Content uploaded by Erkinjon G. Nazarov
Author content
All content in this area was uploaded by Erkinjon G. Nazarov on Oct 09, 2017
Content may be subject to copyright.
89
Planar Differential Mobility Spectrometry
for Clinical Breath Diagnostics
Erkinjon G. Nazarov, Timothy Postlethwaite,
Kenneth Markoski, Sophia Koo, and Jeffrey T. Borenstein
CONTENTS
6.1 Introduction .................................................................................................. 90
6.2 Ambient Pressure Ionization Instrument Classication:
Gas-Phase Mobility Spectrometry ............................................................. 92
6.2.1 Technology Origins ..........................................................................93
6.2.2 Chronological History of DMS and FAIMS Developments .......93
6.2.3 DMS versus FAIMS..........................................................................94
6.2.4 Technology Transfer ........................................................................94
6.3 Comparison between Conventional IMS and DMS Operation.............98
6.3.1 Operational Principle of Conventional IMS ................................98
6.3.2 Phenomenological Description of DMS Sensor Operation ......100
6.3.3 Comparison between the IMS and DMS Spectra ......................102
6.3.4 Use of DMS for Detection of Chemical Warfare Agents:
GA, AC, and CK .............................................................................102
6.3.5 Planar DMS Sensor Design Optimization for a
Particular Application ................................................................... 104
6.3.6 Peak Position in Planar DMS Spectra .........................................104
6.3.7 Peak Width in Planar DMS Spectra ............................................. 106
6.3.8 Optimization of the DMS Design: Gap Dimension Effect .......106
6.4 Categorization of DMS Sensor Operation in Integrated Systems.......109
6.4.1 Motivation for Integration of DMS with Other Analytical
Instruments ..................................................................................... 109
6.4.2 List of Products Where DMS Serves as System Engine ........... 110
6.4.3 Operation of a Single DMS as a Spectral Detector
in a Gas Analyzer ........................................................................... 111
6.4.4 DMS Operation as a Spectral Detector for Fractionating
Systems ............................................................................................ 113
6.4.5 DMS Operation as an Ambient Pressure Ion Pre-Filter for
Sophisticated Analytical Instruments ......................................... 113
6.5 Advantages of Tandem Systems for Specic Applications ..................115
6
90 Diagnostic Devices with Microuidics
6.5.1 The GC-DMS System: Protecting Crew Health and Safety
Aboard the International Space Station (ISS) .............................115
6.5.2 DMS Operation as a GC Detector for Measurements in
Harsh Environments .....................................................................117
6.5.3 Example of Chemical Noise Reduction in DMS-MS ................119
6.5.4 Separation of Isobaric Ion Species in DMS-MS .........................120
6.5.5 Selection and Identication of Isomeric Ion Species in
DMS-MS .......................................................................................... 121
6.5.6 DMS Operation as Interface in Front of IMS:
DMS-IMS2 System .......................................................................... 122
6.6 Biodefense Applications: Aerosolized Pathogens as Potential
Bioweapons ................................................................................................. 125
6.7 Medical Applications .................................................................................131
6.7.1 Breath Analysis ...............................................................................131
6.7.2 Cardiovascular Applications ........................................................132
6.7.3 Diagnosis and Monitoring of Tuberculosis ................................133
6.7.4 Rapid Detection of Invasive Aspergillosis ................................. 134
6.8 Conclusions .................................................................................................135
References .............................................................................................................136
6.1 Introduction
Recent experience in emerging infectious diseases ranging from severe
acute respiratory syndrome to avian inuenza highlights a critical need
for point-of-care diagnostics to immediately identify pulmonary infections.
For decades, investigators have recognized that many diseases result in
the production of distinctive volatiles or patterns of volatiles in human
breath. Until recently, however, the technology to sample, identify, and
classify these volatiles has been lacking. The result is that only certain dis-
eases, such as Helicobacter pylori infection, are being clinically diagnosed
through a breath test. The last several decades have brought advances in
microsystems technology, computing power, and algorithms capable of
rapid multivariate analysis. These new technologies enable inexpensive,
portable, automated point-of-care systems capable of detecting biomark-
ers generated by specic infectious agents. A new era of clinical diagno-
sis could develop around technologies and techniques using noninvasive
breath analysis. This advance could dramatically reduce the time spent by
patients in clinical settings, increase the precision of differential diagnosis,
shorten the response time for emerging epidemics, and reduce the cost bur-
den for patient care.
