a thin-film which is then used as an
etch mask in a subsequent etch process that would be too harsh for the photoresist. The
etch mask films ? typically silicon dioxide, silicon nitride or silicon ? are called hard masks.
Different materials are etched at different rates. Usually it is important to have etch
selectivity between materials so that some materials are etched while some are affected as
little as possible. Etching can be based on chemical reactions or physical erosion and it can
take place in either wet or dry phase. Categorization of a few etch methods are shown in
Table 2.2. In wet chemical etching the samples are immersed in liquid solutions and the
Table 2.2: Categorization of a few different etch methods.
Physical etching Both physical and chemical etching Chemical etching
Dry etching
Ion milling
Powder blasting
FIB
RIE
DRIE HF vapour
Wet etching Water jet machining ECDM
HF
KOH
TMAH
2.2 Patterning and micromachining 19
etch chemicals in the solution will react at the sample surface to produce soluble reaction
products. Etching can also be done in the dry phase either with chemically reactive etch
gasses producing volatile etch products with the substrate or by physical etching with
mechanical bombardment of the surface breaking off small pieces out of the substrate.
Some etch methods, such as reactive ion etching (RIE), combine both physical and chemical
etching.
Different etchants produce different etch profiles. Four typical etch profiles are
presented in Figure 2.2. An isotropic etch has no preferential directionality and the etch
proceeds spherically under the edges of the mask. The undercut of the etch mask equals the
etch depth. Therefore, the maximum aspect ratio (width:height) is isotropically etched
structures is 2:1. Wet etching of amorphous materials is always isotropic. Wet etching of
crystalline materials, on the other hand, may be anisotropic due to crystal orientation
dependent etch rates. For example, with caustic etch solutions, such as potassium
hydroxide (KOH), the {111} planes of silicon are etched slowly compared to the other
crystal planes. Thus, etched cavities are defined by {111} sidewalls as shown in Figure 2.2
(b). Plasma etching can be either isotropic or anisotropic. If the etching is purely chemical
the profile is usually isotropic as in wet etching. Physical etching is directional and results
in anisotropic etch profile. However, purely physical etching has low selectivity against the
masking materials and, therefore, for deep structures with high aspect ratios and vertical
sidewall profiles special processes are required. DRIE methods for silicon and glass will be
discussed in Sections 2.3.2 and 2.4.2, respectively. Powder blasting produces an anisotropic
channel profile with maximum aspect ratios around 1:3. Powder blasting is further
discussed in Section 2.4.3.
Figure 2.2: Etch profiles with different etch processes: (a) isotropic, (b) anisotropic wet etching of
<100> silicon, (c) ideally anisotropic DRIE and (d) powder blasting.
2.2.3 Lift-off patterning and shadow masks
In lift-off patterning lithography is done before thin-film deposition. The material that is
deposited over the photoresist will be removed during resist stripping (Figure 2.1 (c)). In
general, negative tone photoresists provide negative sidewall profile for lift-off patterning
than positive tone ones. Positive tone photoresist have typically slightly slanted sidewalls
leading to continuous film with thin connection between the deposited film on the
substrate and over the resist. Naturally, this makes resist stripping difficult and the pattern
quality may deteriorate due to the tearing of flakes at the pattern edges. Because the lifted-
off thin-film will end up as particles and flakes in the resist stripping solution there is a
danger of contaminating the sample, especially with microfluidic chips with channels.
54.7?
(a) (b)
(c) (d)
?{11
1}
{100}
20 2 Microfabrication for fluidics
Therefore, etch patterning is generally preferred and lift-off patterning is only applied to
materials that are difficult to etch.
Lift-off patterning is practically limited to evaporated and sputtered metals, because the
deposition temperature has to be close to room temperature in order to avoid damage to
the photoresist. Even with these deposition methods there is a chance that the wafer surface
will overheat because photoresists have poor thermal conductivity. Therefore, the
deposition power or time may need to be limited in order to avoid damage to the
photoresist. Otherwise the photoresist may become very difficult to remove or it may crack
leading to metal deposition to unwanted areas. Also, there may be outgassing from the
photoresist that affects the properties of the deposited film, for example, the sheet
resistance of the metal film may become higher.
Lift-off patterning is particularly useful for metals that are difficult to etch, such as
platinum. Lift-off patterning the thin-film heaters of the HN chips will be discussed in
Section 3.1.4.
Shadow mask deposition is a special kind of lift-off process where an external mask
with through-holes is used and no lithography is needed. Extremely high resolution (down
to 10 nm) can be achieved with the most advanced shadow masks [68]. In addition to
masked deposition, shadow masks can be applied to masking of plasma etching [69] as
done in publication III. If needed, the shadow mask can be accurately aligned in a bond
aligner and temporarily bonded with a small amount of photoresist, glue, vacuum grease,
or wax.
2.2.4 Polymer micromachining
Different polymers are frequently used for the fabrication of microfluidics. Polymers offer a
range of chemical, mechanical and optical properties and polymer microfluidic devices
they can be much cheaper compared to silicon and glass devices. There are several
micromachining techniques available for polymers.
Master replication techniques are used with thermoplastics and curable polymers.
Figure 2.1 (f) illustrates the casting of liquid pre-polymers, such as PDMS [70]. The inverse
topography of the mould is copied to the cured polymer. Publication I uses a wet-etched
silicon mould for the PDMS connectors. In hot-embossing, the temperature is increased
over the glass transition point of the thermoplastic and then the master is compressed
against the polymer substrate. Then the temperature is decreased below glass transition
point and the master is released. The embossed substrate will have the inverse topography
of the master. Injection moulding is used on a massive scale in the manufacturing of
compact disks for digital audio and data storage. Microinjection moulding processes and
master fabrication were recently reviewed [71]. The moulding approach is also feasible for
glasses given the moulding temperature is high enough [72].
Photoresists are polymers and in addition to their typical use as an etch mask they can
also be used as structural materials. SU-8 is chemically quite stable and transparent in the
visible and near-ultraviolet range. Layering of several SU-8 films enables complex 3D
structures made entirely of SU-8 [32]. In addition, SU-8 is frequently used to make masters
for PDMS moulding.
2.2 Patterning and micromachining 21
2.2.5 Other micromachining methods
Several other approaches can be used for machining in micro and nano scale. Conventional
machining methods can be scaled down. Very small diameter tools and mechanical
positioning stages with sub-nanometre resolution are available off-the-shelf. Milling with
rotating abrasive or cutting tool tips and electric (EDM) or electrochemical (ECDM) [73]
discharge machining can be used for machining with down to a few micrometres'
resolution. Atomic force microscope (AFM) tips provide the ultimate resolution enabling
the manipulation of single atoms as demonstrated twenty years ago by IBM researchers
[74]. However, all these methods lack the batch nature of lithography and etching and,
therefore, the throughput is low and the expense of a single device will be high.
Laser machining has potential for high throughput machining of patterning with
dimensions in the microscale [75]. Through-holes, trenches and even embedded 3D
structures are possible [76].
Powder blasting is mostly utilized for cleaning and surface texturing, but is suitable for
micromachining as well. In this context, the method is sometimes called as abrasive jet
machining with additional methods such as water jet machining [77] also falling within the
category. The resolution of powder blasting is determined by the size of the abrasive
particles that are readily available down to a few micron size. Elastic and ductile materials
can be used for masking of brittle and hard workpieces. Powder blasting of glass will be
further discussed in Section 2.4.3.
Focused ion beam (FIB) is a true nanomachining method. It is similar to EBL but instead
of electrons, it uses heavy ions, such as gallium [78]. FIB can do both physical ion-milling
and direct writing of an etch mask by doping the surface layer. FIB machining is suited for
practically any material, including glasses [79].
2.3 Silicon micromachining
This section reviews the basics of bulk micromachining of silicon using both wet and dry
etching methods. Further information can be found in several books [58-60,80] and review
articles [81].
2.3.1 Wet etching of silicon
Silicon wafers are single crystals and exhibit anisotropic wet etching in alkaline etch
solutions, such as KOH, tetra methyl ammonium hydroxide (TMAH) and ethylenediamine
pyrocatechol (EDP). The etch rate in <100> directions is tens or hundreds of times faster
than in <111> direction. The {110} and higher order planes are etched even faster. In
addition to the crystal plane, the etch rate depends on concentration, temperature, the
wafer's doping, mechanical agitation, masking layer stress and possible additives and
applied electric fields [82-88]. High levels of p-type doping can be used to slow down the
etch rate of the surface layer which enables etching of membranes with good thickness
control. An etch-stop to a p-type silicon can also be obtained by applying a voltage over a
diffused pn-junction. Addition of isopropyl alcohol (IPA) in KOH or TMAH reduces the
surface roughness and changes the crystal plane selectivity [88]. Changes in crystal plane
selectivity have significant implications. For example, mask undercutting is significantly
reduced for islands on <100> wafers as shown in Figure 2.3. KOH+IPA etching has also
22 2 Microfabrication for fluidics
been applied to make capillary tube stopper structures in the nebulizer chip channel which
will be further discussed in Section 3.1.2.
There are also isotropic wet etchants for silicon. They are based on mixtures of nitric
(HNO3) and hydrofluoric (HF) acids. The nitric acid oxidises the silicon surface and the HF
etches the oxide away. The properties of HNO3?HF?H2O and HF?HNO3?CH3COOH silicon
etch systems have been thoroughly documented in [89].
2.3.2 Dry etching of silicon
Dry etching of silicon can be done with fluorine or chlorine containing atmospheres with
[90] or without [91,92] plasma. Deep anisotropic etching requires sidewall passivation to
prevent underetching by chemical etching. Two approaches can be used for this: the pulsed
Bosch process and the cryogenic process. Two reviews of silicon DRIE technology were
recently published [93,94].
The patented Bosch process uses alternating etch and passivation steps repeatedly [95].
A similar gas chopping etch technique had previously been presented for etching
polyimide [96]. During the passivation step a fluoropolymer layer is deposited on the
sample. The passivation layer is removed from the bottom of trenches due to physical
etching. Both PECVD and RIE are equipment-wise very close to each others and, in
principle, both deposition and etching can easily be done using the same equipment just by
changing the process gasses, pressure and plasma power. The pulsed passivation and etch
process leads to characteristic ripple at the sidewalls of Bosch DRIE etched structures as
shown in Figure 2.4 (a). The size of the ripple depends on the exact process parameters,
such as the pulse lengths. Parameter optimization requires multivariate analysis [97,98].
However, minimal sidewall scallop is obtained at the expense of etch rate or mask
selectivity [99]. Alternatively, the ripple can be reduced by post processing, for example, by
wet etching in KOH+IPA [100] or using a doped silicate glass deposition and reflow
annealing [101].
(a) (b)
Figure 2.3: Square masked silicon island etched for 30 minutes in 82 ?C (a) 20% KOH and (b) 20%
KOH + 5% IPA. (Micrographs courtesy of Kestas Grigoras.)
10
0 ?
m
100 ?m
10
0 ?
m
100 ?m
2.3 Silicon micromachining 23
Another approach for silicon DRIE uses a non-pulsed process at cryogenic
temperatures around ?110 ?C [102]. The processes gasses are SF6 and O2. The process gasses
form a thin passivating SiOxFy layer on silicon surfaces but due to physical etching the layer
is removed from the bottom leaving only the sidewalls protected. The process is sensitive
to the oxygen flow rate. Too little oxygen leads to insufficient sidewall protection and mask
undercutting whereas too much oxygen will lead to silicon nanograss ? black silicon ? at
the bottom because the physical etching is not capable of removing the passivation layer
fast enough. With cryogenic DRIE it is easier to get smooth sidewalls compared to the
Bosch process, as shown in Figure 2.4 (b).
Even for a perfectly anisotropic etch profile the resulting sidewall profile is affected by
the sidewall profile of the etch mask. There is never an infinite etch selectivity between the
mask and substrate. Therefore, the exposed area can change along the thinning or
undercutting of the mask layer as illustrated in Figure 2.5.
Three important phenomena that lead to variations in etched depths with the same
plasma etching process are loading, the RIE lag and aspect ratio dependent etching (ARDE)
[103,104]. The ratio between the etched and protected surface area determine the loading.
The etch rate may decrease if loading increases. Small features are etching slower than
large ones because of RIE lag. This is due to the dynamics of the process, i.e. the ion
bombardment and gas transport rates. ARDE is similar to RIE lag, but it is related to
changes in etch rate of a chosen linewidth as the etch proceeds and aspect ratio of the
feature increases, whereas RIE lag is for a specific etch time and different linewidths. It is
possible to take advantage of RIE lag to fabricate channels with custom depth profile [105].
However, uniform etch depth are generally more favourable. It requires careful process
(a) (b)
Figure 2.4: Silicon DRIE sidewall profiles with different processes: (a) the pulsed-mode Bosch
process and (b) the cryogenic etch process (courtesy of Lauri Sainiemi).
Figure 2.5: Effect of mask profile on the etch profile in a perfectly anisotropic etch process but finite
selectivity between substrate and mask. (a) Vertical mask sidewalls can provide vertical channel
sidewalls but (b) slanted sidewalls or mask undercutting lead to slanted channel profile. (c) Having
higher etch selectivity helps.
(a) (b) (c)
3 ?m 5 ?m
24 2 Microfabrication for fluidics
optimization and may require settling for lower etch rates [106]. Ramping the process
parameters during the etch run can also improve the results [107].
2.4 Glass micromachining
This section will cover the literature about the three most typical glass micromachining
methods: wet etching, dry etching and powder blasting. In addition, other off-mainstream
glass micromachining methods are briefly discussed and compared. Recently, a book on
glass microfabrication was published [57].
2.4.1 Wet etching of glass
Because most glasses compose mostly of silicon dioxide, they are etched in HF-based
solutions. The etch rate depends on the concentration of the solution and can vary from
tens of nanometres per minute with buffered HF [108] close to 10 ?m min?1 with
concentrated HF [109,110]. Because glass is amorphous, the etching is isotropic. However,
there are some special glasses that do exhibit anisotropic wet etching. These include single
crystal quartz and photoactive glasses, such as Foturan [111].
Relatively shallow, up to a few tens of micrometer, etches can be done using a
photoresist mask [112,113]. Deeper etches require the use of a hard mask There are several
suitable hard mask materials that are not etched in HF [114]. Silicon is one obvious choice.
Both LPCVD [115] and PECVD deposited [116] silicon thin-films and even bonded silicon
wafers [117] with through etched patterns have been used. In addition, different
combinations of layers of chromium, gold, copper and photoresist [118-123] and PECVD
silicon carbide [109] have been successfully applied as etch masks for glass.
For defect free etching the masking layer stresses have to be low [118]. Otherwise there
will be pinholes and cracking of the under etched hard mask. These defects are illustrated
in Figure 2.6. Tensile stress in the hard mask layer often results in pinhole defects at the
surface where the etchant has penetrated through the mask through a microscopic hole.
Several glasses ? including Pyrex ? have constituents (e.g. K2O, CaO, MgO or Al2O3)
that do not produce soluble etch products with HF. During etching they lead to
micromasking and added surface roughness. The problem can be overcome by the addition
of H2SO4 or HCl in the etch solution [110,124]. Alternatively, the precipitated particles can
be removed by periodical 10 second immersion of the wafer in HCl solution after every 5
minutes of etching in BHF [112].
Figure 2.6: Mechanisms for mask induced etch defects. Pinhole formation (a-c) and notch defects (d).
Redrawn from [118].
defect tensile stress
creep
HF
pinhole
notch defect
underetching
hard mask
glass
(a) (b) (c) (d)
2.4 Glass micromachining 25
2.4.2 Dry etching of glass
Isotropic vapour phase etching of SiO2 and glasses can be done with HF vapours, which
can be used, for example, in the release etch of MEMS devices with sacrificial oxide [125].
However, anisotropic etching is often desirable due to inherent limitations of isotropic
etching. Anisotropic profiles can be obtained with RIE.
In general, fluorine chemistry is used in plasma etching of glasses. Various
combinations of process gasses have been studied: CF4 [126-128], CF4/O2 [129], CF4/Ar
[129,130], CF4/CHF3 [126,127], CHF3/Ar [129,130], CHF3 [131], SF6 [132-135], SF6/Ar
[130,132,133,136], SF6/CHF3 [137], SF6/C4F8 [137], C4F8 [131,137], C4F8/Ar [138], C4F8/Ar/CHF3
[139], C4F8/O2 [138], C4F8/O2/He [140], C4F8/He [140], C4F8/CH4 [141] and Ar [126,127].
Etch reaction products from SiO2 are highly volatile SiF4, CO, CO2 and H2O [127]. Many
glasses contain B2O3 with volatile reaction products BF3 and B2F4 [126] . However, several
other compounds that are used in glasses and that may be as trace impurities also in high
purity silica glasses have non-volatile etch products: AlF3, BaF2, CaF2, KF, LiF and NaF
[126,127]. The removal of these non-volatile species is achieved with physical etching, but
process optimization is required for good surface quality, profile control and etch
selectivity between masking material and the glass substrate. The glass compositions which
were studied include different quartz glasses [126,127,130,132,137,138,140], Pyrex and
similar borosilicate glasses [127,128,131-134,136,139,140], soda-lime glass [129] and several
other silica-based glasses [126,127,132].
The etch mask materials used include standard photoresist [127,129,138], thick SU-8
resist [128,130], aluminium [129], chromium [126,132], evaporated [136] and electroplated
[133-135] nickel, sputtered aluminium nitride [137], ALD deposited Al2O3 [137], LPCVD
amorphous silicon [140], polysilicon [138,141] and through etched silicon wafers ? either
anodically bonded [131,139] or as reusable shadow masks [140]. Publication III shows the
use of different C4F8/He/O2 plasma etches with four different mask materials: electroplated
Ni, LPCVD a-Si, SU-8 and a silicon shadow mask.
DRIE of glass with etched depths over 100 ?m deep is shown in [128,131,135,139,140].
In contrast to silicon DRIE with the pulsed Bosch process dominating, glass DRIE processes
are continuous. Glass DRIE etch rates are about an order of magnitude lower (up to a few
hundred nm min?1) compared to silicon DRIE. Major challenge is obtaining good etch
selectivity between the etch mask and glass. Fairly high selectivity (30:1) AlN and Al2O3
masks have been demonstrated for silica [137]. However, selectivity reduces with other
glasses that require significant physical etching for the non-volatile etch products. The non-
volatile etch products are sputtered on the sidewalls affecting the resulting etch profile
[134]. The sidewall taper angle depends on both mask opening width and glass
composition. In general, thick masking layers are needed for glass DRIE. For deposited
films, stress-related problems may limit the maximum mask thickness. Also, the thicker the
mask is the more difficult it is to make small patterns due to an increased aspect ratio
which contributes to RIE lag and ARDE of both the mask and subsequent glass etching. As
already illustrated in Figure 2.5 the sidewall profile of the mask affects the sidewall profile
of the etched trench. The optimal etch process depends on the used masking material. For
example, SF6 is not a good choice with silicon masks, because it rapidly etches silicon.
Instead, polymer forming etch gasses ? such as C4F8 or CHF3 ? should be used with silicon
masks to reduce the mask erosion.
26 2 Microfabrication for fluidics
2.4.3 Powder blasting
Since the mid-1990s powder blasting has become a fairly well established micromachining
method [142-145]. In powder blasting, high velocity particles (e.g. fine grain alumina sand)
erode the surface via cracking upon impact. It is well suited for brittle and hard materials,
such as glass. Powder blasting is anisotropic and the etch rate depends on the mass flow
rate of the particles and their velocity, that is, the applied kinetic energy.
Large areas are etched fairly uniformly but the resulting side wall profile of the pattern
edges is dependent on aspect ratio, the used particle size, shape and velocity [146]. The
profile changes with depth. An example of a powder blasted glass structure is shown in
Figure 2.7 (a).
As a rule of thumb, the minimum linewidth is about three times the particle diameter.
However, ARDE is much more significant compared to RIE and linewidths up to a few tens
of times the particle size are affected. With 30 ?m particles the etched depth starts reducing
below 400 ?m linewidth as shown in Figure 2.7 (b). Of course, also the mask thickness
contributes and thinner masking layer can be used alleviate ARDE to some extent.
Although in general ARDE is considered an unwanted side effect, it is possible to utilize
ARDE to implement channels with varying depth [147]. The etch profile can be tailored by
varying the incident angle of the powder jet [142,148]. This enables, for example, the
fabrication of released bridges and high aspect ration structures [149].
Various metal and polymer masks can be used with powder blasting [150,151]. These
include laser machined steel stencils [145], laminated and spin-on elastomer photoresists
[152,153], PDMS [154] and laser ablation patterned polyurethane with gold nanoparticles
[155].
(a)
(b) (c)
Figure 2.7: Powder blasting with 30 ?m alumina particles. (a) SEM of a powder blasted channel
with through-hole. (b) ARDE with 0.4 mm thick steel stencil mask. (c) Maskless etch profile of a
single sweep with 2 mm diameter blasting nozzle.
200 ?m
2.4 Glass micromachining 27
Scanning of the powder jet is required to etch large surfaces. For uniform etch depth the
scanning path needs to be considered. In addition, the erosion rate should remain constant.
However, there are several factors that can lead to variance in the erosion rate. First of all,
the blast powder quality should be uniform which means little variation in the size and
shape distribution of the grains. Secondly, the feed rate of the powder should not vary. The
moisture content in the powder and compressed air plays an important role here. Too
much moisture leads to agglomeration of the powder causing unstable powder feed and
clogging of the transport lines and nozzle. To avoid moisture-related problems low dew
point (?30 ?C) air dryers in the compressed air line and optional heating should be used.
Maskless sub-millimetre resolution patterning is possible with small powder blasting
nozzles that are available at least down to 250 ?m diameter [156]. However, maskless
powder blasting has limited applicability. A circular powder blasting nozzle results in
Gaussian etch depth with a single sweep (Figure 2.7 (c)) and, therefore, the pattern edges
are not sharp unless a masking layer is used.
Powder blasting is very useful for making through-holes for fluidic interconnections.
For example, a cryogenic cooler made of a stack of three bonded glass wafers has been
made using HF etching for the channels and powder blasting for the fluidic inlet and outlet
holes [157]. However, in several fluidic applications powder blasting cannot be used to
make the fluidic channels due to high surface roughness which affects both fluid flow and
optical properties. Surface roughness of powder blasted surfaces depends on the process
parameters, such as particle size and velocity, but typically the roughness is on the order of
1 ?m [143,144]. Different approaches, such as wet etching and thermal annealing, have
been studied to reduce the roughness of a glass surface [158,159] but their effect is rather a
removal of the damaged and porous surface layer without a significant reduction in the
magnitude of the surface roughness. To conclude, a powder blasted surface cannot
compete with the smoothness of wet or plasma etched channels that result in nanometre
scale roughness when properly optimized. Variation of roughness in glass wet etching has
been studied in [110].
2.4.4 Discussion about glass machining
Other glass micromachining methods in addition to wet etching, RIE and powder blasting
include micro-ultra-sonic machining [111], ECDM and similar spark-assisted etching
methods [73,160,161], laser machining [162-165], FIB [79], embossing [166,167], moulding
[72] and glass blowing [168]. Wafer dicing saws can be set to cut only a part of the substrate
thickness, resulting in straight grooves and crossings that can be used as simple fluidic
channels [169,170]. Width of the channel is determined by the width of the saw blade.
Price, throughput, available aspect ratio and surface quality are important factors for
comparing between the methods. For pricing one has to consider the whole process flow.
For example, wet etching may seem like a cheap approach but the equipment needed for
high quality masking, e.g. a LPCVD silicon furnace, may render the method very expensive
for small production volumes. Powder blasting is a very convenient and economical way of
making through-holes in glass. Wet etching requires expensive masking and the cross-
section of isotropically etched hole is hardly optimal. Glass DRIE can provide high quality
through-holes, but it is expensive and not yet mature technology. However, due to the
mechanical nature of powder basting there is chipping of larger flakes near the edges of the
hole as seen in the cross-section in Figure 2.8. Therefore, DRIE might be the only choice if
28 2 Microfabrication for fluidics
smooth surfaces and anisotropy are required. Mechanical drilling of glass with diamond
coated drill tips usually induces large cracks at the surface near the entrance and exit points
and the wafer needs grinding and polishing afterwards. Also, the lifetime of the tool is
limited to tens or hundreds of holes at most.
In general, mechanical machining methods cannot compete with the surface quality
achieved with chemical etching. The etch selectivity in photopatternable glasses is based on
change in crystal structure, i.e. the development of polycrystalline regions. The surface
roughness at the pattern edges is proportional to the crystal size and might be several
micrometers [57]. Rough surfaces are generally not good either for microfluidic or optical
applications. Increased surface roughness of fluidic channels increase surface area and
adsorption which results in band broadening (i.e. reduced resolution) in chromatographic
and electrophoretic separation channels. Planar surfaces can be polished to reduce surface
roughness. However, the roughness of features below the surface plane is determined by
the selected machining method and very little can be done to reduce the roughness
afterwards.
2.5 Wafer bonding
In wafer bonding, two substrates are attached together (Figure 2.1 (g)). There are
applications for both permanent and temporary bonds. Wafer bonding is used to make
advanced substrates, such as silicon-on-insulator (SOI) wafers [171], layer transfer [172], 3D
microstructures [173] and wafer-level packaging [174]. Several MST devices are impractical
or impossible to implement with mere bulk or surface microfabrication. By attaching two
or more substrates together the device complexity can be easily increased.
There are various wafer bonding methods ? anodic, fusion, glass frit, eutectic,
thermocompession, solder, ultrasonic and (polymer) adhesive bonding ? to name a few.
Wafer bonding procedure starts with cleaning of the wafers followed with alignment and
contacting of the wafers. Depending on the bonding method there may be additional steps
involved such as application of adhesive layers. The choice of the bonding method depends
on the used materials and requirements from the application: temperature tolerance
(formation and service), thermal expansion, surface chemistry, price, throughput, yield,
hermetic or non-hermetic, electrically conductive or insulating. Wafer bonding methods
can be divided in two categories: bonding with or without an intermediate layer.
Figure 2.8: Cross-section of a glass wafer etched with DRIE from top and by powder blasting from
bottom.
100 ?m
2.5 Wafer bonding 29
Two mirror polished surfaces will spontaneously bond to each others with van der
Waals force. Van der Waals bonding is relatively weak but thermal annealing can lead to
formation of strong covalent bonds provided that the substrates have sufficiently matching
coefficients of thermal expansion, the surface chemistry is suitable, the surface polish and
the planarity are good enough and that the trapped particles are few enough.
Different adhesives can be used to fix the substrates together. Low temperature melting
glass paste or metal solders are frequently used when hermetic seals are required. Polymer
adhesives are useful when high temperatures (>200 ?C) have to be avoided, but they are
quite permeable to gasses and hence are not applicable when a hermetic seal is required.
Water glass bonding enables low temperature bonding of glass wafers [175].
The next subsections will present silicon and glass wafer bonding techniques that are
also applied to the nebulizer and LC chips of this dissertation (further discussed in Section
3.1.3). Comprehensive information about bonding can be found in a historical review [176],
a semiconductor direct bonding review [177], an adhesive bonding review [178], a bond
characterization review [179] and several books dedicated to wafer bonding [180-182].
2.5.1 Direct bonding
Polishing methods for silicon wafers provide sub-nanometre surface roughness on a
routine basis and, therefore, silicon direct bonding or fusion bonding of silicon wafers is
fairly straightforward. There are two different mechanisms in silicon-to-silicon fusion
bonding [183]. In hydrophilic bonding there is an oxide layer on the wafers or the last
washing solution leaves the silicon surface hydroxyl group terminated. This will lead to
covalent bonding with an insulating oxide layer between the wafers. Heating of the sample
promotes rearrangement of trapped water and transformation of silanol groups (Si?O?H)
into covalent siloxane (Si?O?Si) bonds at the interface. Siloxane formation reaction
produces water which further oxidises silicon near the interface zone. The remaining
hydrogen can diffuse out through the silicon [184]. In hydrophobic bonding all oxide layers
are removed in HF solution from the bonding surfaces of the silicon wafers prior to
bonding resulting in direct Si?Si bonds during annealing. The necessary annealing
temperature for fully developed bond strength in silicon fusion bonding is similar to
thermal oxidation temperature (around 1000 ?C). The high process temperatures required
limit the applicability of fusion bonding. For example, no organic matter or metals are
allowed and ICs with sub-micron transistors are likely to be destroyed by the diffusion of
the dopants.
Fusion bonding of two glass wafers is also widely applied. The typical annealing
temperature for Pyrex 7740 wafers is around 650 ?C [160,185]. This is above the annealing
point of Pyrex (560 ?C) allowing small glass flow to compensate for surface roughness.
Slow cooling rate is required to relieve strain induced by the temperature change. Typical
glass wafer surface roughness is about 2 nm. Practical considerations in fusion bonding of
glass include elimination of unwanted deformation of the wafers and avoidance of voids.
The wafers should be placed in the annealing furnace on top of a horizontal supporting
plate to prevent wafer bow and also because room temperature bond strength is too weak
to hold the wafers together.
Conventional fusion bonding requires thermal annealing to promote bond strength.
However, bulk comparable bond strengths can also be achieved near room temperature by
chemically activating the surfaces prior to contacting. However, the activation and
30 2 Microfabrication for fluidics
contacting has to be done in a vacuum chamber because the activation will rapidly degrade
at room atmosphere with very rapid adsorption of, for example, moisture. A cluster tool
with chambers for O2 and N2 plasma cleaning and surface activation for room temperature
glass-to-glass bonding is presented in [186]. Compared to conventional fusion bonding,
surface activated low-temperature bonding is more sensitive to surface imperfections
because there is no thermal annealing that allows some material flow to compensate small
surface mismatch. For glass wafers with generally higher surface roughness compared to
silicon wafers this is problematic and thermal annealing is still required for the highest
bond strengths.
2.5.2 Anodic bonding
Anodic bonding is a well established method for bonding silicon to glass. The bonding is
done at elevated temperature where the Na2O in glass is dissociated into sodium and
oxygen ions with Na+ having the highest mobility. A negative voltage is applied on the
glass outer surface causing electrophoretic drift of Na+ ions away from the interface
between silicon and glass wafers creating a strong electrostatic force and leading to
permanent bonding between the substrate. Covalent siloxane bonds are formed at the
interface. Typical bonding conditions for silicon and Pyrex 7740 glass are approximately
350 ?C and 600 V. At this temperature the thermal expansion of glass and silicon is well
matched leading to low stresses at room temperature. The cathode electrode contacting the
glass is subject to corrosion due to accumulating sodium which also reacts with moisture to
form NaOH caustic [182]. Cathodic corrosion of deposited platinum film at the glass
surface will be discussed in Section 3.2.
With proper intermediate layers anodic bonding can also be applied to bond materials
that are not anodically bonded as such. Silicon-to-silicon anodic bonding is possible with a
sputtered borosilicate glass layer on the other bonding surface [187,188]. Glass-to-glass
anodic bonding is possible with thin-films that prevent electrostatic drift of sodium and
thus enable the formation of the depletion layer. Possible materials include standard thin-
films, such as polysilicon, amorphous silicon, nitride and carbide [189,190]. Oxide layer
does not work as a sodium barrier, but it can be used in some multilayer configurations
[189].
2.6 Packaging and chip-to-world interfacing
After everything is ready on the cleanroom processed wafer, it is diced to separate
individual chips (Figure 2.1 (h)). Dicing is a dirty process and it is thus done outside the
cleanroom. Before dicing, the wafer is attached to a frame with a sticky film. The dicing is
done with a fast spinning thin resin blade with 3-axis automated motion. Thin silicon
wafers can be cut with a 20 ?m-thick blade, but other materials and thicker samples require
thicker, e.g. 150 ?m blades.
After dicing the chips are usually packaged before they are ready to be shipped to the
customer. The purpose of packing is to protect the chip from the environment, ease the
handling and to provide chip-to-world interconnections. For ICs, the packaging involves
attaching the die to a carrier, wire bonding the die's contact pads to the carriers metal leads
and final encapsulation. This is illustrated in Figure 2.1 (i). However, MEMS and
microfluidic devices can seldom use standard IC packaging. Special modifications to the
2.6 Packaging and chip-to-world interfacing 31
standard packages or even completely custom made solutions are required. For example, a
pressure sensor cannot be fully encapsulated because its sensing element must be in
contact with the ambient air in order to get the pressure reading.
In contrast to electrical circuits, there is no standard packaging solution for microfluidic
devices. Depending on the application, the requirements may include uncovered surfaces,
connections to tubing, electrical connections and optical viewports. Several microfluidic
chip companies have developed their own connection solutions. The quartz chips of
Affymetrix DNA microarrays are packaged in plastic cartridges that ease the handling of
the chips and contain identification data [191]. Optical fluorescence detection is done after
incubating the chip in hybridization and washing solutions. Agilent's LC?ESI chips are
packed in metal cartridges [41]. Dolomite offers a set of microfluidic standard chips with a
range of matching top and edge connectors [192]. Micronit Microfluidics offers a connector
frame for use with different chips with both fluidic and electrical connectors [193].
Microfluidic ChipShop uses different kinds of commercial fittings, including Luer taper
ports that are glued on the chip [194]. The next section reviews different methods for
fluidic interconnections in more detail.
2.6.1 Fluidic interconnections
Microfluidic devices operate on liquids and gasses and fluidic interconnections are
required to introduce them to the chip and extract them out. Sometimes open sample
reservoirs are sufficient for introducing liquids by pipetting. Open sample wells also allow
the use of electric probe needles for applying electric potentials needed, for example, with
CE chips. Sometimes open sample wells are sufficient but very often capillary tubes are
connected to the chip. Figure 2.9 illustrates these basic approaches.
Several aspects need consideration when planning fluidic interconnections: influence
on the chip fabrication process, pressure tolerance, dead volume, material properties such
as chemical compatibility and temperature tolerance, required space, ease of operation,
reusability and price. Due to the nature of the machining processes, the available inlet port
geometries are different depending on whether it is located on the top or edge surface of
the chip. Dead volume ? unnecessary channel volume ?has detrimental effect to the
device performance and it needs to be minimized. For example, if the composition of the
inflow is changed, the earlier composition stored in the dead volume will slowly mix to the
new one through diffusion. In chromatography, dead volumes lead to band broadening
and peak tailing.
Different fittings for tube connectors for industrial and chromatography applications
are available from various manufacturers, such as Svagelok [195], Upchurch Scientific1
[196] and Valco Instruments [197]. Some of these fittings and the applied principles are also
applicable to making fluidic connections to chips. A typical fitting consists of a nut that
compresses a ferrule against the tubing and the body of the connector for gripping the tube
in place and for sealing the connection. There are various types of nuts and ferrules
available. In general, coned fittings offer higher pressure tolerance compared to flat-bottom
fittings. The pressure tolerance depends on the properties of the used materials. Higher
pressures mean that harder materials have to be used because of the pressure induced
deformation of soft tubing and ferrules. Metal tubing and fittings offer the highest pressure
tolerance, but on the other hand plastic ones have several benefits, too [198]. A variety of
1 Part of IDEX Health & Science LLC.
32 2 Microfabrication for fluidics
different polymer fittings are available with extensive chemical compatibility range,
reusability and competitive pricing. In addition, polymer tubing is easy to cut.
There are various approaches for chip-to-world interfacing as shown in a review [199].
Exemplary approaches for connecting tubes with microfluidic chips are illustrated in
Figure 2.10. A simple approach is to push the tube against an inlet hole on the chip surface
(Figure 2.10 (a)). Gluing is usually required for making the connection leak-tight provided
that the glue is chemically compatible with the application. However, gluing is laborious
manual work and involves a great chance of miss-alignment or clogging the tube or chip
with the glue. The inevitable loss in the yield of glued connections renders the approach
unsuitable for microfluidic devices with high number of fluidic connections. In addition,
the permanent nature of gluing prevents alterations in the configuration. Gluing can be
avoided by using pliant silicone tubing and a special fixture that compresses the tubes
against the chip [200]. Polymer films can be used for sealing [201]. The are also different
connectors that can be attached on the chip either by gluing [202], anodic bonding [203], or
ultrasonic welding [204].
Problems with alignment and clogging by glue can be alleviated by making a recess in
the chip which closely matches the outer diameter of the tubing as shown in Figure 2.10 (b).
Hourglass shaped holes have been machined to glass using ECDM [205]. Different flanging
Figure 2.10: Examples of different solutions for connecting tubing to chip: direct coupling (a, b),
NanoPort connectors (c, d), custom made fixtures (e?f) and a PDMS connector.
(a) (b) (c) (d)
(g)(f) (h)
(e)
Figure 2.9: (a) Open sample wells allow simple application of liquid droplets and the use of probe
needles for electrical connections. (b) Tubing can be connected from the edge or surface of the chip.
(a)
V
? +
(b)
2.6 Packaging and chip-to-world interfacing 33
processes can be used to reshape the end of the tube for sealing [206,207]. Thermoplastic
cyclic polyolefin2 (COC) has been used to make direct sealing with metal needles [208].
Unfortunately, depending on the fabrication process it is not always possible to make
tightly fitting slots for the tubing. Also, a simple slot might not provide sufficient grip to
hold the tubing in place.
One of the few commercial general purpose off-the-shelf solutions for microfluidic
interconnections are the NanoPort connectors from Upchurch Scientific [209]. Two different
NanoPort configurations are shown in Figure 2.10 (c) and (d). They compose of a cylinder
block with internal threading that is glued on the chip surface. Different NanoPort
assembly configurations allow the use of either coned or flat-bottom fittings and they can
also be used as sample reservoirs. NanoPort connectors enable repeated attachment and
release of the tubing, but because they are glued on the chip they usually have to be
discarded with the chip. Therefore, NanoPorts can be expensive solution if the lifetime of
the chip is short. With the preformed adhesive ring the NanoPort connectors take up about
a 10 mm diameter area on the chip surface. With this and the general issues with gluing in
mind, NanoPorts are not feasible for high density inlets.
An alternative to gluing of fluidic ports to the chip is to use custom made chip holders.
Typically, a two part approach is used with top and bottom plates accommodating a
housing for the chip and fixed positions for the fittings. The same fittings can be used as
with NanoPort connectors, but there is no need for gluing and less surface area is required
for the connection possibly enabling reduction of the chip size and the fabrication of more
chips per wafer. On the other hand, a chip holder might take up more space than a
NanoPort connector which may be of importance in some applications. Figure 2.10 (e)
shows a chip holder with a flat-bottom fitting and (f) with an o-ring. Reusable magnetic
connectors were recently published [210]. They offer a freely configurable alternative to
dedicated holders for each chip design that can also be used with whole wafers.
Density of inlets using standard nut fittings is limited by the diameter of the nuts. For
higher density connections the tubes cannot be brought in direct contact with the chip as
shown in Figure 2.10 (g). A thin sealing membrane between the frame and chip offers
higher pressure tolerance [211]. Alternatively, o-rings can be used [212-214].
Edge connections can potentially greatly simplify the fabrication process because there
is no need for the through-holes. Fairly good edge surface quality can be achieved in wafer
dicing. Standard fittings require some space and can be only be used with fairly thick
chips. For example, Dolomite uses 4 mm thick chips with their edge connectors. A more
universal approach is to insert the capillary inside the channel and settle for gluing. UV
curable glue can be used to accurately control the penetration depth of the glue into the
capillary insertion channel [215]. Reversible edge connections using PDMS as the sealant
and small metal wire to prevent clogging have been demonstrated [216] but this method is
hardly universal.
The PDMS connector of publication I introduced in Section 1.4.1 utilizes the self-sealing
nature of PDMS that provides reversible seals between PDMS and tubing, and PDMS and
the chip. The PDMS connector is moulded on a master with matching surface textures of
the PDMS connector and the fluidic chip enabling easy alignment of multiple tubes
simultaneously. The inverted pyramid profile of KOH etched <100> silicon is particularly
useful for this purpose. The matching surfaces of silicon pyramids are also used in the
2 Also known as cyclo olefin polymer (COP).
34 2 Microfabrication for fluidics
fluidic couplers [202]. In publication I the holes for the capillary tubes are made using
dummy capillaries during moulding. Others have made PDMS connector blocks with
punctured holes [217] or machined pillars on a PMMA mould [218].
Interconnection schemes for the HN chips will be further discussed in Section 3.3.
35
3 Heated nebulizer chips
The HN chips are used to introduce a vaporized and ionized sample to a mass
spectrometer. This chapter will cover the different fabrication processes used and discuss
the different chip designs. Also the chip lifetime, jet shape and the different analytical
applications will be discussed.