Infectious diseases are the number one cause of mortality in the world
and disproportionately ravage the developing world. Respiratory infectious
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
91Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
diseases such as pneumonia and tuberculosis account for more than 10%
of deaths in the world each year. In addition, many of the most common
opportunistic infections that accompany AIDS in the developing world are
also respiratory infections caused by bacteria or fungi. Together, respiratory
infections cause millions of deaths each year and result in untold personal
suffering and loss of productivity. In areas lacking robust diagnostic facilities
(typical of a vast majority of health care facilities in the developing world),
therapeutic approaches to respiratory illnesses are often empiric. Since many
respiratory diseases have similar presentation, lack of diagnostic accuracy
of specic pathogens often results in high rates of morbidity and mortal-
ity post-therapy, waste of scarce therapeutic resources, and, in cases such as
cystic brosis, the risk of additional infection.
Most pulmonary infections are diagnosed and treated based on patient his-
tory and clinical ndings rather than denitive lab-based tests. This can be
problematic because in many cases this information is insufcient to narrow
the etiology to a single pathogen, and the physician must treat with broader
spectrum medications than necessary. In recent years, signicant advances in
the laboratory diagnostics available to detect respiratory viral infections have
been achieved, including assays that use tissue culture, serology, and direct
examination, as well as some rapid diagnostic techniques and molecular assays.
Bacterial infections may also be conrmed via sputum and deep tracheal aspi-
rates that are cultured and tested with standard microbiological techniques.
In practice, however, these diagnostic tests have little impact on treatment
decisions because patients are treated empirically based on clinical suspicion
well before the studies are completed. If a point-of-care diagnostic tool were
available, treatment could be immediately tailored to the relevant pathogen,
improving the quality of care and reducing the spread of antibiotic resistance.
This chapter describes a device, and its technology underpinnings, which
is capable of revolutionizing clinical breath diagnostics based on a combina-
tion of its small size and power requirements, ease of use, and applicability
to a broad range of clinically relevant biomarker signatures. This technology,
based upon the principle of ion mobility spectrometry, generates signature
spectral patterns as distinctive as mass spectrometry, MS, but with far greater
simplicity and adaptability. The differential mobility spectrometer (DMS) is a
portable, handheld device that generates multidimensional biological spec-
tra from volatile compounds found in exhaled breath. The current prototype
model of the DMS ts in the palm of the hand, is highly durable, operates
at atmospheric pressure, and can be operated with standard batteries. The
technology is based upon the principles of ion mobility, where the specic
mobility of a target species, rather than the molecular weight, is tracked as a
function of electric eld. The ion mobility spectra provide enhanced resolu-
tion because species with similar molecular weights can be distinguished,
and further because the mobility of ions is a strong function of applied elec-
tric eld. In contrast to existing time-of-ight ion mobility spectrometers
(TOF-IMS), in which sensitivity is reduced as the instrument is miniaturized,
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
92 Diagnostic Devices with Microuidics
the DMS sensitivity is enhanced as the device is made smaller, due to greater
precision in the microscale device dimensions and control over local electric
elds. A substantial body of data has been gathered using the DMS device
in measurements of biomarkers of bacterial spores, human blood and urine,
and chemical weapons simulants at parts-per-trillion levels. In the following,
the technology for various implementations of the DMS will be described,
along with applications in chemical sensing (to demonstrate the mechanisms
involved in obtaining unique signature detection), biodefense, and in clinical
breath diagnostics.
6.2 Ambient Pressure Ionization Instrument
Classification: Gas-Phase Mobility Spectrometry
Ion-mobility-based spectrometers operating with one of a number of possible
ionization means are capable of providing highly sensitive, lab-quality detec-
tion of trace targeted chemicals in eld conditions. In this family of instru-
ments, sample molecules are ionized directly at ambient pressure, and then
the formed ions are identied according to their coefcient of mobility in a
drift gas, typically N2 or air. Because ionization and separation occur at ambi-
ent pressure, these instruments do not require cumbersome vacuum pumps,
and therefore they are suitable for miniaturization and eld use. Currently,
there are a number of analyzers employing atmospheric pressure ionization,
for example, conventional time-of-ight ion mobility spectrometry (IMS),1
differential ion mobility spectrometry or eld asymmetric ion mobility spec-
trometry (DMS/FAIMS),2 aspirated IMS (AIMS),3 and ambient pressure ion-
ization mass spectrometer (API MS). In these instruments, ions are generated
directly at ambient pressure, and then introduced into a vacuum for subse-
quent mass spectrometric (m/z) identication of ion species.4 Many types of
atmospheric pressure ionization analyzers have proven their effectiveness in
addressing analytical needs when used in laboratory conditions. In a labora-
tory environment, users have access to functions enabling extensive sample
preparation, including collection, pre-concentration, cleanup, and extrac-
tion. The advantage of incorporating these time-consuming pretreatment
steps is that sample complexity is decreased, and the overall quality of the
analysis is improved. In elded applications, there are limited resources and
time available for effective collection, pretreatment, and detection of targeted
components. Therefore, for these applications, it is desirable to identify novel
and rapid methods for sample collection and pretreatment. This need has
served as a motivation for intensive efforts over the past two decades toward
the development of instruments such as ion-mobility-based systems for ion
pre-ltration and pre-separation. In this chapter, we will focus on relatively
new differential mobility ion separation technology, which can be considered
as a next step in the evolution of ion mobility spectrometry technology for
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
93Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
applications in environmental detection, and, most importantly, to revolu-
tionize point-of-care clinical diagnostics.
6.2.1 Technology Origins
The DMS/FAIMS method for ion species separation/identication in gases
exploits the nonlinear dependence of an ion’s coefcient of mobility under
the effect of a strong electric eld was rst proposed in the Soviet Union dur-
ing the 1980s by M. Gorshkov (Siberian Academy of Science, Novosibirsk).5
The principle of operation for this new method involves separation of ions
in a gas ow stream subjected to a superimposed electric eld composed of a
high-frequency, asymmetric waveform RF and DC eld applied transverse to
the direction to gas stream. Development efforts for this technology contin-
ued in the Soviet Union by researchers from laboratories at the Siberian and
Uzbek Academies of sciences.2,6–11 One of the principal motivations behind
these efforts was the development of eld-deployable instruments for the
detection of explosives in the environment.
6.2.2 Chronological History of DMS and FAIMS Developments
During the late 1980s and early 1990s, the development process for DMS
and FAIMS diverged with the efforts of two independently operated
teams. The rst group, Gorshkov’s team at the Institute of Thermophysics
of the Siberian Academy of Science, built curved geometry devices where
ion separation occurs in analytical gap between two concentric cylindri-
cal electrodes. In this design, the nonhomogeneous electric eld is formed in
the analytical gap when superimposed RF and DC voltages are applied to
each of electrodes. This design was initially called eld ion spectrometry
and was eventually transferred to Mine Safety Appliances in Pittsburgh,
Pennsylvania, to further develop and explore commercialization opportu-
nities in this area.12 A general advantage of the cylindrical device is that
in the process of ion trajectory along analytical gap with inhomogeneous
electric elds, the analyzed ions can be focused, which results in enhancing
the efciency of ion transmission through the sensor. In cylindrical designs,
the ion’s transmission can be changed from zero at low electric elds up
to a maximum value at enhanced electric eld conditions. In addition, the
efciency of focusing also depends on the relative difference between high
eld and low eld ion mobility, dened by the ion’s “alpha parameter.”
Therefore, different ion species with distinct alpha parameters display dif-
ferent level of focusing, which complicates quantitative analysis and sys-
tem calibration. Over the next decade, the cylindrical design would become
known as High Field Asymmetric Waveform Ion Mobility Spectrometry or
FAIMS.13 One motivation for this approach was to try to take advantage
of the ion focusing properties of the inhomogeneous elds created with
curved electrodes.14,15
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
94 Diagnostic Devices with Microuidics
A second scientic team, which included members from two research insti-
tutes, Novosibirsk and Tashkent, worked on a planar sensor design.8,16 In this
conguration, ion separation occurs in a homogeneous electric eld in the gap
between two planar electrodes. This sensor design was initially referred to
as a drift spectrometer, which eventually changed to the more phenomeno-
logically descriptive name Differential Mobility Spectrometer (DMS). The
motivation for a planar approach was to create a device in which different
ion species (independent of their polarity) would be transmitted through
the sensor with the same efciency, without being inuenced by the value
and sign of each ion’s alpha parameters. Straightforward planar electrode
designs provide the highest efciency ion transmission even when separa-
tion voltages are turned off. This mode of operation is called the “transpar-
ent” regime of measurement, and enables the possibility of comparing the
DMS output with and without ion separation. These characteristics were
known to be lacking in the curved geometry cells.
6.2.3 DMS versus FAIMS
Both DMS and FAIMS share the same physical separation principle con-
ceived by Gorshkov, fundamentally based on the difference between the
high and low eld mobility of a particular gas-phase ion. They both pro-
duce a continuous spatial separation of ion species, which takes place under
superimposed effect of dispersion RF and compensation DC electric elds.
It was also recognized that different geometries imparted different analytical
properties, warranting distinguishing them with the separate terms FAIMS
and DMS. For instance, Figure 6.1 provides a comparison between mobil-
ity spectra obtained for the same bradykinin samples (with formation of
the same ion species m/z = 531 Da), which were obtained in two different
designs: planar and cylindrical. This direct comparison shows that planar
DMS provides higher resolution and sensitivity versus a cylindrical design.