3.1 Different chip designs and fabrication processes
The HN chips have been implemented using various layouts, materials and fabrication
processes. Photographs of four different nebulizer chips are shown in Figure 3.1. The
author has made the chip in some 40 different variations since taking over their
development work. Before this, the chips had been made from wet etched [52] and DRIE
etched [56] silicon with anodically bonded glass cover. During this dissertation work,
different glass microfabrication processes were explored resulting in all-glass HN chips.
Different nozzles and their influence on the jet shape were studied. A meander pre-heating
channel for the nebulizer gas was realized. Chip size was be reduced thanks to new
interconnection methods.
The fabrication of all the different versions of the nebulizer chips follows the same
process outline with two main process stages: (1) double sided through-wafer etching of a
wafer with nebulizer channel, nozzle and fluidic interconnections and (2) bonding to a
Figure 3.1: Different heated nebulizer chips: (a) original silicon?glass nebulizer chip (top and
bottom views), (b) improved silicon?glass chip, (c) wet etched glass chip and (d) glass chip without
NanoPort for use with a chip holder (top and bottom views). (Photos courtesy of Markus Haapala.)
(a)
18
m
m
29 mm
Nanoport connectors
heater
contacts
nebulizer gas in
sample in
nozzle
heater
(c)(b)
10
m
m
18 mm
10
m
m
26 mm
Nanoport for
nebulizer gas nozzle
heater
contacts
sample
capillary
Nanoport for
nebulizer gas nozzle
heater
contacts
sample
capillary
(d)
5 m
m
33 mm
heater
nozzle
sample
capillary
through hole for
nebulizer gas
nebulizer gas pre-heat channel
36 3 Heated nebulizer chips
cover wafer with a heater element. Figure 3.2 illustrates the fabrication process using wet
etching of a glass wafer. The upcoming subsections will discuss the different materials and
fabrication processes used for the nebulizer chips.
3.1.1 Substrate materials
The nebulizer chips have been fabricated using either one silicon and one glass wafer
anodically bonded together or two glass wafers. Because the on-chip operating
temperatures of the nebulizer chip typically range from 300 ?C to 500 ?C polymers
substrates are out of the question. For silicon and borosilicate glass this temperature range
is suitable.
Surface chemistry and chemical compatibility of silicon and glass are similar. However,
thermal conductivity values are very different. Thermal conductivity of silicon is about two
order of magnitudes higher than that of glass. Thermal images of silicon?glass and all-glass
chips are shown in Figure 3.3. Glass allows much more localized heating with high on-chip
temperature gradients whereas silicon will efficiently spread out the heat over the whole
surface. This has implications, for example, for the durability of fluidic interconnections
that may involve the use of materials with lower temperature tolerance than silicon and
glass. Even though all the chips in Figure 3.3 are heated with the same power and the same
amount of gas flows through the channels, there are considerable differences with the peak
temperatures of the chips. In addition to the material choice, the layout of the heater also
affects the peak temperature. On the meander channel chip the heating power is
distributed over a larger area resulting in much lower peak temperature compared to the
chip with the straight channel chip with a narrow heater. A photo of the meandering chip is
shown in Figure 1.5 on the left.
Figure 3.2: Fabrication process of wet etched glass nebulizer chip: (a) LPCVD silicon deposition, (b)
double sided lithography, (c) silicon wet etching, (d) glass wet etching, (e) photoresist and silicon
stripping, (f) bonding to another glass wafer, (g) platinum sputtering, (h) lithography, platinum
etching, (i) photoresist stripping and (j) dicing. A sample capillary is inserted from the edge of the
chip and sealed with high temperature epoxy glue.
(a)
(c)
sample in
nebulizer gas in
(b)
(d)
(e)
(f)
(g)
(h)
(i)
(j)
3.1 Different chip designs and fabrication processes 37
3.1.2 Channel fabrication processes
Different etch processes can be applied to both silicon and glass wafers as discussed in
Sections 2.3 and 2.4. Figure 3.4 shows scanning electron microscope (SEM) images of
nebulizer chip nozzles produced by different fabrication methods. This subsection
compares the different methods applied for making the nebulizer chips.
Wet etched glass
Fabrication process for nebulizer chips with wet etched channels are described in
publication II. Chips made with a similar process are also used in [1-6] and in the jet shape
measurements of publication IV. The process is outlined in Figure 3.2. LPCVD silicon hard
Figure 3.3: Infrared thermal images with peak temperatures (Tmax) of HN chips with 4 W heating
power and 150 sccm nitrogen flow. Silicon?glass chip size is 18 mm ? 10 mm and both all-glass
chips are 26 mm ? 10 mm.
silicon?glass chip
from heater side
silicon?glass chip
from NanoPort side
all-glass straight channel chip
from heater side
all-glass straight channel chip
from NanoPort side
all-glass meander channel chip
from heater side
all-glass meander channel chip
from NanoPort side
Tmax = 274??C
Tmax = 508??C
Tmax = 406??C
Tmax = 610??C
Tmax = 329??C
Tmax = 474??C
Figure 3.4: Nebulizer chip nozzles produced with various fabrication processes: (a) anisotropic wet
etching of silicon, (b) DRIE of silicon, (c) wet etched glass, (d) powder blasted glass and (e) DRIE
etched glass channel with powder blasted through-holes.
400 ?m 600 ?m
(a) (c)(b)
400 ?m 400 ?m 400 ?m
(d) (e)
38 3 Heated nebulizer chips
mask deposition (about 400 nm thick) was ordered from CSEM (Neuch?tel, Switzerland).
Standard double sided lithography (AZ5214E resist, AZ Electronic Materials GmbH,
Germany) follows with wet etching of the silicon hard mask (HNO3/NH4F/H2O, 44:1:18),
glass wet etching (HF/HCl, 10:1), photoresist stripping (ultrasonic acetone bath) and hard
mask removal (25% TMAH, 80 ?C). Alternatively the photoresist can be removed after the
silicon etching prior to glass etching, but the photoresist layer can potentially help to
reduce pinhole defects. The isotropic silicon etchant also etches glass and leads to higher
surface roughness compared to stripping with TMAH [108]. Without stirring of the glass
etch solution visible waviness on the channel surface will result. The etch process is
endothermic and the magnitude of natural convection in the solution depends on the area
of the exposed surface. Therefore, continuous stirring of the solution is important for
uniform etch depth. Ultra- or mega-sonic agitation can also be used for improving etch
uniformity [219].
The nozzle shape obtained with this process is shown in Figure 3.4 (c). The isotropy is
very good with almost perfect radial etch profile. Using this process with simultaneous two
sided etching the wafer thickness defines the channel depth and minimum channel width.
Some over-etching is required to ensure proper through-etching throughout the wafer. For
standard 500 ?m thick glass wafers the channel depth will be about 250 ?m and the
channels will be about 500 ?m wider at the wafer surface compared to the channel width
on the mask. If the mask openings are too small the etch rate may be locally reduced
because of limited transport of fresh etchant to the surface. For the 250 ?m deep etch 50 ?m
wide openings are large enough to be reproduced without problems. Isotropic etching can
produce at best 1:2 (height:width) aspect ratio which gives little room to play with nozzle
shape and possible extra structures monolithically integrated to the channels.
Wet etched silicon
The first nebulizer chips were fabricated using anisotropic wet etching of silicon. Both
TMAH [52,53,56] and KOH [220] etching was used with 600 nm?1 ?m thermal oxide hard
mask material. The oxide mask was patterned on both sides via lithography and etching.
Typically, about 90% of the oxide thickness was first etched with RIE and then the
remainder with BHF. Compared to etching fully with BHF this approach results in better
edge profile which has some effect on the resulting channel profile as discussed in Section
2.3.2. However, the nebulizer chips do not have very high aspect ratio structures so this
phenomena has little practical significance. Alternatively, LPCVD Si3N4 (~100 nm) was
sometimes used as the hard mask instead of thick thermal oxide. Nitride has a lower etch
rate in both TMAH and KOH compared to oxide. However, nitride has to be patterned
using RIE and nitride stripping requires either RIE or 160 ?C phosphoric acid, which is
complicated compared to oxide stripping in HF. Also, for good etch results a thin oxide
layer underneath nitride is required to reduce stress and to improve nitride adhesion. All
and all, the use of a simple thick enough oxide hard mask is favoured over nitride.
For typical 380 ?m thick DSP wafers, the maximum outer diameter (OD) of the sample
capillary is 190 ?m. Although capillaries with OD less than 90 ?m are available there was
an interest to use commercially available 220 ?m OD deactivated capillaries that minimize
the possibility of unwanted surface effects involved with untreated capillaries. However,
the use of thick capillaries would have required either the use of thicker DSP wafers or two
phase etching to enable other than half wafer thickness channel depth. Alternatively, the
3.1 Different chip designs and fabrication processes 39
etching can be continued after penetration of the wafer but this adds to the undercutting of
the nozzle and other convex corners. KOH with added IPA has smaller undercutting and it
was used to etch 240 ?m deep structures from both sides of 380 ?m thick silicon wafer
[218]. Reduced undercutting of KOH+IPA also enabled a capillary stopper structure in the
nebulizer channel which could not be made with TMAH as shown in Figure 3.5. The
purpose of the capillary stopper was to provide repeatable positioning of the sample
capillary. However, the capillary stoppers were left out after a couple of chip generations
because the influence of the capillary positioning was yet to be determined and the
structures could interfere with the sample and the nebulizer gas flow at the capillary tip
with an effect on the stability of the mixing. Unfortunately, it is very difficult to get definite
experimental verification of these and many other effects. The absolute ion intensity in the
mass spectra is affected by a plethora of different factors, such as, ambient conditions, exact
positioning of the ion source with respect to the ion inlet of the mass spectrometer and the
electrical drift of the detector. In quantitative analysis with mass spectrometers calibration
is established by a reference standard with known composition and concentration mixed in
the sample.
DRIE etched silicon
The DRIE etched silicon process is very similar to the wet etched silicon process except the
etching is done using DRIE. Again, thermal oxide is good masking material but there is no
need to remove the photoresist after oxide patterning. DRIE etched silicon?glass nebulizer
chips are used in [7-10,56].
Anisotropic DRIE etching offers freedom of layout because there will be very little
etching under the mask. The channels have rectangular cross-section. Especially the nozzle
design can be easily tailored and there is no need for corner compensation structures like in
anisotropic wet etching. Sharp and tapering nozzles are easy. As DRIE is one sided process,
there is no problem to etch different depths from each side. For the 220 ?m OD deactivated
capillaries, the target depth of the channels was 250 ?m to give some tolerance.
Powder blasted glass
A nebulizer chip nozzle etched with powder blasting is shown in Figure 3.4 (d). A
photopatternable LF55GN elastomer was used as masking material and powder blasting
was done in a custom-made chamber equipped with sample stage and nozzle oscillators. A
(a) (b)
Figure 3.5: Nebulizer gas inlet hole and capillary stopper structure etched in (a) KOH+IPA and (b)
TMAH. The dashed and the solid lines indicate the original etch mask edges on the inlet and channel
sides of the wafer, respectively. (Micrographs courtesy of Kestas Grigoras.)
10
00
?m
10
00
?m
40 3 Heated nebulizer chips
Texas Airsonics, Model HP-2 powder blasting machine was used with 30 ?m alumina
particles (EKF 320, Strahltechnik Bachmann AG).
Compared with wet etching of glass, powder blasting allows higher aspect ratios. This,
combined with possibilities from oblique angle blasting, enables the micromachining of
more complex structures. Also, powder blasting is economical compared to wet etching
thanks to easy masking processes available. However, for the nebulizer chips the rough
surface of the powder blasted nebulizer channel could be detrimental to the analytical
performance of the chip. Sample droplets from the inlet capillary are pushed to the
sidewalls where they spread on the rough surface and are flash evaporated by nucleate
boiling. The analytes with the highest boiling point may desorb slowly causing a memory
effect. Also, the hot surface may cause thermal degradation of the analytes temporarily
dried on it. A smooth inner surface of the nebulizer channel is better because if the surface
is hot enough it can provide vapour film boiling of the sample droplets where the droplets
bounce of the channel surface due to the Leidenfrost effect. Much less thermal degradation
of the analytes will results as most of the droplet evaporation will take place in gas phase
with heat transfer from the nebulizer gas. These issues are further discussed in [16].
Unfortunately, no powder blasted chips were made ready for testing with a mass
spectrometer to get experimental verification whether the added surface roughness makes
a difference or not.
DRIE etched glass
The DRIE of glass was also applied for making nebulizer chip channel wafers as presented
in publication III and Figure 3.4 (e). Out of the three glass micromachining methods used
in this work, glass DRIE offers the highest aspect ratios and most flexible channel and
nozzle geometries similar to silicon DRIE. A silicon wafer shadow mask was used to mask
the DRIE etching from the channel side. Etching from the other side was done with powder
blasting. Back side alignment for LF55GN lithography for powder blasting masking was
easier compared to alignment of a shadow mask for DRIE. Also, through-etching with
DRIE would require additional protection for the sample stage, for example, a dummy
wafer underneath the sample wafer.
Glass DRIE can provide very low surface roughness and there are no apparent reasons
why nebulizer chips would not perform well in their applications. Unfortunately, the
prepared wafer broke during bonding and no chips could made ready for testing with a
mass spectrometer. For the current nebulizer chip designs there are no foreseeable
advantages resulting from using glass DRIE and the process is expensive and cumbersome
compared to wet etching of glass. In contrast, for a nebulizer chip with integrated LC
column glass DRIE could be advantageous as will be discussed in Section 4.2.
3.1.3 Cover wafer bonding
For nebulizer chips with the channel wafer made of silicon anodic bonding of a glass cover
wafer is an obvious choice as it is a fairly routine microfabrication process as discussed in
Section 2.5. Anodic bonding provides very strong bonding and it is not too sensitive to
surface roughness or minor particle contamination on the bonding surfaces. For the
nebulizer chips an RCA-1 cleaning step was done to both silicon and glass wafers prior to
bonding. Anodic bonding can also conform to thin (~100 nm) heater elements between the
wafers. The on-chip heater will be further discussed in Section 3.1.4. Anodic bonding often
3.1 Different chip designs and fabrication processes 41
results in minor corrosion (visible staining) on the glass (cathode) surface. The corrosion
depends on the local electrical contact quality between the glass and the bonder electrode.
Corrosion results from sparking due to poor contact and from sodium agglomeration.
Fusion bonding of two glass wafers was used for all glass nebulizer chips. The wafers
are RCA-1 cleaned, dried, contacted and thermally annealed at 650 ?C resulting in very
strong bonding, comparable to bulk glass. During annealing the glass wafers are placed in
a horizontal position on a quartz plate with a non-polished surface. Polished surface quartz
plates would bond with glass and the wafers would break during chamber cool down due
to about one order of magnitude difference in thermal expansion between silicon and
quartz. If cleaned wafers are directly put on top of each other for annealing there will be a
varying amount of voids between the wafers. To reduce the voids a pre-bond step in a
vacuum bonding chamber with compression and heating up to 500 ?C was found to
provide good results. Using higher temperatures with compression would result in
imprinting of the bonder platen patterns (the alignment view ports, for example) on the
glass. Furnace annealing at 650 ?C after the pre-bond step was still required to increase the
bond strength.
Also anodic bonding experiments between two glass wafers with intermediate silicon
thin-film were conducted as mentioned in publication II. Initial experiments with PECVD
deposited a-Si lead to poor bond strength presumably due to out gassing from the film.
The LPCVD silicon could provide stronger bonding but the over hanging silicon hard mask
on channel edges imposes some problems. During water rinsing after glass etching the
hanging silicon tends to crack away with poor edge quality affecting the bond interface at
channel edges. Overall, fusion bonding is preferred.
3.1.4 On-chip heater
The nebulizer chips have thin-film metal resistors that are used for heating. In initial
prototypes aluminium was used but these heaters had a very short lifetime even with low
heating powers. In addition, the heater was directly in the vaporization channel of the chip
and was exposed to the vapours of the sample. Some analytical applications use dilute
caustics as a solvent which is not a good combination with aluminium because they are
aluminium etchants. Therefore, the heater material was changed to platinum which
provided much longer heater lifetime. Typically, the platinum heaters of the all-glass HN
chips have lifetimes of one to a couple of weeks corresponding to operating times from tens
to hundreds of hours, which can be considered sufficiently long.
Typically, a 300 nm thick sputtered platinum with a 17 nm thick chromium adhesion
layer was used as the heater. The sheet resistance of this kind of heater is around 1 ?/sq
providing a reasonable heater resistance from 100 ? to 200 ? depending on the heater
layout. This resistance range is suitable for the typical 3 W?5 W heating powers with
ordinary 60 V laboratory DC power sources.
Thin-film platinum heater elements are used in various MEMS devices, because
platinum tolerates high temperatures well and has fairly constant temperature coefficient
of resistivity over wide temperature range. However, platinum is also used as a catalyst for
various chemical reactions. With this in mind it did not seem like a good idea to have
platinum in contact with the sample flow. Therefore, the heater was moved away from the
channel to the outer surface of the chip. Naturally, this slightly reduces the energy
42 3 Heated nebulizer chips
efficiency of the heater because the heat must be conducted through the glass cover wafer
to the nebulizer channel.
Another benefit of having the heater on the outer surface is that the heater can be made
after wafer bonding. Anodic bonding can easily conform to approximately 100 nm thick
heater film between the wafers, but a thicker film ? that would have both lower resistivity
and longer lifetime ? would result in voids. This problem can be solved by embedding the
heater in the surface level of the wafer by etching a recess before deposition but this adds
the process complexity and requires good film thickness uniformity [122]. Patterning the
heater elements after the bonding has the advantage that no alignment is needed for the
bonding. Also, anodic bonding causes cathodic corrosion (further discussed in Section 3.2)
and spark erosion of platinum.
Both lift-off and etch patterning were applied to the platinum heater. Typically, etch
patterning was used and it was done in diluted aqua regia (3:1:2 HCl:HNO3:H2O, 70 ?C,
etch rate some tens of nm min?1) with AZ5214E photoresist. A chromium adhesion layer
was etched in a ceric ammonium nitrate-based etchant (200 g of Ce(NH4)2(NO3)6 + 50 ml of
HClO4 + 1090 ml of H2O). Wet etching of platinum is quite a sensitive process because even
a small amount of oxidized platinum at the surface inhibits dissolution of platinum in aqua
regia [221]. A sputtered platinum film would usually etch without problems, but
evaporated platinum had such a slow etch rate that the photoresist was dissolved faster.
This could be due to some remaining oxygen in the evaporation chamber. Also, wet etching
of platinum that was deposited prior to anodic bonding was unsuccessful. With lift-off it is
important to have suitable photoresist and avoid its overheating during deposition. If lift-
off patterning is done after bonding there is a danger that the channels will be
contaminated with released platinum particles. By using etch patterning particle
contamination of the channels is avoided, but the channels are filled with the etchant and it
is somewhat laborious to properly rinse them.
For glass-to-glass fusion bonding the coefficient of thermal expansion between glass
and platinum poses a problem. An experiment to bond glass wafer with platinum resulted
in large voids throughout the wafer due to the deformation of the glass with the platinum
heaters. The problem could be overcome with low temperature glass-to-glass bonding or
by using compression during the annealing, but the necessary machinery was not available
in our facilities. The surface texture of the platens of the compressing machine will be
imprinted in the surface of glass and surface polishing may be required after the bonding if
the surface quality is an issue. Fortunately, the performance of the HN chips is not affected
by the added roughness on the outer surface. Integrated metal electrodes between glass
wafers have been demonstrated in [122] with a block of alumina placed on top of the
aligned wafers during annealing.
In addition to platinum, also aluminium, copper, tungsten, chromium and gold thin-
film heaters were tried out. Out of these metals, gold was the only one that showed a
similar temperature tolerance and lifetime as platinum. However, pure gold is soft and
susceptible to scratching. With ordinary handling gold heaters would often fail
prematurely because of scratching. Therefore, platinum seems like the best choice.
3.2 Lifetime and failure mechanism of the platinum heater 43
3.2 Lifetime and failure mechanism of the platinum heater
Earlier studies have identified different phenomena that play a role in the failure of
platinum heaters: electromigration, stress-induced morphological changes
(recrystallization), interlayer diffusion and chemical reactions [213]. For investigating the
failure mechanism and the lifetime of the platinum heater used in the HN chips a set of test
wafers were prepared. The layout of the heater test chips is shown in Figure 3.6 (a). The
influence of three different adhesion layers (Cr, Ti, Ta) and different thermal annealing
temperatures (no annealing, 450 ?C, 600 ?C) were tested for platinum heaters on Pyrex
glass wafers. Annealing time was 30 minutes and it was done in nitrogen atmosphere with
?10 ?C min?1 temperature rampings. The test chips were operated with a computer
controlled power source with data logging of the current, voltage and the the chip
temperature measured with a thermocouple. The chips were operated at ambient air with
6 W power corresponding a temperature of about 550 ?C at the centre of the chip. This is
higher than the typical operating temperature of the chips and it was chosen to accelerate
testing. The measured lifetimes are shown in Figure 3.6 (b) and various other properties of
the samples in Table 3.1.