For example, the analysis presented in Figure 6.1 shows that when both sys-
tems are tuned to operate with comparable sensitivity, the resolving power
of DMS is better by a factor of 2–4 times, and the operational speed is at least
20 times faster in the planar design.17
Table 6.1, which summarizes results reported from various scientic teams,
compares the performance for these two approaches. A review of this infor-
mation shows that the choice of geometry is particularly important, because
the output and analytical performance of the DMS/FAIMS devices for the
same samples can be signicantly different.
6.2.4 Technology Transfer
During the 1990s, the DMS/FAIMS technologies were transferred from the
East to the West. Figure 6.2 shows a chronological history of development
and migration between institutions for both designs that started in the 1980s.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
95Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
It also provides a list of the companies and institutions that have played a
major role in the growth of this technology from its starting points. The cylin-
drical version moved from the Siberian Academy of Sciences (Novosibirsk)
to the Mine Safety Appliances (MSA) company in Pittsburgh (see bottom
white color trajectory in Figure 6.2). The rst commercial prototype of
the cylindrical design was built and reported in 1995.12 The Mine Safety
Appliances cylindrical FAIMS device was utilized by the National Research
Council in Ottawa to couple to a mass spectrometer.14,15,18 This became the
basis of Ionalytics Corp., which was subsequently bought by Thermo Fisher
Scientic, Inc. in 2005 to commercialize this design as a FAIMS-MS system.
Currently, Thermo Scientic’s commercial product is called a FAIMS inter-
face, which works in combination with a heated electrospray ionization
source (HESI) and atmospheric pressure chemical ionization (APCI) of ion
probes commercially available in combination with a triple-stage quadru-
pole mass spectrometer.
In 2004, Dr. Richard D. Smith’s scientic team from Pacic Northwest
National Laboratory (PNNL) began work with a cylindrical FAIMS system.19
But later, after testing and comparison of performance of a new prototype
planar version of DMS sensors versus cylindrical designs, the group transi-
tioned to the at design approach.17
Prof. Rick Yost’s research team from the University of Florida also actively
worked (from 2004) with the FAIMS system, as shown in the chronologi-
cal map in Figure 6.2. In 2010, Yost’s team offered a new three-dimensional
(3D) geometry of the sensor with curved electrodes, which can be considered
as derivative of two-dimensional (2D) cylindrical designs. These sensors
are designed with hemispherical and spherical geometries.20 They reported
systematic experimental data for nine regular explosives. It conveyed that
3D designs provided enhanced sensitivity, resolution (up to twofold) and
Pl
anar
desi
gn Cylindrica
l
design
×25
f
12
–CV (V)
Signal
1110987
e
d
II
I
b
c
a
FIGURE 6.1
Comparison between the full spectra obtained by DMS and FAIMS for the same sample com-
pounds (bradykinin): solid line corresponds to planar design and the dotted lines are for FAIMS
design. The dashed line is fragment of DMS spectra (expanded ×25 for visualization) showing
the separation power of a planar versus a cylindrical design.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
96 Diagnostic Devices with Microuidics
TABLE 6.1
Comparison of Specic Features of Planar (DMS) and Cylindrical (FAIMS) Designs
Ions’ Residence
Time
Focusing/
Defocusing
Ion Species
RF
Dependence
on Peak
Width and
Intensity
Selectivity
(Relative
to DMS)
Modes of
Operation
Detection
(+) or (−)
ions
Time to
Mount/
Demount
from MS Cost
DMS 1–2 ms None—this
provides an
option to
generate
dispersion
plots
Moderate 1.0 Two modes
operation:
1. Filtering
(when
RF on)
2. Transparent
(when
RF off)
Simulta-
neously
detect
both
polarity
ions
A few minutes $
FAIMS Signicantly longer
At least 100 ms
Some ions
focused, some
defocused
Strong 0.2–0.3 Single mode
operation:
Filtering only
Detect
single
polarity of
ions {(+)
or (−)}
A few hours $$
DMS
advantage
over
FAIMS
1. Reduced ion
reaction losses
due to
diffusion
2. High-speed
measurements
possible
Coefcients of
ion
transmission
for different
species is
similar, which
make easy
quantitative
measurements
Operate with
different
values of
RF voltage
to turn
sensor in
optimal
regimes for
specic ion
species
Ref: Anal.
Chem.
2006,
78,3706
Fast
comparison of
ltered and
unltered
mass spectra
Simple to
switch
from one
polarity to
another
Rapid change
from
ESI-DMS-MS
to any other
conguration
Disposable
sensor
chips
under
develop-
ment.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
97Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
resolving power (1.5-fold to twofold) for six different explosives. The best ion
transmission was obtained in the spherical FAIMS cell, while a hemispheri-
cal FAIMS provided the best resolution and resolving power.21
A parallel trajectory related to the development of the DMS with pla-
nar electrodes started in 1996 from Tashkent (see upper green color trail in
Figure 6.2). This technology was brought to Prof. Gary Eiceman’s laboratory
at New Mexico State University (NMSU), a leading center for new develop-
ments in ion-mobility-based technologies. The rst tested prototype was a
micromachined DMS sensor built by the Charles Stark Draper Laboratory
in Cambridge, MA, as part of a Draper-funded collaboration with NMSU.22
During this Draper-initiated and funded project (1998–2001), the planar
geometry evolved further, in terms of a better understanding of the funda-
mentals of the DMS phenomena, miniaturization, and for providing new
analytical applications of the technology. In addition to scientic investiga-
tions, the rst prototypes of compact gas analyzers23 and DMS detectors for
use with gas chromatography (GC) were designed and built.24 At that time,
the technology began to be referred to in the literature as DMS, more closely
reecting the basic physical principles of operation.25 Later, these prototypes
ThermoFisher
Scientific
20152010200520001995199019851980
PNNL
PNNL
Ionalytics
Univ. of Florida
LU
UNC, Bruker
AB SCIEX
Univ. of Florida
MSA
NMSU
Owlstone
NEU
Sionex Corp.
Draper
Lab
Electronics
Institute.
Tashkent
Geophysical
Institute.
Novosibirsk
FAIMS/DMS Planar
Siberian Acad.
of Science,
Novosibirsk
Coaxial
FIGURE 6.2
A chronological history of DMS (planar) and FAIMS (coaxial) developments, including the
migration process of these two geometries between institutions. Trail for cylindrical is located
below the horizontal axis and for planar is presented above the horizontal axis. This historical
information was drawn from the literature and from personal communications with develop-
ers, which made contributions to this historical perspective. Used abbreviations: NMSU, New
Mexico State University; NEU, Northeastern University; UNC, University of North Carolina;
LU, Loughborough University; PNNL, Pacic Northwest Laboratory; MSA, Mine Safety
Appliances.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
98 Diagnostic Devices with Microuidics
were deployed as commercial prototypes at a Draper Laboratory spin-off
company called Sionex Corporation. During the period 2001–2010, Sionex
Corporation (following an OEM business model) provided numerous DMS
sensors and electronics kits to companies specializing in the production of
analytical systems for the detection and identication of traces of hazardous
chemicals in complex mixtures.
In 2004, Owlstone Ltd was established. This company utilized a modern-
ized planar multichannel design DMS sensor. The sensor design includes
multiple (>15) parallel operated microchannels with an individual channel
size height of 35 μm and length of 300 μm. Microscale analytical gaps enable
a signicant reduction in the residence time of ions, and can reach electric
elds almost two times higher without breakdown. Decreasing the residence
time of ions leads to increasing the speed of measurement and enhancing the
magnitude of the electric eld, leading to deeper fragmentation of analyte
ions in the process of passing ions through the analytical gap. Fragmentation
patterns can be used as a complementary piece of information about the
ion species, which can improve identication accuracy. A consequence of
using microscale channels is the necessity to increase the frequency of the
RF generators, which imposed additional technical challenges.26 The higher-
frequency version of the analytical cell was called ultrahigh-frequency
FAIMS. Based on this type of system, Owlstone produces several types of gas
analyzers, including a stand-alone instrument (LonestarTM) and an OEM sen-
sor module. Current micro- and macroscale gap (0.03–2 mm) DMS/FAIMS
technologies have been adopted by various researchers, working in collab-
oration with instrumentation companies: Prof. Richard Yost (University of
Florida) in collaboration with Agilent, and Prof. Gary Glish (University of
North Carolina at Chapel Hill) with Bruker Daltonics Inc.
6.3 Comparison between Conventional
IMS and DMS Operation
A primary focus of this chapter is to provide an overview of the different
implementations of DMS analyzers (stand-alone and in combination with
other technologies) that have been built and characterized on the basis of
a planar DMS sensor. To clarify the advantages of this relatively new tech-
nology, a technical description of DMS operation is presented, along with a
description of its features relative to conventional IMS.