The rate of corrosion depends on the applied electric field. The resistances of the tested
Cr/Pt films were lowest resulting in lowest operating voltages and longest lifetimes.
150 ?m wide heaters were also tested and their lifetimes were only 20 %?40 % of the
lifetime of the 300 ?m wide heaters. The narrow heaters have higher resistance and they
require higher operating voltage for the same heating power. In all chips the failure
of the heater took place near the contact pad for negative electrode, i.e. the cathode end.
There was visible smudge on the platinum film around the failure point. Figure 3.7 shows
an SEM and an energy dispersive spectroscopy (EDS) elemental analysis map of a failed
platinum heater. The EDS analysis revealed the smudge crystals are mainly composed of
sodium, carbon and oxygen ? probably in the form of sodium carbonate (Na2CO3). Other
sodium compound such as Na2O, Na2O2 and NaOH may also be present but eventually
they should react with the atmospheric CO2 to form sodium carbonate. Unfortunately, EDS
does not provide reliable quantification of light elements (carbon, hydrogen and oxygen)
and other measurements would be needed to get verification of the exact composition of
the smudge. A few microlitre drop of phenolphthalein pH indicator solution turned pink
on the corroded platinum surface indicating an increase in the pH of the drop. This is in
good agreement with the proposed composition of the smudge as all of the above
mentioned sodium compounds are bases. It is well known that at elevated temperatures
Na+ ions in glass become mobile and will drift in an applied electric field. This
phenomenon is the basis of anodic bonding as discussed in Section 2.5. Sodium ions
migrate towards the cathode, diffuse through platinum film and react with ambient air at
the surface causing corrosion on the platinum. Galvanic corrosion of platinum in molten
Na2CO3 around 500 ?C has been reported and oxidation of the platinum at the cathode,
transport to the anode and reduction to metallic platinum at the anode was proposed as an
explanation [222]. Because in the case of the heater there is no separate anode, this can also
be interpreted to be electromigration or recrystallization of platinum. Naturally, there is
also considerable compressive stress in the platinum film which can be relieved through
hillock formation. However, because the heaters always fails at the cathode end, the
sodium agglomeration undoubtedly plays an important part in the failure mechanism.
44 3 Heated nebulizer chips

(a) (b)
Figure 3.6: (a) Heater test chip layout (5 mm x 25 mm). DC power source was connected to large
contact pads and voltage was measured via reference electrodes to determine the resistance of the
heater outside the contact area. (b) Lifetime of the individual 300 ?m wide Pt-heaters (three of a
kind) at 6 W operating power with different adhesion layers (Cr, Ti, Ta) and post deposition
annealing temperatures (20 ?C, 450 ?C, 600 ?C).
Table 3.1: Prepared test chips: Cr/Pt by sputtering, Ti/Pt and Ta/Pt by e-beam evaporation.
Cr/Pt Ti/Pt Ta/Pt
Nominal adhesion layer thickness 17 nm 10 nm 10 nm
Nominal Pt layer thickness 200 nm 150 nm 150 nm
Profilometer measured total thickness 211 nm 180 nm 208 nm
Average resistance (a) of chips without thermal annealing 197 ? 302 ? 313 ?
Average resistance (a) of chips annealed at 450 ?C 197 ? 286 ? 338 ?
Average resistance (a) of chips annealed at 600 ?C 200 ? 273 ? 345 ?
(a) at 6 W operation
(a) (b)
Figure 3.7: (a) SEM of a platinum heater after operation showing small agglomerated crystals and
(b) EDS map from the same area with silicon showing in green, sodium in red and platinum in blue.
100 ?m100 ?m
3.2 Lifetime and failure mechanism of the platinum heater 45
The electrophoretic agglomeration of sodium could be avoided by using sodium-free or
low sodium containing glass. For example, quartz glass is pure SiO2 and has much higher
softening temperature compared to Pyrex which, theoretically, should enable the use of
higher operating temperatures. However, the thermomechanical properties of quartz glass
are inferior to Pyrex. Heater test chips made on quartz glass would spontaneously break
during heating due to stresses from on-chip temperature gradients or due to phase
transition that takes place for single crystalline ?-quartz at 573 ?C. Redesigning a heater
that would provide better temperature uniformity might enable the use of quartz glass.
However, quartz-to-quartz bonding is more difficult than glass-to-glass bonding [223,224].
Alternatively, sodium diffusion from glass can be reduced with a diffusion barrier layer
(e.g. Si, SiNx or SiCx) between glass and heater as in glass-to-glass anodic bonding
[189,225,226].
With the conducted test set, no decisive conclusions can be drawn regarding the effect
of the adhesion layer or annealing. For sputtered Cr/Pt films the annealing had no effect,
evaporated Ti/Pt film shows reduction in resistance and Ta/Pt film increase in resistance.
For titanium adhesion, the average lifetime shows approximately three-fold increase for
chips annealed at 600 ?C compared to chips without any post deposition annealing.
To maximize the lifetime of a platinum heater on a glass substrate one should use as
thick a platinum layer as possible, minimize local electric fields in the layout of the heater
and make the cathode end wider so that it takes longer to fail by corrosion. A diffusion
barrier either for the sodium from the glass or the oxygen from the air might help, but
could also lead to issues with adhesion or film stress.
3.3 Fluidic and electrical interconnections
The nebulizer chip requires connections for sample and nebulizer gas flows and electrical
connection to the heaters. Various approaches for making the chip-to-world connections
have been applied. The first nebulizer chips had three NanoPort connectors (Figure 3.1 (a)):
one for the sample flow and two for the nebulizer gas. This was not an economical
approach for two reasons. Firstly, NanoPort assemblies are fairly expensive (around ?20
each) and they are difficult to reuse. Secondly, three NanoPorts take plenty of surface area
reducing the number of chips per wafer. A two NanoPort design with only one NanoPort
for nebulizer gas connection behind the sample NanoPort was demonstrated in [56] but it
was almost immediately replaced with a single NanoPort design.
In the single NanoPort design a sample capillary is inserted directly into the nebulizer
channel from the chip edge and sealed with high temperature tolerant epoxy glue (Duralco
4703, Cotronics). The NanoPort is used for the nebulizer gas. A glass nebulizer chip with
the single NanoPort approach is shown in Figure 3.8 (a).
As discussed in Section 2.6.1 there are several possibilities to make top surface fluidic
connections. Different chip holders that used flat-bottom connectors were used to replace
the NanoPort connector for nebulizer gas. The reversible nut-and-ferrule gas connection
eases the use of the nebulizer chips as only the sample capillary needs to be glued to the
chip and the chip can easily be replaced in the set-up, when needed. The holder material
has to tolerate the on-chip temperatures, preferably up to 500 ?C. Also, it must not short
circuit the heater on the chip surface. A monolithic holder made of machinable glass
ceramic (MACOR, Corning Inc.) is shown in Figure 3.8 (a). MACOR is a non-porous high
46 3 Heated nebulizer chips
end material that can be machined with ordinary metal working tools and it withstands
continuous use at 800 ?C. However, it is brittle and requires careful machining whereas
metals, such as aluminium, are much easier to machine. An aluminium-MACOR two piece
holder is shown in Figure 3.8 (b). The latest chip holders (Figure 3.8 (c)) are made from high
temperature tolerant polyimide polymer (Sintimid NT, Ensinger Sintimid GmbH) and they
have an o-ring (Perlast G75B, Perlast Ltd.) seal for the tubing. A schematic cross-section of
this holder is shown in Figure 2.10 (f).
For electrical connection to the heater it is possible to solder wires directly to the heater.
Although the peak temperature on the chip is higher than a typical solder melting point,
the inlet end of the chip remains cooler. Nevertheless, it may be necessary to use a high
melting point solder depending on the chip layout and operating power. For solder-less
connections the recent chip holders are equipped with gold plated spring loaded
connectors (90041, Preci-Dip SA; cross-section in Figure 3.8 (d)) that provide easy and
reversible heater contacts.
Publication I presents a fluidic inlet scheme that used PDMS to seal the tubing against
the chip. Unfortunately, this approach was not compatible with the heated chips as the
outgassing from the PDMS leads to a significant increase in the signal background noise
and contaminates the mass spectra.
3.4 Jet shape and temperature
The jet shape produced by the nebulizer chips influence both ionization and ion collection
efficiencies. Therefore, knowing and controlling the jet shape is important. However, it is
(a) (b)
(c) (d)
Figure 3.8: Different interconnection schemes for the nebulizer chips. (a) NanoPort connector and a
monolithic ceramic holder with flat-bottom fitting. (b) Aluminiumacor holder with flat-bottom
connector. (c) Polyimide holder with integrated o-ring seal. (d) Cross-section of the spring loaded
electrical connector used in (b) and (c).
3.
5 m
m
1.83 mm 1/8? mounting rod
nebulizer
gas tube clamping screws
heater cables
sample capillary
(not v isible)
1/8? mounting rod
nebulizer
gas tube
clamping screwsheater cables
flat-bottom fitting
sample capillary
(not v isible)
flat-bottom fitting
sample capillary
nebulizer gas tube
NanoPort fitting
monolithic
connector block
3.4 Jet shape and temperature 47
not trivial to measure the shape of a small gaseous jet. Optical contrast between ambient air
is too weak for direct photography. Liquid fluorescent droplets containing spray can be
imaged with fluorescence microscopy, but fluorescence intensity from a fully gaseous jet is
too weak for conventional fluorescence image sensors. Methods demonstrated for
measuring gaseous jets are planar laser-induced fluorescence (PLIF) [227,228], infrared
imaging [56,229] and microcantilever sensors attached to a computer controlled xyz stage
[230].
Infrared imaging has been applied for measuring the jet shape of the nebulizer chip by
capturing the heat trace of the jet on a piece of paper along the jet [56]. As such, the hot gas
would not stand out from the background. However, the capture sheet distorts the jet
reducing spatial and temperature accuracy of the measurement. The method is also very
sensitive to the positioning and deformation of the capture sheet. Despite the limitations of
the applied infrared jet shape measurement the results clearly showed that a narrow plume
can be produced by the nebulizer chip. However, the apparent difference between DRIE
and wet etched silicon nozzles is probably exaggerated by measurement artefacts and in
reality the jet shapes are much closer to each other.
Publication IV presents a method for measuring 3D temperature distribution of the jet.
A miniature thermocouple is attached to a computer-controlled xyz table and an array of
data points is measured at different positions. A schematic of the measurement set-up is
shown in Figure 3.9 (a). The thermocouple has a sharp about 7 mm long V-shaped tip
(Figure 3.9 (b)) made of 25 ?m diameter wire with the junction at the apex. A set of 2D
temperature scans is shown in Figure 3.10. The measurements were done under a box that
provided some protection from the random air flows in the laboratory, but nevertheless
some thermal noise is seen around the jet due to fluctuations in the ambient air.
The measured temperature is the temperature of the thermocouple junction and that
temperature depends on the sum of heat fluxes. The gas jet induces fairly rapid convective
heat transfer and thermal conduction of heat takes place along the thermocouple wires.
Depending on the position of the thermocouple either heat flux can be either positive or
negative. Thermal conduction along the wires depends on the cross-sectional area of the
wires and the temperature gradient along the wire. To study the effect of thermocouple size
thee different thermocouple wire diameters (13 ?m, 25 ?m and 50 ?m) were compared.
The measured temperature profiles along a line scan across the jet are shown in Figure 3.11
(a). The 13 ?m and 25 ?m diameter thermocouples provided practically identical results,
but the measured peak temperature was lower with the 50 ?m diameter thermocouple.
This experiment shows that with sufficiently small thermocouple the conduction along the
thermocouple wires can be neglected. The 25 ?m thermocouple was used in other
(a) (b)
Figure 3.9: (a) Schematic measurement set-up. (b) Close-up of the 25 ?m wire diameter
thermocouple tip. [IV]
48 3 Heated nebulizer chips
measurements because it was small enough for accurate measurements but mechanically
more rigid than the 13 ?m one.
It takes a while for the thermocouple temperature and the heat fluxes to reach
equilibrium. The measurement programme has a stabilization algorithm, which allows the
user to adjust stabilization parameters. There is a trade-off between the measurement time
and accuracy. Longer temperature signal averaging would allow more accurate results but
would increase the overall measurement time. Too fast measurements result in a hysteresis
effect in a reciprocating scan as seen in Figure 3.10 (a). The measurement time of a single
data point was about 0.5 s on average, but it varied between different temperature regions.
All in all, the thermocouple can be assumed to have reached equilibrium in this time as the
measured 63 % rise and fall time constants of the thermocouple were 28 ms and 60 ms,
respectively. Naturally, the exact time constants depend on the velocity of the jet, but the
order of magnitude should remain the same. Figure 3.11 (b) shows the measured
temperature as a function of gas flow rate. The measured temperature is highest at
120 sccm nitrogen flow rate. With higher flow rates the temperature decreases, which is
natural because the energy needed to heat the increased mass flow of the gas increases but
the heating power is kept constant. Also, the velocity of the gas is increasing and there
might be insufficient time to reach thermal equilibrium with the channel wall temperature
(a)
(b)
(c) (d) (e)
Figure 3.10: Temperature maps of a heated nitrogen jet produced by an all-glass nebulizer chip. The
chip nozzle is at the origin and gas flow is to the right. (a) xy cross-section, (b) xz cross-section and
yz cross-sections at (c) 1 mm, (d) 5 mm and (e) 10 mm distances from the nozzle. Nitrogen flow rate
100 sccm, 10 ?L min?1 water/methanol (50/50) and 3 W heating power. [IV]
3.4 Jet shape and temperature 49
due to decreasing residence time in the channel. At lower flow rates the measured
temperature also decreases which can be explained by the more rapid spreading of the low
velocity jet and decreased heat flux to the thermocouple which might be insufficient to
outpower the conduction losses via thermocouple wires.
The thermocouple scanning method provides temperature distribution but velocity
distribution of the jet is also of considerable interest. The different physical and chemical
magnitudes are interconnected by the laws of physics. To get an idea of the correlation
between temperature and velocity profiles an axial symmetry 2D computer simulation
model was made using the COMSOL Multiphysics 3.5 program with a non-isothermal flow
application mode. The boundary conditions were selected to provide peak temperature
and mass flow analogous to the thermocouple measured jets. The simulation results are
shown in Figure 3.12.
The simulated velocity profile is parabolic throughout the jet, which is characteristic for
a laminar flow. The jet exits the nozzle with uniform temperature but after some 2 mm
from the nozzle also the temperature profile has a Gaussian profile and after this both
temperature and velocity decrease at similar rate as shown in Figure 3.12 (c). The heat of
the jet is dissipated to the ambient air mostly by conduction because convection is small
due to laminar flow. Velocity decreases due to viscous friction and due to decreasing gas
volume with decreasing temperature. Based on the simulation results, it can be concluded
that an approximate velocity profile is obtained by measuring the temperature profile of a
heated jet.
(a) (b)
Figure 3.11: (a) Temperature profile along the y-axis at a distance of 5 mm from the nozzle with
different size thermocouples (100 sccm N2, 3 W heating power). (b) The measured peak temperature
of the jet 5 mm from the nozzle as a function of gas flow rate (25 ?m thermocouple, 3 W heating
power). [IV]
50 3 Heated nebulizer chips
3.5 Applications of the heated nebulizer chips
The nebulizer chips can be applied to several ionization methods and used in various
configurations (e.g. in combination with different separation methods). Table 3.2 lists all the
ionisation principles that have so far been demonstrated with the nebulizer chips. The first
demonstrations were with APCI [52] and APPI [53]. In APPI, a dopant (e.g. toluene) is
mixed in vaporized sample. The dopant has an important role in the photoionization
process.
Performance of APCI using the nebulizer chip as the source has been found better
compared to a commercial macroscopic APCI source [52] enabling much lower sample
flow rates without compromising the detection limits. This can be explained with the shape
of the nebulizer chip's plume. The chip provides a very narrow and confined jet compared
to the commercial APCI, which has roughly one order of magnitude larger tubular nozzle.
As the range of a corona discharge is in the order of a few hundred micrometers the
analytes in a jet with the similar diameter are ionized much more efficiently compared to a
larger plume. The jet shape also determines how efficiently the ions are collected to the MS
inlet. The nozzle shape of the chip and the resulting jet shape can easily be varied and,
therefore, knowing the jet shape is important. Publication IV and Section 3.4 shows that the
jet is sharp and it spreads out very little in the measured 13 mm distance which is similar to
the distance between the chip and the inlet of MS.
In addition to APCI and APPI, the nebulizer chip can be used for other API techniques.
Sonic spray ionization (SSI) is demonstrated in [7] and atmospheric pressure thermospray
ionization (APTSI) in [1]. In SSI the nebulizer gas flow is much higher than with APCI and
(a)
(b) (c)
Figure 3.12: (a) 2D axial symmetry computer simulation of the temperature distribution of a heated
nitrogen gas jet. The black velocity streamlines follow the temperature profile closely. (b) Axial
temperature and velocity profile at distances of 1 mm and 5 mm from the nozzle. (c) Temperature
and velocity along the centreline. [IV]
3.5 Applications of the heated nebulizer chips 51
the velocity of the jet is near the speed of sound, i.e. sonic speed. No heating of the chip is
required. SSI is best suited for analytes that are already ionized in the solution whereas
ionization of nonpolar compounds is inefficient. Nevertheless, SSI is a robust and soft
ionization technique, which makes it suitable for thermolabile compounds. APTSI differs
from APCI and APPI by the lack of additional excitation. Only heating and nebulizer gas
are used. Certain analytes are ions already in the solution or may be spontaneously
protonated in the gas phase after vaporization. Out of the the above mentioned API
techniques, APPI is the most universal one providing efficient ionization of the widest
range of analytes.
Direct sample infusion has limited applicability in practical applications, but it can be
useful, for example, with crude oil analysis with extremely powerful Fourier transform ion
cyclotron resonance mass spectrometry (FT-ICR MS) as shown in [2]. Using the nebulizer
chip instead of a conventional APPI source has clear advantages. Similar sensitivity is
obtained with much lower sample flow rates and the contamination of the mass
spectrometer chamber is reduced. In addition, the nebulizer chips are easy to replace after
accumulation of non-volatile crude oil components inside the chip.
The optimal sample flow rate range with the nebulizer chip is similar to the flow rates
used in capillary liquid chromatography. External LC systems have been connected with
the nebulizer chip via capillary tubing [6,8,9]. Commercial ion sources offer only ESI for
capillary LC whereas the HN chip enables other ionization methods extending the range of
suitable analytes.
The nebulizer chips can also be used with gaseous samples as shown in [9,10,12] where
a gas chromatograph is connected to the chip via a heated transfer line for APPI-MS
analysis. When coupling the chips to gas chromatography (GC) it is important that there
Table 3.2: Summary of different ionization methods demonstrated with the heated nebulizer chips.
Method Description Applicability
APCI Atmospheric pressure chemical
ionization; heated jet, high voltage
applied to a needle electrode for
corona discharge
For thermally stable polar and less polar
analytes, m/z ? 1.000
APPI Atmospheric pressure
photoionization; heated jet, dopant
mixed to nebulizer gas
For thermally stable polar and nonpolar
analytes, m/z ? 1.000
APTSI Atmospheric pressure thermospray
ionization; heated jet
Thermally stable very polar analytes
DAPPI Desorption APPI; hot jet with dopant
applied towards sample surface
Analysis of surfaces and solid samples with
little or no preparation; suitable for same
analytes as APPI
IS Ionspray; room temperature, voltage
applied to the liquid sample,
nebulizer gas flow
Biomolecules and other (polar) analytes
similar to ESI; allows higher flow rates and
more water in solvent than ESI (thanks to the
nebulizer gas, which is not used in ESI)
SSI Sonic spray ionization; room
temperature, high nebulizer gas flow
Biomolecules and other polar or ionic
analytes; more gentle than ESI
52 3 Heated nebulizer chips
are no cold spots in the part between the GC oven and the chip. Cold spots are problematic
because analytes with higher boiling point will condense compromising the GC separation.