6.3.1 Operational Principle of Conventional IMS
Classical time-of-ight ion mobility spectrometers (IMS) apply an instru-
mental approach, enabling characterization of ion species according to its
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
99Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
absolute values of coefcient mobility.1 This parameter can be considered
as a unique “ngerprint” feature of individual ion species.27 At moderate
electric elds, the coefcient of mobility reects structural information of ion
species, including a reduced mass μ and cross-sectional area Ω (or size) of the
analyzed ions, as described by Equation 6.1:
K
Ee
E
e
Nk
TT
ef
fe
ff
()
=
()
=æ
è
ç
ç
ö
ø
÷
÷
()
mn
p
mW
3
16
21
11
½
.
(6.1)
The schematic and principle of operation of time-of-ight IMS is presented in
Figure 6.3, in the left panel. An ion mobility spectrometer includes an ioniza-
tion chamber, an ion injection shutter (or ion gate), and an ion drift tube with
Detector
In time-of-flight IMS the absolute
value of coefficient mobility {K(0)} is
measured
In DMS the alpha parameter (dependence of
coefficient mobility from electric field) is measured
IMS spectrum
Drift time (ms)
–Ions
detector
+
DC or
scanning
voltage
K(E)=K(0)[1+α(E)]
RF Vc
DMS spectrometer
IMS spectrometer
Ion source
region Drift region
+Ions
detector
Planar electrode
Ion species trajectories
Ionizer Gas exhaust
Gas flow
neutrals
DMS spectrum
Compensation voltage (V)
–30–20 –100 10
FIGURE 6.3
Comparison between principles of operation for conventional IMS and DMS. Left panel:
Schematic and examples of output of IMS. Right panel: Operational schematic and examples of
output for planar differential mobility spectrometer.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
100 Diagnostic Devices with Microuidics
a homogeneously distributed low DC electric eld (E = 100–200 V/cm) for
pulling ion species in the direction of the detector (Faraday plate), the detec-
tor plate located at the end of drift tube. To inject ions from the ionization
chamber into the drift tube area, periodic electric pulses are applied to the
ion gate electrodes. After injection of a packet of ion cloud containing dif-
ferent ion species, all ions start to move through the drift area with distinct
velocities, because the velocity of individual ion species is proportional to
their coefcient mobility (Ki) and strength of electric eld (E):
J
ii
KE=*
(6.2)
When the ion gate is operated in periodic mode, the detector records IMS
mobility spectra, which include a series of ion pulses with different arrival
time td. The value of td is used for calculation of coefcient mobility, and con-
sequently for the identication of ion species. As can be seen, the principle
of operation of an IMS is similar to the operation of the time-of-ight mass
spectrometer (TOF MS), with the difference between IMS and MS occurring
only in pressure conditions in analytical areas. Because MS operates in a vac-
uum, the arrival time of analyzed ions characterizes only the mass-to-charge
ratio (m/z). Nevertheless, similar measurements in IMS operating at ambient
pressure provide additional ion structural information.
6.3.2 Phenomenological Description of DMS Sensor Operation
A schematic and the operational principles of differential mobility spectrom-
eter (DMS) are shown on the right panel of Figure 6.3. Construction of a DMS
sensor is relatively simple in comparison with IMS. It contains the follow-
ing three aligned segments: an ionization area, an analytical area between
two electrodes for ion separation, and a pair of two detector electrodes for
recording ions of each polarity (positive and negative). A formal description
of the DMS shows that it operates in a manner similar to a quadrupole mass
analyzer; ion separation occurs due to superimposed effect of dispersing RF
and compensation DC electric elds in the analytical gap. When sample mol-
ecules are continuously introduced with transport gas ow into ionization
area, the positive and negative resultant ions are formed and transported in
analytical gap between two electrodes, where they then can be separated by
adjusting specic combination of RF and DC electric elds. By regulation of
the value of compensation voltage, it is possible to straighten trajectories for
certain ion species. In this condition when values of Vc and RF are appropri-
ately adjusted, only selected ion species can pass through the analytical gap
and be continuously recorded on detector electrodes located in the exhaust
of the analytical gap. Other ion species will have tilted trajectories; therefore,
these ion species eventually reach one of the analytical electrodes (upper or
bottom) and are neutralized. Instead of a constant compensation voltage,
sweeping the voltage the output of DMS detectors can provide a sequence
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
101Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
of peaks, as shown on DMS spectrum in Figure 6.3. The shape of the elec-
trodes can be planar, as shown in the upper-left corner of Figure 6.2. This
design is usually associated with the term DMS. Another cylindrical shape
electrode design, shown in bottom-left corner of Figure 6.2, is associated with
the term FAIMS. In any case, ion movement through the gap under the effect
of gas stream along analytical gap and superimposed asymmetric waveform
RF and DC electric elds applied across the two electrodes, causes ions to
move in the analytical gap with a “zigzag” or up/down motion. Different
ion species trajectories have different slopes. When the asymmetric wave-
form electric eld is large enough (E> 1000 V/cm), the resulting trajectories
of the different ion species diverge. The ion’s alpha parameter characterizes
the dependence of the coefcient of mobility of individual ion species on the
strength of the RF electric eld, E, as shown in Equation 6.3:
KE
KE
ii i
()
=
()
+
()
é
ëù
û
01a (6.3)
where Ki(0) is the ion’s low eld coefcient of mobility. As ions travel in
zigzag trajectories, their effective trajectories depend on their specic alpha
parameters and can be shifted by either of the two electrodes. Ion species
whose net movement under gas ow, the superimposed DC and asym-
metric RF voltage adjusted to pass through the analytical gap, are sub-
sequently detected at one of two Faraday plates placed on the exhaust of
the analytical gap. Positive ions are detected at the bottom Faraday plate,
which has a negative DC bias, and negative ions are detected at the upper
detector plate, which has a positive bias. In order to “tune” the DMS sen-
sor to pass the desired targeted ion species, a variable DC potential, known
as the compensation voltage (Vc), is also applied across the two analyti-
cal electrodes. By varying either the RF or DC electric elds, it is possible
to reach conditions where only specic ion species pass through the gap
and are detected. Other ion species are neutralized on the top or bottom
analytical electrodes. As such, DMS can be considered as a continuously
operated ion species lter, when Vc is constant and tuned for ltering the
targeted ions. So, in contrast to IMS, which provides ion separation in a
pulsed regime according to the absolute value of ion coefcient mobility
Ki(0), the DMS separates ions continuously and provides ions ltration on
the basis of the ion’s alpha parameters. Equation 6.4, which is a derivative
of Equation 6.3, shows that the alpha parameter of individual chemicals
expresses the ratio of the differential mobility ΔK(E) = K(E) − K(0) to the
low eld ion mobility K(0):
aD
i
ii
i
i
i
EKE K
K
KE
K
()
=
()
-
()
()
=
()
()
0
00
(6.4)
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
102 Diagnostic Devices with Microuidics
The value of the alpha parameter, which characterizes the effect of a strong
electric eld on ion coefcient mobility Ki(E), is specic for individual ions,
and therefore can be used for the identication of an ion species. The sign of
αi can be positive, when ion coefcient mobility increases with increasing E,
or can be negative, when mobility decreases with increasing values of E.
6.3.3 Comparison between the IMS and DMS Spectra
Figure 6.4 presents IMS and DMS spectra for DMMP (124 Da) and lutidine (107
Da) samples that were obtained separately in IMS (a) and DMS (b). As shown,
the IMS spectra for both of these chemicals are fairly similar, and naturally
include a reactant ion peak (RIP) with drift time td = 4.2 ms and analyte peaks
in both spectra with same drift time td = 5.4 ms. In contrast, the DMS spectra
of DMMP+Lutidine mixture probe (on Figure 6.4b) contain a combination of
DMMP peaks (monomer Vc = −4 V and dimer Vc = +4.5 V) and a lutidine peak
at Vc = −1.5 V. A control experiment of a solely lutidine sample unambiguously
shows that the lutidine peak in fact has Vc = −1.5 V and is located between two
DMMP peaks. So, for this particular component, we can see that while DMMP
and lutidine are unresolved in IMS, they can be easily resolved in DMS. This
was an expected result, because as already discussed, the IMS and DMS meth-
ods exploit different properties of ion species: IMS operation is based on mea-
surement of the absolute value of coefcient mobilityKi; the DMS exploits the
alpha parameter (αi(E)), which characterizes the ability of changing of ions
conformation (cross section Ω) under the effect of strong electric eld.
6.3.4 Use of DMS for Detection of Chemical
Warfare Agents: GA, AC, and CK
Historically, ion mobility-based spectrometers were the most commonly
deployed systems for chemical monitoring by the military for quantitative
measurements and identications. In this section, results are presented to
1357911
Drift time (ms)
Intensity (a.u.)
DMMP
Lutidine
–30 –25 –20
(b)
(a)
–15 –10 –5 0510
Compensation voltage (V)
Lut
DMMP+Lut
RIP
RIP
(DMMP)2H
+
(DMMP)H+
Lut
FIGURE 6.4
Comparison between IMS (a) and DMS (b) spectra for lutidine and DMMP samples.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
103Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
show the analytical potential and utility of DMS technology for homeland
security applications. In Figure 6.5, in the left column, experimental positive
ion spectra are presented for Tabun (GA—nerve class chemical agent with
MW = 162 Da), and in the right column are two spectra of blood class agents:
(d) CK—agent, which is cyanogen chloride (CNCl), MW = 61 Da; and (c)
AC—agent, which is a very small hydrogen cyanide (HCN) (MW = 27 Da).