The beginning of the chip where the transfer capillary is glued is a potential cold spot and,
therefore, silicon?glass chips were preferred for use with GC because they have less
temperature gradients compared to all-glass chips. However, in order to have suitable all-
glass HN chips for GC, a new heater layout was designed to provide more heating at the
beginning of the chip thus preventing the possibility of a cold spot.
The nebulizer chip can also be used for desorption ionization to analyse the
composition at the surfaces of the sample. Desorption atmospheric pressure
photoionization (DAPPI) has been used to analyse pharmaceutical tablets, illicit drugs,
environmental soil and food samples [3-5,11]. In DAPPI, a heated dopant containing gas jet
from the nebulizer chip and photons from a photoionization lamp is directed towards a
surface. Analytes on the surface are desorbed, ionized and the ions are collected into a mass
spectrometer.
Just recently, the silicon?glass nebulizer chip was demonstrated for the ionspray (IS)
technique where a high voltage is applied to the silicon part of the chip and no heating is
required [231]. IS is suitable for similar analytes as ESI ? that is for polar molecules and
large and thermolabile biomolecules ? but it can operate on higher flow rates and tolerates
a higher water proportion in the solvent.
53
4 On-chip liquid chromatography
This chapter reviews the literature about on-chip LC and discusses the LC chips developed
in this dissertation project.
4.1 Review of LC chips
Liquid chromatography is a very versatile and powerful separation method. However, LC
has gained less attention in the miniaturization community compared to CE, capillary
electrochromatography (CEC) and related separation methods [30,232-234]. There are,
nevertheless, quite a few publications dealing with microchip based LC. Various substrate
materials, stationary phases and detection methods have been applied. A fully integrated
LC system consists of pumps, a sample injector, a column and a detector. However, most of
the LC chip development is still concerning the miniaturization of one or a few
subcomponents and the devices still rely on external ? and usually regular size ?
components.
4.1.1 Materials
Silicon substrates with etched channels and anodically bonded glass cover have been used
by many [235-239] ? also in publications V and VI. Silicon dioxide based materials are
dominant in laboratory glassware and capillary tubing and they have been also applied to
LC chips in the form of glass [38,240,241], fused silica [242] and quartz [243] wafers. The
Agilent LC-ESI chip is made from laminated polyimide films machined by laser ablation
[35]. Other polymers that have been used include hot-embossed [244-246] or injection
moulded [247] COC, PDMS [248], SU-8 [37] and parylene-C [249].
4.1.2 Columns
There are four different types of the LC columns: open-tubular [235,236,250], particle
packed [35,36,38,248,251,249], monolith [37,240-242,244-247,252,253] and pillar array
columns [237,238,243,254-256].
Open-tubular columns have a small cross-sectional area (dimensions 0.5 ?m ? 250 ?m)
and they are quite long (10 cm ? 16 cm). The channel walls are coated with the desired
stationary phase which is typically hydrophobic saturated carbon chains. Open-tubular
columns are limited in injection capacity and subject to blocking [234]. For the best
separation performance long channels and, subsequently, high pumping pressures are
required. In general, the separation quality is proportional to the separation time.
More frequently, the column channel is filled with solid media to increase the surface
area of the column and thus to improve the capacity and to reduce the necessary column
length. Channels can be filled with particles by pumping the channel with a slurry with the
54 4 On-chip liquid chromatography
particles mixed in a solvent. In order to stop the particles from going through the channel a
frit or a tapering channel is needed at the column end. Although the cross-section of a
tapering channel end would be larger than the particle size the keystone-effect leads to
blocking of the particles at the column [248,257,251]. Elastic columns can block the particles
also by a clamping effect where the channel expands and contracts due to pressure change
or an anchoring effect with particles adhering to the sidewalls [248]. The LC chips in
publications V and VI use a high aspect ratio pillar array as a frit etched in silicon (see
Figure 4.1) to stop particles during column packing. The pillar frit works fine for 2.5 ?m
particles even though the distance between neighbouring pillars is at most 6 ?m.
Porous solid media filled columns are called monolithic columns. They are made by
first filling the column with a precursor which is then cured during which the solvent in
the precursor evaporates leaving a solid backbone which is then coated with the desired
functional coating (e.g. hydrophobic octadecylsilica, C18, for reversed-phase LC).
A different kind of monolithic column which also utilises the batch fabrication nature of
microfabrication is the pillar array column which consists of lithographically defined
regular array of etched pillars. The integrated pillars eliminate the need for a separate
packing step and offer higher capacity compared to open-tubular columns. With properly
optimized distances between pillars and channel sidewalls a flat flow profile can be
obtained for minimal band broadening [258]. A perfect LC column is perfectly ordered, i.e.
the porosity is uniform [232]. However, it is difficult to eliminate all non-idealities that
decrease the theoretical performance. The pillars always have some variation in size, shape
and sidewall profile. Again, the column needs to be coated and the quality of the coating
affects the separation performance. The pillar array column has a limited capacity because
of limited surface area. Surface area of the silicon pillars can be increased by
electrochemical anodization [238,259].
Temperature of the LC column affects the retention properties of the analytes and
heating of the column can potentially provide better and faster separation results. The LC
chip in publication VI has a separate platinum heater element for the column, but so far it
has not been used. An LC column heater has also been presented in [235].
Figure 4.1: Detail from the LC chip.
400 ?m
4.1 Review of LC chips 55
4.1.3 Interconnections, sample injection and pumping
Fluidic interconnections for microchip LC are challenging because they need to tolerate the
high pumping pressures required by the small channels. Depending on the column
dimensions and the packing density the pumping pressures can be several hundreds of
atmospheres. Gluing of capillary tubing is a straightforward solution [240], but is has its
weaknesses, as discussed in Section 2.6.1. If the chip is not made of brittle and hard
material syringe needles can be pressed in providing tight connections up to 200 bar
without additional gluing [245]. Alternatively, the chip can be compressed inside a custom
made connector block [260]. In the LC chips of publication V, the capillaries were inserted
and glued from the edge of the chip directly in the column. A custom made holder with
standard flat-bottom fittings was used with integrated LC-nebulizer chip of publication VI.
The connections did not show leakage with the maximum 150 bar pumping pressure used
in the experiments.
In LC, the column is filled with eluent and then a small volume of sample is injected
and the eluent pumping is continued. The two main principles of sample injection are
shown in Figure 4.2. Sample injection can be done using crossing channels with the
injection volume determined by the channel volume at the crossing [240,261]. Simple
crossing of two channels or a double-T configuration are frequently used. This method is
ideal for microchip CE where the direction of the flows is determined simply by the
applied voltage at each inlet. However, mechanical valves are needed to avoid leakage
between the additional injection channels in LC because the flow is pressure driven.
Therefore, many LC chips rely on external valves for the injection [245]. The injection
volume is determined by the volume of the injection loop. The unique feature of the
Agilent chip is that it is placed between the stator and rotor of the injection valve [35]. This
enables high pressure tolerant connections, low dead volumes and eliminates possible
cross-talk between different inlets thanks to physical separation between them.
(a) (b)
Figure 4.2: Different sample plug injection schemes: (a) on-chip injection using crossing channels
and (b) off-chip injection with a mechanical 6-way injection valve.
Unlike CE, LC is based on pressure driven flow which is more challenging to
miniaturize. Integrated pumps can rely on a separate electroosmotic flow (EOF) pump
[261] or on electrochemical pumping via electrolysis in a closed chamber [36]. However,
column
waste
pump
sample in
sample loop
column
waste
pump
sample in
sample loop
Step 1: fill sample loop
Step 2: injection to column 60?
Step 1: fill w ith eluent
waste
I
t
A B
AB
pump
wastepump
pump
Step 2: introduce sample
Step 3: injection
Step 4: separation and detection
column
56 4 On-chip liquid chromatography
frequently external pumps are applied with LC chips. Syringe pumps or commercial LC
pumps can provide high pressures and stable flow rates. A moving cover lid can provide
shear driven flow [256]. In practical LC applications the composition of the mobile phase
eluent is usually changed during the run. This is achieved by using gradient pumps that
can generate a desired mixture from several sources. A clever on-chip gradient generator
channel network has been demonstrated to create gradient flow using only a single pump
[39].
4.1.4 Detection
Several detection methods are applicable to LC. Mass spectrometric detection with an on-
chip ESI source is frequently applied [35-38]. On-chip detection methods that rely heavily
on external instrumentation include fluorescence [242,245] and UV absorbance [261]
detections. Detection methods that can be integrated inside the chip include amperometric
[236], heat transfer based [262], conductometric [250] and electrochemical [263] detectors.
Naturally, external control electronics and data analysis software are required even with
these detection methods. Although some of these methods may be highly sensitive or
specific, none can provide the performance, versatility and identification capability of MS.
The integrated LC?HN of publication VI chip extends the applicability of LC chips by
enabling other API-MS techniques than ESI to be used for the ionization of a wider range of
analytes as discussed in Section 3.5.
4.2 All-glass LC-nebulizer chip
A major problem with the integrated LC?HN chip of publication VI is the conduction of
heat from the hot (up to 500 ?C) nebulizer section to the LC column. For good separation
efficiency the column temperature should be as uniform as possible. The high thermal
conductivity of silicon (150 W m?1 ?C?1 ) is a problem when trying to keep different parts of
the chip at different temperatures. Minimizing the silicon cross-sectional area will improve
thermal insulation, but this was insufficient to provide the frit and column temperatures
low enough in publication VI. Therefore, an external cooler block was used in to eliminate
excess heating of the column. Similar problems with thermal conductivity of silicon have
been encountered by others, too. For efficient thermal isolation the silicon is completely
removed and replaced with a thermally isolating material. Parylene-C has very low
thermal conductivity (0.08 W m?1 ?C?1 ) and it can be conformally coated which makes it a
feasible choice [263] but only for temperatures up to its glass transition point around
100 ?C. Suspended low stress silicon nitride is applicable also for higher temperatures
[264]. The thermal conductivity of glass is roughly two orders of magnitude lower than that
of silicon. Therefore, a process for making the LC-nebulizer chip out of two glass wafers
was developed.
An LC-nebulizer chip made entirely out of glass should eliminate the need for active
cooling between the LC and HN chip parts that was required with the silicon?glass chip. In
addition, the all-glass design has reduced flushed volume after the LC column prior to the
mixer with nebulizer gas implemented with double layer masking process with aluminium
nitride and SU-8 was used for making different channel depths for the pillar frit and the
column portions of the chip. In theory, the LC column channel depth would smoothly
decrease prior to the frit due to aspect ratio dependent etch phenomenon.
4.2 All-glass LC-nebulizer chip 57
The fabrication process of the all glass LC HN chip is illustrated in Figure 4.3. First,
reactive sputtering (Von Ardenne Anlagentechnik GmbH, Dresden, Germany) of 1.3 ?m
thick AlN hard mask layer is done on a clean glass wafer. AlN is patterned with channel
and pillar frit structures using standard lithography (AZ5214E resist) and RIE (STS AOE).
After resist removal, a 50 ?m thick layer of SU-8 is applied over the patterned AlN. The
photomask for SU-8 is otherwise the same as that for AlN except that the LC column
channel tapers down just prior to the frit and the frit and narrow optical detection channel
are covered with SU-8. DRIE of glass is done in two steps. First the SU-8 is used as the
mask to etch part of the desired channel depth (100 ?m). Then SU-8 is stripped in O2/CF4
plasma and glass DRIE is continued with AlN as the masking layer. The AlN mask was
necessary as the pillar size structures could not be patterned by the thick SU-8. Because the
glass DRIE process is not optimal for making through-wafer etching, the necessary fluidic
interconnections would be made via the bonded glass cover wafer with through-holes.
Unfortunately, there were issues that could not be overcome and no chips could be
completed for testing with LC and MS. The AlN to glass etch selectivity was too small
which limited the channel depth to about 5 ?m. Unfortunately, process optimization could
not be carried out properly due to a limited number of samples. The AlN sputter had poor
availability and there were repeatability issues involving AlN film or wafer cracking and
problems with adhesion. The high selectivity glass DRIE process with AlN masking that
had worked well for fused silica wafers [137] was not compatible with Pyrex glass because
of the non-volatile etch products. Therefore, a process with higher ion bombardment was
used with the expense of rapid wearing of the masking layer. Figure 4.4 shows an example
of the obtained structures.
Figure 4.3: All-glass LC-nebulizer chip fabrication process: AlN deposition and patterning on a
glass wafer (a-c), SU-8 (d), glass DRIE (e), SU-8 strip (f), continued glass DRIE for pillar frit (g),
AlN stripping (h), Pt sputtering & patterning (i-k) and bonding with cover wafer with inlet holes
(l).
(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
(k)
(l)
(j)
(i)
58 4 On-chip liquid chromatography
Figure 4.4: Fish bone shaped pillar frit structure in Pyrex glass made with double mask layer mask
and DRIE. (SEM by Kai Kolari)
50 ?m
LC column
channelfish bone fr it
59
5 Conclusions and outlook
The HN chips have shown several advantages over conventional API mass spectrometer
ion sources. The chip sources can be operated at a fraction of the sample flow rates without
compromising sensitivity. This leads to reduced sample, solvent and nebulizer gas
consumption and allows higher mass spectrometer uptime thanks to less frequent need to
clean the chamber from accumulated contamination from the sample flow. All these lead to
cost savings. The HN chips cover all the most used API techniques (APCI, APPI, SSI, APTSI
and SI) and it can be applied to desorption ionization (DAPPI) as well. The nebulizer chips
can operate with any API mass spectrometer whereas commercial ion sources are
dedicated to a single model or brand. However, there is still work to be done before the
chips can be commercialized. Most important, it should be easy to equip a mass
spectrometer with the necessary mountings required by the chips and the routine
operation of the chips must be straightforward. Custom fixtures and power sources can be
used in experimental research environments but this is unacceptable for routine mass
spectrometer users. For added operator safety and reduction of the risk of contamination of
the mass spectra from compounds in the laboratory air, the ion source has to be enclosed
inside a casing. Chip design, its lifetime and the production economics can still be
improved. Different ionization methods have different requirements for electric
conductivity and temperature of the chip as well as the nozzle and the jet shape. The
channel and the nozzle dimensions used in the HN chips so far, are quite large from the
microfabrication point of view and their further miniaturization could be worth further
studies. Combined ionization modes such as simultaneous or alternating ESI and APPI are
also subject to future study. For this, some ideas could be taken from the silicon micropillar
ESI chip developed in our group [33].
The LC?HN chip was the first published demonstration of this kind of integrated LC
chip for MS, but so far its analytical performance has been less impressive and the reasons
for the suboptimal chromatographic peak shapes need to be investigated. Other future
studies will include channel sidewall deactivation coating and high temperature LC with
the use of the separate column heater element. One advantage of the current integrated
chip is that post column dead volume is minimal. However, the post column dead volume
can be negligible even with a conventional capillary LC connected to a HN chip and, as
such, the current LC?HN chip has little advantage to offer. However, integration of further
functionality could take the usefulness of the chip up to a higher level. Integration of a pre-
column or a sample injector could be one possibility. Changing the column from the
particle packed one to a microfabricated pillar array column would reduce the required
post-processing of the chips after wafer dicing. Despite the faced challenges during our
attempt to make one, the all-glass integrated LC?HN chip is still an intriguing prospect as
it should solve the issues concerning excess heating of the LC column part of the chip
60 5 Conclusions and outlook
without resorting to an active cooling element. A parallel LC chip with multiple columns
could be useful for high throughput analysis. Multidimensional separations [265] or
transformation into a CEC [266,267] chip which combines both CE and LC in a single
column could open up interesting new separation capabilities.
Another future prospect could be the integration of a GC column and a heated
nebulizer. The LC?HN chip would be a good starting point for such a GC?HN chip. As GC
is typically done at a similar temperature as the HN operating temperature there should be
no problems with the heat transport even with a silicon?glass design.
A lot of know-how about silicon and glass microfabrication, on-chip heaters, chip-to-
world interconnection schemes and jet shape measurements was acquired during this
work. The HN chips and LC columns discussed in this dissertation have only scratched the
surface of the possible applications where this kind of microfabrication could be applied to.
The multidisciplinary nature of the field is both a challenge and an opportunity. Successful
collaboration between experts with different backgrounds can lead to innovations having a
revolutionizing impact on health care, security and the environment.
61
References
[1] P. Keski-Rahkonen, M. Haapala, V. Saarela, S. Franssila, T. Kotiaho, R. Kostiainen, and S. Auriola,
Atmospheric pressure thermospray ionization using a heated microchip nebulizer, Rapid
Communications in Mass Spectrometry 23 (2009) 3313-3322. doi:10.1002/rcm.4252
[2] M. Haapala, J.M. Purcell, V. Saarela, S. Franssila, R.P. Rodgers, C.L. Hendrickson, T. Kotiaho, A.G.
Marshall, and R. Kostiainen, Microchip Atmospheric Pressure Photoionization for Analysis of
Petroleum by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry, Analytical Chemistry
81 (2009) 2799-2803. doi:10.1021/ac802427m
[3] L. Luosujarvi, V. Arvola, M. Haapala, J. P?l, V. Saarela, S. Franssila, T. Kotiaho, R. Kostiainen, and
T. Kauppila, Desorption and ionization mechanisms in desorption atmospheric pressure
photoionization, Analytical Chemistry 80 (2008) 7460-7466. doi:10.1021/ac801186x
[4] M. Haapala, J. P?l, V. Saarela, V. Arvola, T. Kotiaho, R. Ketola, S. Franssila, T. Kauppila, and R.
Kostiainen, Desorption atmospheric pressure photoionization, Analytical Chemistry 79 (2007) 7867-
7872. doi:10.1021/ac071152g
[5] T. Kauppila, V. Arvola, M. Haapala, J. P?l, L. Aalberg, V. Saarela, S. Franssila, T. Kotiaho, and R.
Kostiainen, Direct analysis of illicit drugs by desorption atmospheric pressure photolonization,
Rapid Communications in Mass Spectrometry 22 (2008) 979-985. doi:10.1002/rcm.3461
[6] L.L. Ahonen, M. Haapala, V. Saarela, S. Franssila, T. Kotiaho, and R. Kostiainen, Feasibility of
capillary liquid chromatography-microchip atmospheric pressure photoionization-mass
spectrometry in analyzing anabolic steroids in urine samples, Rapid Communications in Mass
Spectrometry 24 (2010) 958-964. doi:10.1002/rcm.4468
[7] J. P?l, T. Kauppila, M. Haapala, V. Saarela, S. Franssila, R. Ketola, T. Kotiaho, and R. Kostiainen,
Microchip sonic spray ionization, Analytical Chemistry 79 (2007) 3519-3523. doi:10.1021/ac070003v
[8] P. ?stman, S. J?ntti, K. Grigoras, V. Saarela, R. Ketola, S. Franssila, T. Kotiaho, and R. Kostiainen,
Capillary liquid chromatography-microchip atmospheric pressure chemical ionization-mass
spectrometry, Lab on a Chip 6 (2006) 948-953. doi:10.1039/b601290f
[9] M. Haapala, L. Luosujarvi, V. Saarela, T. Kotiaho, R.A. Ketola, S. Franssila, and R. Kostiainen,
Microchip for combining gas chromatography or capillary liquid chromatography with
atmospheric pressure photoionization-mass spectrometry, Analytical Chemistry 79 (2007) 4994-4999.
doi:10.1021/ac070157a
[10] L. Luosujarvi, M. Karikko, M. Haapala, V. Saarela, S. Huhtala, S. Franssila, R. Kostiainen, T.
Kotiaho, and T. Kauppila, Gas chromatography/mass spectrometry of polychlorinated biphenyls
using atmospheric pressure chemical ionization and atmospheric pressure photoionization
microchips, Rapid Communications in Mass Spectrometry 22 (2008) 425-431. doi:10.1002/rcm.3379
[11] L. Luosuj?rvi, S. Kanerva, V. Saarela, S. Franssila, R. Kostiainen, T. Kotiaho, and T.J. Kauppila,
Environmental and food analysis by desorption atmospheric pressure photoionization-mass
spectrometry, Rapid Communications in Mass Spectrometry 24 (2010) 1343-1350. doi:10.1002/rcm.4524
[12] L. Luosuj?rvi, M. Haapala, M. Thevis, V. Saarela, S. Franssila, R.A. Ketola, R. Kostiainen, and T.
Kotiaho, Analysis of selective androgen receptor modulators by gas chromatography-microchip
atmospheric pressure photoionization-mass spectrometry, Journal of the American Society for Mass
Spectrometry 21 (2010) 310-316. doi:10.1016/j.jasms.2009.10.019
[13] G.E. Moore, Cramming more components onto integrated circuits, Electronics 38 (1965) 114.
[14] Silicon Technology from Intel, http://www.intel.com/technology/architecture-silicon/silicon.htm?
iid=tech_as_lhn+silicon
[15] ITRS Starting Materials Sub-TWG, Advantages and Challenges Associated with the Introduction of
450mm Wafers, 2005.
[16] T.R. Covey, B.A. Thomson, and B.B. Schneider, Atmospheric pressure ion sources, Mass
Spectrometry Reviews 28 (2009) 870-897. doi:10.1002/mas.20246
62
[17] Inkjet Head Market flyer, Yole D?veloppement,
http://www.yole.fr/pagesAn/products/pdf/Ink_jet_head_market.pdf
[18] M.C. Louwerse, H.V. Jansen, M.N.W. Groenendijk, and M.C. Elwenspoek, Nozzle fabrication for
micropropulsion of a microsatellite, Journal of Micromechanics and Microengineering 19 (2009) 045008.
[19] J.D.J.S. Samuel, R. Steger, G. Birkle, R. Zengerle, P. Koltay, and J. Ruhe, Modification of
Micronozzle Surfaces Using Fluorinated Polymeric Nanofilms for Enhanced Dispensing of Polar
and Nonpolar Fluids, Analytical Chemistry 77 (2005) 6469-6474. doi:10.1021/ac050878e
[20] W.G. Gardner, J.W. Jaworski, A.P. Camacho, and J.M. Protz, Experimental results for a microscale
ethanol vapor jet ejector, Journal of Micromechanics and Microengineering 20 (2010) 045019.
doi:10.1088/0960-1317/20/4/045019
[21] S. Zimmermann, S. Wischhusen, and J. M?ller, Micro flame ionization detector and micro flame
spectrometer, Sensors and Actuators B 63 (2000) 159-166. doi:10.1016/S0925-4005(00)00353-1
[22] S. Terry, J. Jerman, and J. Angell, A gas chromatographic air analyzer fabricated on a silicon wafer,
IEEE Transactions on Electron Devices 26 (1979) 1880-1886.
[23] A. Manz, N. Graber, and H. Widmer, Miniaturized total chemical analysis systems: A novel
concept for chemical sensing, Sensors and Actuators B 1 (1990) 244-248. doi:10.1016/0925-
4005(90)80209-I
[24] J. West, M. Becker, S. Tombrink, and A. Manz, Micro Total Analysis Systems: Latest Achievements,
Analytical Chemistry 80 (2008) 4403-4419. doi:10.1021/ac800680j
[25] M.A. Burns, Everyone's a (Future) Chemist, Science 296 (2002) 1818-1819.
doi:10.1126/science.1073562
[26] M.A. Burns, B.N. Johnson, S.N. Brahmasandra, K. Handique, J.R. Webster, M. Krishnan, T.S.
Sammarco, P.M. Man, D. Jones, D. Heldsinger, C.H. Mastrangelo, and D.T. Burke, An Integrated
Nanoliter DNA Analysis Device, Science 282 (1998) 484-487. doi:10.1126/science.282.5388.484
[27] W.R. Vandaveer IV, S.A. Pasas-Farmer, D.J. Fischer, C.N. Frankenfeld, and S.M. Lunte, Recent
developments in electrochemical detection for microchip capillary electrophoresis, Electrophoresis
25 (2004) 3528-3549. doi:10.1002/elps.200406115
[28] K.B. Mogensen, H. Klank, and J.P. Kutter, Recent developments in detection for microfluidic
systems, Electrophoresis 25 (2004) 3498-3512. doi:10.1002/elps.200406108
[29] T. Sikanen, S. Franssila, T.J. Kauppila, R. Kostiainen, T. Kotiaho, and R.A. Ketola, Microchip
technology in mass spectrometry, Mass Spectrometry Reviews 29 (2010) 351-391.
doi:10.1002/mas.20238
[30] S. Koster and E. Verpoorte, A decade of microfluidic analysis coupled with electrospray mass
spectrometry: An overview, Lab on a Chip 7 (2007) 1394-1412. doi:10.1039/b709706a
[31] J. Lee, S.A. Soper, and K.K. Murray, Microfluidic chips for mass spectrometry-based proteomics,
Journal of Mass Spectrometry 44 (2009) 579-593. doi:10.1002/jms.1585
[32] S. Tuomikoski, T. Sikanen, R.A. Ketola, R. Kostiainen, T. Kotiaho, and S. Franssila, Fabrication of
enclosed SU-8 tips for electrospray ionization-mass spectrometry, Electrophoresis 26 (2005) 4691-
4702. doi:10.1002/elps.200500475
[33] L. Sainiemi, T. Nissil?, V. Jokinen, T. Sikanen, T. Kotiaho, R. Kostiainen, R.A. Ketola, and S.
Franssila, Fabrication and fluidic characterization of silicon micropillar array electrospray
ionization chip, Sensors and Actuators B 132 (2008) 380-387. doi:10.1016/j.snb.2007.09.077
[34] T. Sikanen, S. Tuomikoski, R.A. Ketola, R. Kostiainen, S. Franssila, and T. Kotiaho, Fully
Microfabricated and Integrated SU-8-Based Capillary Electrophoresis-Electrospray Ionization
Microchips for Mass Spectrometry, Analytical Chemistry 79 (2007) 9135-9144. doi:10.1021/ac071531+
[35] H. Yin, K. Killeen, R. Brennen, D. Sobek, M. Werlich, and T. van de Goor, Microfluidic Chip for
Peptide Analysis with an Integrated HPLC Column, Sample Enrichment Column, and
Nanoelectrospray Tip, Analytical Chemistry 77 (2005) 527-533. doi:10.1021/ac049068d
[36] J. Xie, Y. Miao, J. Shih, Y. Tai, and T.D. Lee, Microfluidic Platform for Liquid
Chromatography?Tandem Mass Spectrometry Analyses of Complex Peptide Mixtures, Analytical
Chemistry 77 (2005) 6947-6953. doi:10.1021/ac0510888
[37] J. Carlier, S. Arscott, V. Thomy, J.C. Camart, C. Cren-Oliv?, and S. Le Gac, Integrated
microfabricated systems including a purification module and an on-chip nano electrospray
ionization interface for biological analysis, Journal of Chromatography A 1071 (2005) 213-222.
[38] I.M. Lazar, P. Trisiripisal, and H.A. Sarvaiya, Microfluidic liquid chromatography system for
proteomic applications and biomarker screening, Analytical Chemistry 78 (2006) 5513-5524.
[39] R.A. Brennen, H. Yin, and K.P. Killeen, Microfluidic Gradient Formation for Nanoflow Chip LC,
Analytical Chemistry 79 (2007) 9302-9309. doi:10.1021/ac0712805
63
[40] C.S. Chu, M.R. Ni?onuevo, B.H. Clowers, P.D. Perkins, H.J. An, H. Yin, K. Killeen, S. Miyamoto, R.
Grimm, and C.B. Lebrilla, Profile of native N-linked glycan structures from human serum using
high performance liquid chromatography on a microfluidic chip and time-of-flight mass
spectrometry, Proteomics 9 (2009) 1939-1951. doi:10.1002/pmic.200800249
[41] K. Cottingham, Nano-LC/ESI on a chip, Analytical Chemistry 77 (2005) 51A. doi:10.1021/ac0533321
[42] S. Ehlert, K. Kraiczek, J.A. Mora, M. Dittmann, G.P. Rozing, and U. Tallarek, Separation Efficiency
of Particle-Packed HPLC Microchips, Analytical Chemistry 80 (2008) 5945-5950.
[43] K. Reinsberg, U. Effelsberg, and U. Tallarek, Microchip electrospray performance during gradient
elution with bulk conductivity changes, Lab on a Chip 9 (2009) 2914-2923. doi:10.1039/b905052c
[44] H. Yin and K. Killeen, The fundamental aspects and applications of Agilent HPLC-Chip, Journal of
Separation Science 30 (2007) 1427-1434. doi:10.1002/jssc.200600454
[45] T.J. Kauppila, T. Kuuranne, E.C. Meurer, M.N. Eberlin, T. Kotiaho, and R. Kostiainen, Atmospheric
Pressure Photoionization Mass Spectrometry. Ionization Mechanism and the Effect of Solvent on
the Ionization of Naphthalenes, Analytical Chemistry 74 (2002) 5470-5479. doi:10.1021/ac025659x
[46] S. Tuomikoski, K. Huikko, K. Grigoras, P. Ostman, R. Kostiainen, M. Baumann, J. Abian, T.
Kotiaho, and S. Franssila, Preparation of porous n-type silicon sample plates for
desorption/ionization on silicon mass spectrometry (DIOS-MS), Lab on a Chip 2 (2002) 247-253.
[47] A. Sen, J. Darabi, and D. Knapp, Design, fabrication and test of a microfluidic nebulizer chip for
desorption electrospray ionization mass spectrometry, Sensors and Actuators B 137 (2009) 789-796.
doi:10.1016/j.snb.2009.02.002
[48] S. Tuomikoski, K. Huikko, K. Grigoras, P. Ostman, R. Kostiainen, M. Baumann, J. Abian, T.
Kotiaho, and S. Franssila, Preparation of porous n-type silicon sample plates for
desorption/ionization on silicon mass spectrometry (DIOS-MS), Lab on a Chip 2 (2002) 247-253.
doi:10.1039/b207634a
[49] Z. Ouyang and R.G. Cooks, Miniature mass spectrometers, Annual Review of Analytical Chemistry 2
(2009) 187?214. doi:10.1146/annurev-anchem-060908-155229
[50] Torion Technologies Inc., http://www.torion.com/
[51] Z. Ouyang, R.J. Noll, and R.G. Cooks, Handheld Miniature Ion Trap Mass Spectrometers,
Analytical Chemistry 81 (2009) 2421-2425. doi:10.1021/ac900292w
[52] P. ?stman, S. Marttila, T. Kotiaho, S. Franssila, and R. Kostiainen, Microchip atmospheric pressure
chemical ionization source for mass spectrometry, Analytical Chemistry 76 (2004) 6659-6664.
doi:10.1021/ac049345g
[53] T. Kauppila, P. Ostman, S. Marttila, R. Ketola, T. Kotiaho, S. Franssila, and R. Kostiainen,
Atmospheric pressure photoionization-mass spectrometry with a microchip heated nebulizer,
Analytical Chemistry 76 (2004) 6797-6801. doi:10.1021/ac049058c
[54] S. Marttila, Micromechanical ion source made of silicon and glass, Master's thesis, Helsinki University
of Technology, Finland, 2003.
[55] R. Kostiainen, S. Franssila, T. Kotiaho, and S. Marttila, Method and apparatus for mass spectrometric
analysis, Patent #WO2005047848 (A3)
[56] S. Franssila, S. Marttila, K. Kolari, P. Ostman, T. Kotiaho, R. Kostiainen, R. Lehtiniemi, C. Fager,
and J. Manninen, A microfabricated nebulizer for liquid vaporization in chemical analysis, Journal
of Microelectromechanical Systems 15 (2006) 1251-1259. doi:10.1109/JMEMS.2006.879671
[57] D. H?lsenberg, A. Harnisch, and A. Bismarck, Microstructuring of glasses, Springer, 2008.
[58] S. Franssila, Introduction to Microfabrication, Wiley, 2004.
[59] G.T.A. Kovacs, Micromachined Transducers Sourcebook, McGraw-Hill Science/Engineering/Math,
1998.
[60] M.J. Madou, Fundamentals of Microfabrication: The Science of Miniaturization, Second Edition, CRC,
2002.
[61] V. Lindroos, M. Tilli, A. Lehto, and T. Motooka, Handbook of silicon based MEMS materials and
technologies, William Andrew/Elsevier, 2010.
[62] H. Becker and C. G?rtner, Polymer microfabrication technologies for microfluidic systems,
Analytical and Bioanalytical Chemistry 390 (2008) 89-111. doi:10.1007/s00216-007-1692-2
[63] A.N. Tiwari, A. Romeo, D. Baetzner, and H. Zogg, Flexible CdTe solar cells on polymer films,
Progress in Photovoltaics: Research and Applications 9 (2001) 211-215. doi:10.1002/pip.374
[64] Heidelberg Instruments Mikrotechnik GmbH, http://www.himt.de/
[65] A.E. Grigorescu and C.W. Hagen, Resists for sub-20-nm electron beam lithography with a focus on
HSQ: state of the art, Nanotechnology 20 (2009) 292001. doi:10.1088/0957-4484/20/29/292001
[66] H. Schift, Nanoimprint lithography: An old story in modern times? A review, Journal of Vacuum
Science & Technology B 26 (2008) 458-480. doi:10.1116/1.2890972
64
[67] SU-8 Resists, MicroChem Corp., http://www.microchem.com/products/su_eight.htm
[68] G.M. Kim, M.A.F. van den Boogaart, and J. Brugger, Fabrication and application of a full wafer size
micro/nanostencil for multiple length-scale surface patterning, Microelectronic Engineering 67-68
(2003) 609-614. doi:10.1016/S0167-9317(03)00121-7
[69] S.W. Pang, M.W. Geis, W.D. Goodhue, N.N. Efremow, D.J. Ehrlich, R.B. Goodman, and J.N.
Randall, Pattern transfer by dry etching through stencil masks, Journal of Vacuum Science &
Technology B 6 (1988) 249-252. doi:10.1116/1.584016
[70] Y. Xia and G.M. Whitesides, Soft Lithography, Annual Review of Materials Science 28 (1998) 153-184.
doi:10.1146/annurev.matsci.28.1.153
[71] J. Giboz, T. Copponnex, and P. M?l?, Microinjection molding of thermoplastic polymers: a review,
Journal of Micromechanics and Microengineering 17 (2007) R96-R109. doi:10.1088/0960-1317/17/6/R02
[72] C. Lin, Y. Lee, C. Chang, W. Hsu, Y. Chang, and W. Fang, Novel glass microprobe arrays for neural
recording, Biosensors and Bioelectronics 25 (2009) 475-481. doi:10.1016/j.bios.2009.08.006
[73] V. Fascio, R. Wuthrich, D. Viquerat, and H. Langen, 3D microstructuring of glass using
electrochemical dischargemachining (ECDM), 1999 International Symposium on Micromechatronics
and Human Science (1999) 179-183.
[74] D.M. Eigler and E.K. Schweizer, Positioning single atoms with a scanning tunnelling microscope,
Nature 344 (1990) 524-526. doi:10.1038/344524a0
[75] A.K. Dubey and V. Yadava, Laser beam machining--A review, International Journal of Machine Tools
and Manufacture 48 (2008) 609-628. doi:10.1016/j.ijmachtools.2007.10.017
[76] Cunxia Li et al, Fabrication of three-dimensional microchannels inside silicon using a femtosecond
laser, Journal of Micromechanics and Microengineering 19 (2009) 125007. doi:10.1088/0960-
1317/19/12/125007
[77] D.S. Miller, Micromachining with abrasive waterjets, Journal of Materials Processing Tech. 149 (2004)
37?42.
[78] N. Chekurov, K. Grigoras, A. Peltonen, S. Franssila, and I. Tittonen, The fabrication of silicon
nanostructures by local gallium implantation and cryogenic deep reactive ion etching,
Nanotechnology 20 (2009) 065307. doi:10.1088/0957-4484/20/6/065307
[79] C. Chao, S. Shen, and J. Wu, Fabrication of 3-D Submicron Glass Structures by FIB, Journal of
Materials Engineering and Performance 18 (2009) 878-885. doi:10.1007/s11665-008-9318-1
[80] M. Elwenspoek and H.V. Jansen, Silicon Micromachining, Cambridge University Press, 2004.
[81] G.T. Kovacs, N.I. Maluf, and K.E. Petersen, Bulk micromachining of silicon, Proceedings of the IEEE
86 (1998) 1536?1551. doi:10.1109/5.704259
[82] K. Biswas and S. Kal, Etch characteristics of KOH, TMAH and dual doped TMAH for bulk
micromachining of silicon, Microelectronics Journal 37 (2006) 519-525.
[83] J. Chen, L. Liu, Z. Li, Z. Tan, Q. Jiang, H. Fang, Y. Xu, and Y. Liu, Study of anisotropic etching of (1
0 0) Si with ultrasonic agitation, Sensors and Actuators A 96 (2002) 152-156.
[84] K. Sato, M. Shikida, T. Yamashiro, K. Asaumi, Y. Iriye, and M. Yamamoto, Anisotropic etching rates
of single-crystal silicon for TMAH water solution as a function of crystallographic orientation,
Sensors and Actuators A 73 (1999) 131-137.
[85] M. Decarli, V. Guarnieri, R. Pal, F. Giacomozzi, B. Margesin, and M. Zen, Influence of masking
layer stress on anisotropic silicon etching in TMAH solutions, International Conference on
Microelectronic Test Structures (2003) 25-28.
[86] C.R. Yang, P.Y. Chen, Y.C. Chiou, and R.T. Lee, Effects of mechanical agitation and surfactant
additive on silicon anisotropic etching in alkaline KOH solution, Sensors and Actuators A 119 (2005)
263-270.
[87] E. Van Veenendaal, K. Sato, M. Shikida, and J. Van Suchtelen, Micromorphology of single
crystalline silicon surfaces during anisotropic wet chemical etching in KOH and TMAH, Sensors
and Actuators A 93 (2001) 219-231.
[88] I. Zubel and M. Kramkowska, The effect of isopropyl alcohol on etching rate and roughness of (1 0
0) Si surface etched in KOH and TMAH solutions, Sensors and Actuators A 93 (2001) 138-147.
doi:10.1016/S0924-4247(01)00648-3
[89] B. Schwartz and H. Robbins, Chemical Etching of Silicon, Journal of The Electrochemical Society 123
(1976) 1903-1909. doi:10.1149/1.2132721
[90] D.L. Flamm, Mechanisms of silicon etching in fluorine-and chlorine-containing plasmas, Pure and
Applied Chemistry 62 (1990) 1709?1720.
[91] D.E. Ibbotson, J.A. Mucha, D.L. Flamm, and J.M. Cook, Plasmaless dry etching of silicon with
fluorine-containing compounds, Journal of Applied Physics 56 (1984) 2939-2942. doi:10.1063/1.333834
65
[92] L.R. Arana, N. de Mas, R. Schmidt, A.J. Franz, M.A. Schmidt, and K.F. Jensen, Isotropic etching of
silicon in fluorine gas for MEMS micromachining, Journal of Micromechanics and Microengineering 17
(2007) 384?392. doi:10.1088/0960-1317/17/2/026
[93] F. L?rmer, S. Franssila, L. Sainiemi, and K. Kolari, Deep reactive ion etching, in V. Lindroos, M.
Tilli, A. Lehto, and T. Motooka (Eds.), Handbook of Silicon Based MEMS Materials and Technologies,
William Andrew/Elsevier, 2010.
[94] H.V. Jansen, M.J.D. Boer, S. Unnikrishnan, M.C. Louwerse, and M.C. Elwenspoek, Black silicon
method X: a review on high speed and selective plasma etching of silicon with profile control: an
in-depth comparison between Bosch and cryostat DRIE processes as a roadmap to next generation
equipment, Journal of Micromechanics and Microengineering 19 (2009) 033001. doi:10.1088/0960-
1317/19/3/033001
[95] F. Laermer and A. Schilp, Method of anisotropically etching silicon, Patent #US5501893, March 26,
1996.
[96] I.W. Rangelow, High-resolution tri-level process by downstream-microwave rf-biased etching,
Proceedings of SPIE 1392 (1991) 180-184. doi:10.1117/12.48912
[97] S. Chen, Y. Lin, J. Wu, L. Horng, and C. Cheng, Parameter optimization for an ICP deep silicon
etching system, Microsystem Technologies 13 (2007) 465-474. doi:10.1007/s00542-006-0211-2
[98] M. Brunet, P. Dubreuil, H. Mahfoz-Kotb, A. Gouantes, and A. Dorthe, Factorial experimental
design applied to DRIE for optimised process in power electronics applications requiring high-
aspect ratio trenches, Microsystem Technologies 15 (2009) 1449-1457. doi:10.1007/s00542-009-0893-3
[99] B. Volland, F. Shi, P. Hudek, H. Heerlein, and I.W. Rangelow, Dry etching with gas chopping
without rippled sidewalls, Journal of Vacuum Science & Technology B 17 (1999) 2768-2771.
doi:10.1116/1.591061
[100] D. Nilsson, S. Jensen, and A. Menon, Fabrication of silicon molds for polymer optics, Journal of
Micromechanics and Microengineering 13 (2003) S57-S61. doi:10.1088/0960-1317/13/4/309
[101] K. Weng, M. Wang, and P. Tsai, Planarize the Sidewall Ripples of Silicon Deep Reactive Ion
Etching, Technical Proceedings of the 2004 NSTI Nanotechnology Conference and Trade Show 1 (2004)
473-476.
[102] M. de Boer, J. Gardeniers, H. Jansen, E. Smulders, M. Gilde, G. Roelofs, J. Sasserath, and M.
Elwenspoek, Guidelines for etching silicon MEMS structures using fluorine high-density plasmas
at cryogenic temperatures, Journal of Microelectromechanical Systems 11 (2002) 385-401.
doi:10.1109/JMEMS.2002.800928
[103] R.A. Gottscho, C.W. Jurgensen, and D.J. Vitkavage, Microscopic uniformity in plasma etching,
Journal of Vacuum Science & Technology B 10 (1992) 2133-2147. doi:10.1116/1.586180
[104] J. Kiihamaki and S. Franssila, Pattern shape effects and artefacts in deep silicon etching, Jounal of
Vacuum Science and Technology A 17 (1999) 2280-2285. doi:10.1116/1.581761
[105] T. Humphreys, J. Andersson, U. S?dervall, and T. Melvin, Microfluidic interconnects for DNA
loading, Journal of Micromechanics and Microengineering 19 (2009) 105024. doi:10.1088/0960-
1317/19/10/105024
[106] J. Kiiham?ki and S. Franssila, Deep Silicon Etching in Inductively Coupled Plasma Reactor for
MEMS, Physica Scripta T79 (1999) 250-254. doi:10.1238/Physica.Topical.079a00250
[107] A.M. Hynes, H. Ashraf, J.K. Bhardwaj, J. Hopkins, I. Johnston, and J.N. Shepherd, Recent advances
in silicon etching for MEMS using the ASE(TM) process, Sensors and Actuators A 74 (1999) 13-17.
doi:10.1016/S0924-4247(98)00326-4
[108] K.R. Williams, K. Gupta, and M. Wasilik, Etch rates for micromachining processing-Part II, Journal
of Microelectromechanical Systems 12 (2003) 761-778. doi:10.1109/JMEMS.2003.820936
[109] H. Zhang, H. Guo, Z. Chen, G. Zhang, and Z. Li, Application of PECVD SiC in glass
micromachining, Journal of Micromechanics and Microengineering 17 (2007) 775-780. doi:10.1088/0960-
1317/17/4/014
[110] C. Iliescu, J. Jing, F.E. Tay, J. Miao, and T. Sun, Characterization of masking layers for deep wet
etching of glass in an improved HF/HCl solution, Surface and Coatings Technology 198 (2005) 314-
318. doi:10.1016/j.surfcoat.2004.10.094
[111] C. Khan Malek, L. Robert, J. Boy, and P. Blind, Deep microstructuring in glass for microfluidic
applications, Microsystem Technologies 13 (2007) 447-453. doi:10.1007/s00542-006-0185-0
[112] C.H. Lin, G.B. Lee, Y.H. Lin, and G.L. Chang, Fabrication of microfluidic systems on soda-lime
glass, Journal of micromechanics and microengineering 11 (2001) 726?732. doi:10.1088/0960-
1317/11/6/316
66
[113] A. Grosse, M. Grewe, and H. Fouckhardt, Deep wet etching of fused silica glass for microfluidic
devices, Journal of Micromechanics and Microengineering 11 (2001) 257?262. doi:10.1088/0960-
1317/11/3/315
[114] C. Iliescu, B. Chen, and J. Miao, On the wet etching of Pyrex glass, Sensors and Actuators A 143
(2008) 154-161. doi:10.1016/j.sna.2007.11.022
[115] M.A. Gr?tillat, F. Paoletti, P. Thiebaud, S. Roth, M. Koudelka-Hep, and N.F. de Rooij, A new
fabrication method for borosilicate glass capillary tubes with lateral inlets and outlets, Sensors and
Actuators A 60 (1997) 219-222.
[116] C. Iliescu, J. Miao, and F.E.H. Tay, Optimization of an amorphous silicon mask PECVD process for
deep wet etching of Pyrex glass, Surface and Coatings Technology 192 (2005) 43-47.
doi:10.1016/j.surfcoat.2004.03.043
[117] T. Corman, P. Enoksson, and G. Stemme, Deep wet etching of borosilicate glass using an anodically
bonded silicon substrate as mask, Journal of micromechanics and microengineering 8 (1998) 84?87.
doi:10.1088/0960-1317/8/2/010
[118] C. Iliescu, J. Miao, and F.E.H. Tay, Stress control in masking layers for deep wet micromachining of
Pyrex glass, Sensors and Actuators A 117 (2005) 286-292. doi:10.1016/j.sna.2004.03.004
[119] I. Steingoetter and H. Fouckhardt, Deep fused silica wet etching using an Au-free and stress-
reduced sputter-deposited Cr hard mask, Journal of Micromechanics and Microengineering 15 (2005)
2130?2135. doi:10.1088/0960-1317/15/11/019
[120] F. Ceyssens and R. Puers, Deep etching of glass wafers using sputtered molybdenum masks,
Journal of Micromechanics and Microengineering 19 (2009) 067001.
[121] M. Bu, T. Melvin, G.J. Ensell, J.S. Wilkinson, and A.G.R. Evans, A new masking technology for deep
glass etching and its microfluidic application, Sensors and Actuators A 115 (2004) 476-482.
doi:10.1016/j.sna.2003.12.013
[122] R.S. Keynton, T.J. Roussel, M.M. Crain, D.J. Jackson, D.B. Franco, J.F. Naber, K.M. Walsh, and R.P.
Baldwin, Design and development of microfabricated capillary electrophoresis devices with
electrochemical detection, Analytica Chimica Acta 507 (2004) 95?105.
[123] C. Iliescu, F.E. Tay, and J. Miao, Strategies in deep wet etching of Pyrex glass, Sensors and Actuators
A 133 (2007) 395-400. doi:10.1016/j.sna.2006.06.044
[124] T. Arakawa, Y. Sato, T. Ueno, T. Funatsu, and S. Shoji, Pinhole-free Pyrex glass etching using HF-
H2SO4 mixed acid and its applications for a PDMS microflow system, TRANSDUCERS '05, Digest
of Technical Papers 2 (2005) 1489-1492.
[125] W.I. Jang, C.A. Choi, M.L. Lee, C.H. Jun, and Y.T. Kim, Fabrication of MEMS devices by using
anhydrous HF gas-phase etching with alcoholic vapor, Journal of Micromechanics and
Microengineering 12 (2002) 297?306. doi:10.1088/0960-1317/12/3/316
[126] P.W. Leech, Reactive ion etching of quartz and silica-based glasses in CF4/CHF3 plasmas, Vacuum
55 (1999) 191-196. doi:10.1016/S0042-207X(99)00146-3
[127] E. Metwalli and C.G. Pantano, Reactive ion etching of glasses: Composition dependence, Nuclear
Instruments and Methods in Physics Research Section B 207 (2003) 21-27. doi:10.1016/S0168-
583X(03)00517-2
[128] A. Baram and M. Naftali, Dry etching of deep cavities in Pyrex for MEMS applications using
standard lithography, Journal of Micromechanics and Microengineering 16 (2006) 2287-2291.
doi:10.1088/0960-1317/16/11/006
[129] S. Ronggui and G.C. Righini, Characterization of reactive ion etching of glass and its applications
in integrated optics, Journal of Vacuum Science & Technology A 9 (1991) 2709-2712.
doi:10.1116/1.577229
[130] H. Chen and C. Fu, An investigation into the characteristics of the deep reactive ion etching of
quartz using SU-8 as a mask, Journal of Micromechanics and Microengineering 18 (2008) 105001.
doi:10.1088/0960-1317/18/10/105001
[131] T. Akashi and Y. Yoshimura, Deep reactive ion etching of borosilicate glass using an anodically
bonded silicon wafer as an etching mask, Journal of Micromechanics and Microengineering 16 (2006)
1051?1056. doi:10.1088/0960-1317/16/5/024
[132] T. Ichiki, Y. Sugiyama, R. Taura, T. Koidesawa, and Y. Horiike, Plasma applications for biochip
technology, Thin Solid Films 435 (2003) 62-68. doi:10.1016/S0040-6090(03)00370-5
[133] J. Park, N. Lee, J. Lee, J. Park, and H. Park, Deep dry etching of borosilicate glass using SF6 and
SF6/Ar inductively coupled plasmas, Microelectronic Engineering 82 (2005) 119-128.
doi:10.1016/j.mee.2005.07.006
[134] X. Li, T. Abe, and M. Esashi, Deep reactive ion etching of Pyrex glass using SF6 plasma, Sensors and
Actuators A 87 (2001) 139-145. doi:10.1016/S0924-4247(00)00482-9
67
[135] X. Li, T. Abe, Y. Liu, and M. Esashi, Fabrication of high-density electrical feed-throughs by deep-
reactive-ion etching of Pyrex glass, Journal of Microelectromechanical Systems 11 (2002) 625-630.
doi:10.1109/JMEMS.2002.805211
[136] H.C. Jung, W. Lu, S. Wang, L.J. Lee, and X. Hu, Etching of Pyrex glass substrates by inductively
coupled plasma reactive ion etching for micro/nanofluidic devices, Journal of Vacuum Science &
Technology B 24 (2006) 3162-3164. doi:10.1116/1.2388959
[137] K. Kolari, High etch selectivity for plasma etching SiO2 with AlN and Al2O3 masks, Microelectronic
Engineering 85 (2008) 985-987. doi:10.1016/j.mee.2007.12.037
[138] X. Li, L. Ling, X. Hua, M. Fukasawa, G.S. Oehrlein, M. Barela, and H.M. Anderson, Effects of Ar
and O2 additives on SiO2 etching in C4F8-based plasmas, Journal of Vacuum Science & Technology A
21 (2003) 284-293. doi:10.1116/1.1531140
[139] T. Akashi and Y. Yoshimura, Profile control of a borosilicate-glass groove formed by deep reactive
ion etching, Journal of Micromechanics and Microengineering 18 (2008) 105004. doi:10.1088/0960-
1317/18/10/105004
[140] K. Kolari, Deep plasma etching of glass with a silicon shadow mask, Sensors and Actuators A 141
(2008) 677-684. doi:10.1016/j.sna.2007.09.005
[141] M. Pavius, C. Hibert, P. Fluckiger, P. Renaud, L. Rolland, and M. Puech, Profile angle control in
SiO2 deep anisotropic dry etching for MEMS fabrication, Proceedings of IEEE MEMS 2004 (2004)
669-672.
[142] D.S. Park, M.W. Cho, and H. Lee, Effects of the impact angle variations on the erosion rate of glass
in powder blasting process, The International Journal of Advanced Manufacturing Technology 23 (2004)
444?450.
[143] P.J. Slikkerveer, P.C.P. Bouten, F.H. in't Veld, and H. Scholten, Erosion and damage by sharp
particles, Wear 217 (1998) 237-250. doi:10.1016/S0043-1648(98)00187-2
[144] M. Buijs and J.M.M. Pasmans, Erosion of glass by alumina particles: transitions and exponents,
Wear 184 (1995) 61-65. doi:10.1016/0043-1648(94)06560-8
[145] E. Belloy, A. Sayah, and M.A.M. Gijs, Powder blasting for three-dimensional microstructuring of
glass, Sensors and Actuators A 86 (2000) 231-237. doi:10.1016/S0924-4247(00)00447-7
[146] P.J. Slikkerveer, P.C.P. Bouten, and F.C.M. de Haas, High quality mechanical etching of brittle
materials by powder blasting, Sensors and Actuators A 85 (2000) 296-303. doi:10.1016/S0924-
4247(00)00343-5
[147] A. Sayah, P. Thivolle, V.K. Parashar, and M.A.M. Gijs, Fabrication of microfluidic mixers with
varying topography in glass using the powder-blasting process, Journal of Micromechanics and
Microengineering 19 (2009) 085024. doi:10.1088/0960-1317/19/8/085024
[148] E. Belloy, A. Sayah, and M.A.M. Gijs, Oblique powder blasting for three-dimensional
micromachining of brittle materials, Sensors and Actuators A 92 (2001) 358?363.
[149] E. Belloy, A. Pawlowski, A. Sayah, and M. Gijs, Microfabrication of high-aspect ratio and complex
monolithic structures in glass, Journal of Microelectromechanical Systems 11 (2002) 521-527.
[150] H. Wensink, H.V. Jansen, J.W. Berenschot, and M.C. Elwenspoek, Mask materials for powder
blasting, Journal of Micromechanics and Microengineering 10 (2000) 175-180. doi:10.1088/0960-
1317/10/2/313
[151] A. Sayah, V.K. Parashar, A.G. Pawlowski, and M.A.M. Gijs, Elastomer mask for powder blasting
microfabrication, Sensors and Actuators A 125 (2005) 84?90.
[152] A. Sayah, V.K. Parashar, and M.A.M. Gijs, LF55GN Photosensitive Flexopolymer: A New Material
for Ultrathick and High-Aspect-Ratio MEMS Fabrication, Journal of Microelectromechanical Systems
16 (2007) 564?570.
[153] D. Park, M. Cho, H. Lee, and W. Cho, Micro-grooving of glass using micro-abrasive jet machining,
Journal of Materials Processing Technology 146 (2004) 234-240. doi:10.1016/j.jmatprotec.2003.11.013
[154] A. Pawlowski, A. Sayah, and M. Gijs, Precision poly-(dimethyl siloxane) masking technology for
high-resolution powder blasting, Journal of Microelectromechanical Systems 14 (2005) 619-624.
doi:10.1109/JMEMS.2005.844745
[155] H. Yagyu, K. Sugano, S. Hayashi, and O. Tabata, Micropowder blasting with nanoparticles
dispersed polymer mask for rapid prototyping of glass chip, Journal of Micromechanics and
Microengineering 15 (2005) 1236-1241. doi:10.1088/0960-1317/15/6/014
[156] Comco inc., http://www.comcoinc.com/
[157] P. Lerou, G.C.F. Venhorst, C.F. Berends, T.T. Veenstra, M. Blom, J.F. Burger, H.J.M. ter Brake, and H.
Rogalla, Fabrication of a micro cryogenic cold stage using MEMS-technology, Journal of
Micromechanics and Microengineering 16 (2006) 1919?1925. doi:10.1088/0960-1317/16/10/002
68
[158] H. Wensink, S. Schlautmann, M.H. Goedbloed, and M.C. Elwenspoek, Fine tuning the roughness
of powder blasted surfaces, Journal of Micromechanics and Microengineering 12 (2002) 616?620.
[159] T. Diepold and E. Obermeier, Smoothing of ultrasonically drilled holes in borosilicate glass by wet
chemical etching, Journal of Micromechanics and Microengineering 6 (1996) 29-32.
[160] A. Daridon, V. Fascio, J. Lichtenberg, R. W?trich, H. Langen, E. Verpoorte, and N.F. de Rooij,
Multi-layer microfluidic glass chips for microanalytical applications, Fresenius' Journal of Analytical
Chemistry 371 (2001) 261-269.
[161] I. Basak and A. Ghosh, Mechanism of spark generation during electrochemical discharge
machining: a theoretical model and experimental verification, Journal of Materials Processing
Technology 62 (1996) 46-53. doi:10.1016/0924-0136(95)02202-3
[162] X. Zeng, X. Mao, S.B. Wen, R. Greif, and R.E. Russo, Energy deposition and shock wave
propagation during pulsed laser ablation in fused silica cavities, Journal of Physics D: Applied
Physics 37 (2004) 1132?1136. doi:10.1088/0022-3727/37/7/029
[163] H. Niino, Y. Kawaguchi, T. Sato, A. Narazaki, T. Gumpenberger, and R. Kurosaki, Laser ablation of
toluene liquid for surface micro-structuring of silica glass, Applied Surface Science 252 (2006) 4387-
4391. doi:10.1016/j.apsusc.2005.07.084
[164] C.K. Chung, Y.C. Sung, G.R. Huang, E.J. Hsiao, W.H. Lin, and S.L. Lin, Crackless linear through-
wafer etching of Pyrex glass using liquid-assisted CO2 laser processing, Applied Physics A 94 (2008)
927-932. doi:10.1007/s00339-008-4863-x
[165] S. Nikumb, Q. Chen, C. Li, H. Reshef, H.Y. Zheng, H. Qiu, and D. Low, Precision glass machining,
drilling and profile cutting by short pulse lasers, Thin Solid Films 477 (2005) 216-221.
[166] S. Youn, M. Takahashi, H. Goto, and R. Maeda, Fabrication of micro-mold for glass embossing
using focused ion beam, femto-second laser, eximer laser and dicing techniques, Journal of
Materials Processing Technology 187-188 (2007) 326-330. doi:10.1016/j.jmatprotec.2006.11.120
[167] S. Youn, M. Takahashi, H. Goto, and R. Maeda, Microstructuring of glassy carbon mold for glass
embossing - Comparison of focused ion beam, nano/femtosecond-pulsed laser and mechanical
machining, Microelectronic Engineering 83 (2006) 2482-2492. doi:10.1016/j.mee.2006.05.007
[168] E. Eklund and A. Shkel, Glass Blowing on a Wafer Level, Journal of Microelectromechanical Systems
16 (2007) 232-239.
[169] S.W. Tsai, M. Loughran, and I. Karube, Development of a microchip for 2-dimensional capillary
electrophoresis, Journal of Micromechanics and Microengineering 14 (2004) 1693?1699.
[170] A. Sayah, D. Solignac, T. Cueni, and M.A.M. Gijs, Development of novel low temperature bonding
technologies for microchip chemical analysis applications, Sensors and Actuators A 84 (2000) 103-
108. doi:10.1016/S0924-4247(99)00346-5
[171] S. Christiansen, R. Singh, and U. Gosele, Wafer Direct Bonding: From Advanced Substrate
Engineering to Future Applications in Micro/Nanoelectronics, Proceedings of the IEEE 94 (2006)
2060-2106.
[172] Q.Y. Tong and U.M. G?sele, Wafer Bonding and Layer Splitting for Microsystems, Advanced
Materials 11 (1999) 17.
[173] M.A. Schmidt, Wafer-to-wafer bonding for microstructure formation, Proceedings of the IEEE 86
(1998) 1575-1585.
[174] M. Esashi, Wafer level packaging of MEMS, Journal of Micromechanics and Microengineering 18 (2008)
073001. doi:10.1088/0960-1317/18/7/073001
[175] T. Ito, K. Sobue, and S. Ohya, Water glass bonding for micro-total analysis system, Sensors and
Actuators B 81 (2002) 187-195.
[176] J. Haisma and G. Spierings, Contact bonding, including direct-bonding in a historical and recent
context of materials science and technology, physics and chemistry Historical review in a broader
scope and comparative outlook, Materials Science & Engineering R 37 (2002) 1-60.
[177] U. G?sele and Q.Y. Tong, Semiconductor wafer bonding, Annual Review of Materials Science 28
(1998) 215-241. doi:10.1146/annurev.matsci.28.1.215
[178] F. Niklaus, G. Stemme, J.-.Q. Lu, and R.J. Gutmann, Adhesive wafer bonding, Journal of Applied
Physics 99 (2006) 031101-28. doi:10.1063/1.2168512
[179] ?. Vallin, K. Jonsson, and U. Lindberg, Adhesion quantification methods for wafer bonding,
Materials Science & Engineering R 50 (2005) 109-165.
[180] Q. Tong and U. G?sele, SemiConductor Wafer Bonding: Science and Technology, Wiley-Interscience,
1999.
[181] M. Alexe and U. G?sele, Wafer Bonding, Springer, 2004.
[182] J.A. Dziuban, Bonding in Microsystem Technology, Springer, 2006.
69
[183] U. G?sele, Q.Y. Tong, A. Schumacher, G. Kr?uter, M. Reiche, A. Pl??l, P. Kopperschmidt, T.H. Lee,
and W.J. Kim, Wafer bonding for microsystems technologies, Sensors and Actuators A 74 (1999) 161-
168.
[184] A. Van Wieringen and N. Warmoltz, On the permeation of hydrogen and helium in single crystal
silicon and germanium at elevated temperatures, Physica 22 (1956) 849-865. doi:10.1016/S0031-
8914(56)90039-8
[185] J. Lichtenberg, E. Verpoorte, and N.F.D. Rooij, Sample preconcentration by field amplification
stacking for microchip-based capillary electrophoresis, Electrophoresis 22 (2001) 258-271.
doi:10.1002/1522-2683(200101)22:2<258::AID-ELPS258>3.0.CO;2-4
[186] M.M.R. Howlader, S. Suehara, and T. Suga, Room temperature wafer level glass/glass bonding,
Sensors and Actuators A 127 (2006) 31-36.
[187] A.D. Brooks, R.P. Donovan, and C.A. Hardesty, Low-Temperature Electrostatic Silicon-to-Silicon
Seals Using Sputtered Borosilicate Glass, Journal of The Electrochemical Society 119 (1972) 545-546.
doi:10.1149/1.2404250
[188] A. Hanneborg, M. Nese, and P. Ohlckers, Silicon-to-silicon anodic bonding with a borosilicate glass
layer, Journal of Micromechanics and Microengineering 1 (1991) 139-144. doi:10.1088/0960-1317/1/3/002
[189] A. Berthold, L. Nicola, P.M. Sarro, and M.J. Vellekoop, Glass-to-glass anodic bonding with
standard IC technology thin films as intermediate layers, Sensors and Actuators A 82 (2000) 224-228.
doi:10.1016/S0924-4247(99)00376-3
[190] D.J. Lee, B.K. Ju, Y.H. Lee, J. Jang, and M.H. Oh, Glass-to-glass anodic bonding for high vacuum
packaging of microelectronics and its stability, Proceedings of IEEE MEMS 2000 (2000) 253-258.
[191] Affymetrix, Inc., http://www.affymetrix.com/
[192] Dolomite Centre Ltd, http://www.dolomite-microfluidics.com/
[193] Micronit Microfluidics BV, http://www.micronit.com/
[194] microfluidic ChipShop, http://www.microfluidic-chipshop.com/
[195] The Swagelok Company, http://www.swagelok.com/
[196] Upchuch Scientific, Inc., http://www.upchurch.com/
[197] Valco Instruments Co. Inc., http://www.vici.com/
[198] J.W. Batts, all about FITTINGS, Upchurch Scientific, 2003.
[199] C. Fredrickson and Z. Fan, Macro-to-micro interfaces for microfluidic devices, Lab on a Chip 4
(2004) 526-533. doi:10.1039/b410720a
[200] Z. Yang and R. Maeda, Socket with built-in valves for the interconnection of microfluidic chips to
macro constituents, Journal of Chromatography A 1013 (2003) 29-33. doi:10.1016/S0021-9673(03)01125-
7
[201] J.H. Tsai and L. Lin, Micro-to-macro fluidic interconnectors with an integrated polymer sealant,
Journal of Micromechanics and Microengineering 11 (2001) 577-581.
[202] Y. Tai, S. Wu, and E.F. Meng, Micromachined fluidic coupler and method of making the same, Patent
#US6672629, January 6, 2004.
[203] M.T. Blom, E. Chmela, J.G.E. Gardeniers, J.W. Berenschot, M. Elwenspoek, R. Tijssen, and A. van
den Berg, Local anodic bonding of Kovar to Pyrex aimed at high-pressure, solvent-resistant
microfluidic connections, Journal of Micromechanics and Microengineering 11 (2001) 382?385.
[204] S. Ng, Z. Wang, and N. de Rooij, Microfluidic connectors by ultrasonic welding, Microelectronic
Engineering 86 (2009) 1354-1357. doi:10.1016/j.mee.2009.01.048
[205] E.S. Lee, D. Howard, E. Liang, S.D. Collins, and R.L. Smith, Removable tubing interconnects for
glass-based micro-fluidic systems made using ECDM, Journal of Micromechanics and
Microengineering 14 (2004) 535-541.
[206] A. Puntambekar and C.H. Ahn, Self-aligning microfluidic interconnects for glass-and plastic-based
microfluidic systems, Journal of Micromechanics and Microengineering 12 (2002) 35-40.
[207] A.V. Pattekar and M.V. Kothare, Novel microfluidic interconnectors for high temperature and
pressure applications, Journal of Micromechanics and Microengineering 13 (2003) 337-345.
[208] C.F. Chen, J. Liu, L.P. Hromada, C.W. Tsao, C.C. Chang, and D.L. DeVoe, High-pressure needle
interface for thermoplastic microfluidics, Lab on a Chip 9 (2009) 50-55.
[209] IDEX Health & Science 2009-2010 Catalog - Products for Chromatography and Fluidic Applications, IDEX
Health & Science LLC, 2009.
[210] J. Atencia, G.A. Cooksey, A. Jahn, J.M. Zook, W.N. Vreeland, and L.E. Locascio, Magnetic
connectors for microfluidic applications, Lab on a Chip 10 (2010) 246-249. doi:10.1039/b913331c
[211] V. Nittis, R. Fortt, C.H. Legge, and A.J.D. Mello, A high-pressure interconnect for chemical
microsystem applications, Lab on a Chip 1 (2001) 148-152. doi:10.1039/b107836b
70
[212] K. Han, R.D. McConnell, C.J. Easley, J.M. Bienvenue, J.P. Ferrance, J.P. Landers, and A.B. Frazier,
An active microfluidic system packaging technology, Sensors and Actuators B 122 (2007) 337-346.
doi:10.1016/j.snb.2006.06.028
[213] R.M. Tiggelaar, Silicon-technology based microreactors for high-temperature heterogeneous partial
oxidation reactions, Ph.D. thesis, University of Twente, Enschede, The Netherlands, 2004.
[214] R.F. Renzi, Edge compression manifold apparatus, Patent #US6832787, December 21, 2004.
[215] D.M. Hartmann, J.T. Nevill, K.I. Pettigrew, G. Votaw, P. Kung, and H.C. Crenshaw, A low-cost,
manufacturable method for fabricating capillary and optical fiber interconnects for microfluidic
devices, Lab on a Chip 8 (2008) 609-616.
[216] C.H. Chiou and G.B. Lee, Minimal dead-volume connectors for microfluidics using PDMS casting
techniques, Journal of Micromechanics and Microengineering 14 (2004) 1484-1490.
[217] A.M. Christensen, D.A. Chang-Yen, and B.K. Gale, Characterization of interconnects used in PDMS
microfluidic systems, Journal of Micromechanics and Microengineering 15 (2005) 928-934.
doi:10.1088/0960-1317/15/5/005
[218] M. Quaglio, G. Canavese, E. Giuri, S.L. Marasso, D. Perrone, M. Cocuzza, and C.F. Pirri, Evaluation
of different PDMS interconnection solutions for silicon, Pyrex and COC microfluidic chips, Journal
of Micromechanics and Microengineering 18 (2008) 55012-55012.
[219] T.G. Park, J.J. Kim, S.H. Jung, H.J. Song, J.K. Chang, D.C. Han, and S.S. Yang, Fabrication of
Megasonic-Agitated Module for Improving the Characteristics of Wet Etching, Japanese Journal of
Applied Physics 47 (2008) 5262-5269. doi:10.1143/JJAP.47.5262
[220] P. ?stman, L. Luosujarvi, M. Haapala, K. Grigoras, R.A. Ketola, T. Kotiaho, S. Franssila, and R.
Kostiainen, Gas Chromatography-Microchip Atmospheric Pressure Chemical Ionization-Mass
Spectrometry, Analytical Chemistry 78 (2006) 3027-3031. doi:10.1021/ac052260a
[221] M.J. Kim, L.A. Gruenke, R.J. Saia, and S.S. Cohen, Inhibition of acid etching of Pt by pre-exposure
to oxygen plasma, Applied Physics Letters 44 (1984) 462-464. doi:10.1063/1.94767
[222] M. Hara, Y. Shinata, and S. Hashimoto, The galvanic corrosion behavior of coupled nickel-
platinum in molten Na2CO3, Corrosion Science 39 (1997) 627-638.
[223] P. Rangsten, O. Vallin, K. Hermansson, and Y. Backlund, Quartz-to-Quartz Direct Bonding, Journal
of The Electrochemical Society 146 (1999) 1104-1105. doi:10.1149/1.1391728
[224] G. Zhuang, Q. Jin, J. Liu, H. Cong, K. Liu, J. Zhao, M. Yang, and H. Wang, A low temperature
bonding of quartz microfluidic chip for serum lipoproteins analysis, Biomedical Microdevices 8
(2006) 255-261. doi:10.1007/s10544-006-9142-z
[225] F.P. Fehlner, N.J. Binkowski, K.R. Salisbury, and L. Button, Alumina barrier layers on LCD glass,
Journal of Non-Crystalline Solids 195 (1996) 89-94. doi:10.1016/0022-3093(95)00473-4
[226] E. Aubry, M. Ghazzal, V. Demange, N. Chaoui, D. Robert, and A. Billard, Poisoning prevention of
TiO2 photocatalyst coatings sputtered on soda-lime glass by intercalation of SiNx diffusion
barriers, Surface and Coatings Technology 201 (2007) 7706-7712. doi:10.1016/j.surfcoat.2007.03.003
[227] G. Kychakoff, R.D. Howe, R.K. Hanson, M.C. Drake, R.W. Pitz, M. Lapp, and C.M. Penney,
Visualization of turbulent flame fronts with planar laser-induced fluorescence, Science 224 (1984)
382-384. doi:10.1126/science.224.4647.382
[228] M.C. Thurber and R.K. Hanson, Simultaneous imaging of temperature and mole fraction using
acetone planar laser-induced fluorescence, Experiments in Fluids 30 (2001) 93-101.
doi:10.1007/s003480000142
[229] A.P. London, A.A. Ay?n, A.H. Epstein, S.M. Spearing, T. Harrison, Y. Peles, and J.L. Kerrebrock,
Microfabrication of a high pressure bipropellant rocket engine, Sensors and Actuators A 92 (2001)
351-357. doi:10.1016/S0924-4247(01)00571-4
[230] J. Lee, K. Naeli, H. Hunter, J. Berg, T. Wright, C. Courcimault, N. Naik, M. Allen, O. Brand, A.
Glezer, and W.P. King, Characterization of liquid and gaseous micro- and nanojets using
microcantilever sensors, Sensors and Actuators A 134 (2007) 128-139. doi:10.1016/j.sna.2006.05.014
[231] J. P?l, T.J. Kauppila, S. Franssila, T. Kotiaho, and R. Kostiainen, Ionspray microchip, Rapid
Communications in Mass Spectrometry 24 (2010) 2584-2590. doi:10.1002/rcm.4678
[232] J. Eijkel, Chip-based HPLC: the quest for the perfect column, Lab on a Chip 7 (2007) 815.
doi:10.1039/b707464f
[233] K. Ohno, K. Tachikawa, and A. Manz, Microfluidics: Applications for analytical purposes in
chemistry and biochemistry, Electrophoresis 29 (2008) 4443-4453. doi:10.1002/elps.200800121
[234] A. de Mello, On-chip chromatography: the last twenty years, Lab on a Chip 2 (2002) 48N-54N.
doi:10.1039/b207266c
71
[235] A. Benvenuto, V. Guarnieri, L. Lorenzelli, C. Collini, M. Decarli, A. Adami, C. Potrich, L. Lunelli, R.
Canteri, and C. Pederzolli, Fabrication of a MEMS-based separation module for liquid
chromatography, Sensors and Actuators B 130 (2008) 181-186. doi:10.1016/j.snb.2007.07.111
[236] M. McEnery, A. Tan, J.D. Glennon, J. Alderman, J. Patterson, and S.C. O'Mathuna, Liquid
chromatography on-chip: progression towards a ?-total analysis system, The Analyst 125 (2000) 25-
27.
[237] W. De Malsche, H. Eghbali, D. Clicq, J. Vangelooven, H. Gardeniers, and G. Desmet, Pressure-
driven reverse-phase liquid chromatography separations in ordered nonporous pillar array
columns, Analytical Chemistry 79 (2007) 5915-5926.
[238] W. De Malsche, H. Gardeniers, and G. Desmet, Experimental study of porous silicon shell pillars
under retentive conditions, Analytical Chemistry 80 (2008) 5391-5400. doi:10.1021/ac800424q
[239] G. Ocvirk, E. Verpoorte, A. Manz, M. Grasserbauer, and H. Widmer, High-Performance Liquid-
Chromatography Partially Integrated Onto a Silicon Chip, Analytical Methods and Instrumentation 2
(1995) 74-82.
[240] A. Ishida, T. Yoshikawa, M. Natsume, and T. Kamidate, Reversed-phase liquid chromatography on
a microchip with sample injector and monolithic silica column, Journal of Chromatography A 1132
(2006) 90-98. doi:10.1016/j.chroma.2006.07.025
[241] J. Wen, C. Guillo, J.P. Ferrance, and J.P. Landers, Microfluidic-Based DNA Purification in a Two-
Stage, Dual-Phase Microchip Containing a Reversed-Phase and a Photopolymerized Monolith,
Analytical Chemistry 79 (2007) 6135-6142.
[242] D.S. Reichmuth, T.J. Shepodd, and B.J. Kirby, Microchip HPLC of peptides and proteins, Analytical
Chemistry 77 (2005) 527-533.
[243] B. He, N. Tait, and F. Regnier, Fabrication of Nanocolumns for Liquid Chromatography, Analytical
Chemistry 70 (1998) 3790-3797.
[244] A. Bhattacharyya and C.M. Klapperich, Thermoplastic microfluidic device for on-chip purification
of nucleic acids for disposable diagnostics, Analytical Chemistry 78 (2006) 788-792.
[245] J. Liu, C. Chen, C. Tsao, C. Chang, C. Chu, and D.L. DeVoe, Polymer Microchips Integrating Solid-
Phase Extraction and High-Performance Liquid Chromatography Using Reversed-Phase
Polymethacrylate Monoliths, Analytical Chemistry 81 (2009) 2545-2554. doi:10.1021/ac802359e
[246] Y. Yang, C. Li, J. Kameoka, K.H. Lee, and H.G. Craighead, A polymeric microchip with integrated
tips and in situ polymerized monolith for electrospray mass spectrometry, Lab on a Chip 5 (2005)
869-876.
[247] D.A. Mair, E. Geiger, A.P. Pisano, J.M.J. Fr?chet, and F. Svec, Injection molded microfluidic chips
featuring integrated interconnects, Lab on a Chip 6 (2006) 1346-1354.
[248] A. Gaspar, M.E. Piyasena, and F.A. Gomez, Fabrication of Fritless Chromatographic Microchips
Packed with Conventional Reversed-Phase Silica Particles, Analytical Chemistry 77 (2005) 527-533.
[249] C.Y. Shih, Y. Chen, W. Li, J. Xie, Q. He, and Y.C. Tai, An integrated system for on-chip temperature
gradient interaction chromatography, Sensors and Actuators A 127 (2006) 207-215.
[250] A. Manz, Y. Miyahara, J. Miura, Y. Watanabe, H. Miyagi, and K. Sato, Design of an open-tubular
column liquid chromatograph using silicon chip technology, Sensors and Actuators B 1 (1990) 249-
255. doi:10.1016/0925-4005(90)80210-Q
[251] L. Ceriotti, N.F. de Rooij, and E. Verpoorte, An Integrated Fritless Column for On-Chip Capillary
Electrochromatography with Conventional Stationary Phases, Analytical Chemistry 74 (2002) 639-
647. doi:10.1021/ac0109467
[252] S. Le Gac, J. Carlier, J. Camart, C. Cren-Oliv?, and C. Rolando, Monoliths for microfluidic devices
in proteomics, Journal of Chromatography B 808 (2004) 3-14. doi:10.1016/j.jchromb.2004.03.067
[253] K.W. Ro, J. Liu, and D.R. Knapp, Plastic microchip liquid chromatography-matrix-assisted laser
desorption/ionization mass spectrometry using monolithic columns, Journal of chromatography A
1111 (2006) 40-47.
[254] F.E. Regnier, Microfabricated Monolith Columns for Liquid Chromatography. Sculpting Supports
for Liquid Chromatography, Journal of High Resolution Chromatography 23 (2000) 19-26. doi:10.1002/
(SICI)1521-4168(20000101)23:1<19::AID-JHRC19>3.0.CO;2-J
[255] M. De Pra, W. De Malsche, G. Desmet, P.J. Schoenmakers, and W.T. Kok, Pillar-structured
microchannels for on-chip liquid chromatography: Evaluation of the permeability and separation
performance, Journal of Separation Science 30 (2007) 1453-1460. doi:10.1002/jssc.200600468
[256] J. Vangelooven, W.D. Malsche, F. Detobel, H. Gardeniers, and G. Desmet, High-Speed Shear-
Driven Flows Through Microstructured 1D-Nanochannels, Analytical Chemistry 81 (2009) 943-952.
doi:10.1021/ac801691e
72
[257] G.A. Lord, D.B. Gordon, P. Myers, and B.W. King, Tapers and restrictors for capillary
electrochromatography and capillary electrochromatography-mass spectrometry, Journal of
Chromatography A 768 (1997) 9-16. doi:10.1016/S0021-9673(96)01081-3
[258] M. De Pra, W.T. Kok, J.G.E. Gardeniers, G. Desmet, S. Eeltink, J.W. van Nieuwkasteele, and P.J.
Schoenmakers, Experimental study on band dispersion in channels structured with micropillars,
Analytical Chemistry 78 (2006) 6519-6525.
[259] W.D. Malsche, D. Clicq, V. Verdoold, P. Gzil, G. Desmet, and H. Gardeniers, Integration of porous
layers in ordered pillar arrays for liquid chromatography, Lab on a Chip 7 (2007) 1705-1711.
[260] V. Spikmans, S.J. Lane, B. Leavens, A. Manz, and N.W. Smith, On-line on-chip post-column
derivatisation reactions for pre-ionisation of analytes and cluster analysis in gradient ?-liquid
chromatography/electrospray mass spectrometry, Rapid Communications in Mass Spectrometry 16
(2002) 1377-1388. doi:10.1002/rcm.727
[261] J.F. Borowsky, B.C. Giordano, Q. Lu, A. Terray, and G.E. Collins, Electroosmotic Flow-Based Pump
for Liquid Chromatography on a Planar Microchip, Analytical Chemistry 80 (2008) 8287-8292.
doi:10.1021/ac801497r
[262] E. Meng, P. Li, and Y. Tai, A biocompatible Parylene thermal flow sensing array, Sensors and
Actuators A 144 (2008) 18-28. doi:10.1016/j.sna.2007.12.010
[263] C.Y. Shih, Y. Chen, J. Xie, Q. He, and Y.C. Tai, On-chip temperature gradient interaction
chromatography, Journal of Chromatography A 1111 (2006) 272-278.
[264] R.M. Tiggelaar, J.W. Berenschot, J.H.D. Boer, R.G.P. Sanders, J.G.E. Gardeniers, R.E. Oosterbroek,
A.V.D. Berg, and M.C. Elwenspoek, Fabrication and characterization of high-temperature
microreactors with thin film heater and sensor patterns in silicon nitride tubes, Lab on a Chip 5
(2005) 326-336.
[265] S. Tia and A.E. Herr, On-chip technologies for multidimensional separations, Lab on a Chip 9 (2009)
2524 - 2536. doi:10.1039/b900683b
[266] M.M. Robson, M.G. Cikalo, P. Myers, M.R. Euerby, and K.D. Bartle, Capillary
electrochromatography: A review, Journal of Microcolumn Separations 9 (1997) 357-372. doi:10.1002/
(SICI)1520-667X(1997)9:5<357::AID-MCS3>3.0.CO;2-0
[267] L. DeFrancesco, Product Review: Capillary Electrophoresis: Finding a Niche, Analytical Chemistry
73 (2001) 497 A-499 A. doi:10.1021/ac012512h
ISBN 978-952-60-3339-6
ISBN 978-952-60-3340-2 (PDF)
ISSN 1795-2239
ISSN 1795-4584 (PDF)
Search more related documents:TKK Dissertations 238
Download Document:TKK Dissertations 238

Set Home | Add to Favorites

All Rights Reserved Powered by Free Document Search and Download

Copyright © 2011
This site does not host pdf,doc,ppt,xls,rtf,txt files all document are the property of their respective owners. complaint#downhi.com
TOP