GA spectra presented for two RF voltages: panel (a) for 1500 V and panel (b)
for 1000 V. The transport gas was pure air with addition of 0.1 mg/m3 of GA
vapors. GA spectra contain RIP and two analyte peaks, which were identi-
ed as monomer and cluster peaks. Comparison of both spectra shows that
for this class of chemicals, increasing the RF voltage increases the separa-
tion power of a planar DMS sensor. Positions of each peak for a certain RF
value are used for identication of the original ion species. The estimated
LOD level for GA is ~0.007 mg/m3, which corresponds to a (V/V) concen-
tration of 10 ppb.
Both AC and CK agent spectra are presented in Figure 6.5c and d, as obtained
for same concentration of analyte vapors (~22 mg/m3). Each spectrum for these
extremely volatile components contains two peaks: the leftmost peaks cor-
respond to analyte ions that are detected at different compensation voltages.
The rightmost peaks in each spectra are related to background (RIP) ions, and
they naturally have the same peak position, Vc = −11 V in both measurements
0.04 0.130
0.140
0.135
0.130
RIP 0.125
0.120
0.115
0.110
CK
0.125
0.120
0.115
–40 –30 –20 –10
01
0
–40 –30 –20 –10
01
0
0.08
0.06
0.04
0.02
0
–30
0.03
0.02
–20 –10–30
Response
(a)
(c)
(d)
(b)
Response
0.01
0
GA, 0.1 mg/m
3
GA, 0.1 mg/m
3
–20
Vrf =1500
RF = 650 V
RF = 650 V
Vrf =1000
Monomer
Monomer
Cluster
RIP
RIP
AC
Cluster
–10
0
0
10
10
V
c
V
c
V
c
(V)
V
c
(V)
FIGURE 6.5
In the left column, DMS positive ions’ linear spectra of GA agent are presented for two disper-
sion voltages: Panel (a) (1500 V) and (b) for (1000 V). In the right column—examples for negative
ion spectra for two chemicals: panel (c) for AC agent and panel (d) for CK.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
104 Diagnostic Devices with Microuidics
performed with RF = 650 V. These negative reactant peak ions that are
formed in air correspond to oxygen/water molecules cluster ions, (H2O)nO2−.
Comparison of the linear spectra for the two chemicals (in Figure 6.5c and d)
shows that DMS is well matched for the identication of small molecular com-
ponents, which is usually problematic for time-of-ight IMS. The estimated
limit of detection (LOD) for both these blood class agents is ~4.1 mg/m3.
6.3.5 Planar DMS Sensor Design Optimization
for a Particular Application
Well-developed mathematical descriptions of the principle of DMS operation
are available to explain the operation and predict DMS spectra for given pla-
nar designs. Calculation of peak position and width in DMS spectra involves
several steps; a comprehensive description of the calculation can be found in
a recent review article.28 In the following, a common approach is used to get
a nal formula for calculation peak position Vc, peak width FWHM, which
can be used for calculation resolution R = Vc/FWHM, and peak capacity (PC)
used for characterizing resolving power for spectra of multicomponent DMS
spectra PC = (Vc max−Vc min)/FWHM.
6.3.6 Peak Position in Planar DMS Spectra
A mathematical formula for calculating a peak position in a DMS spectra
was suggested in the rst publication regarding DMS/FAIMS technology.2
According to the DMS principle of operation, ion separation occurs when
ions move under the superimposed effect of alternating strong RF (disper-
sion) and weak DC (compensation) electric elds acting in a direction perpen-
dicular to the transport gas stream that transports ions through the analytical
gap. The velocity of all ion species through a cell is determined solely by the
linear velocity of the transport gas. Average dwelling time (τ) through a cell
depends on geometric sizes (height, width, length) of the analytical gap and
ow consumption gas rate Q(cc/min). Usually, an ion’s resident time in the
analytical area τ = Q/(h ∗ w ∗ l) ranges from a few milliseconds to several hun-
dred milliseconds, depending upon design geometry.
During ight time, ions experience the effect of an asymmetric waveform
(alternating between low and very strong electric elds) encountered dur-
ing each period of the RF waveform, roughly at least 10,000 times moving
up and down perpendicular to the gas ow along to gap axis. Ion motion
under effect of the electric eld in gases can be described by simple equation
J= *
KE
. In the case of superimposed very strong oscillated RF and weak
DC elds, the equation is modied as follows:
J^
()
=
()
´
()
=
()
´
()
tKEE tKE
Vt
h
eff
eff
(6.5)
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
105Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
where
Veff(t) = SV(t) − Vc is the effect of superimposed alternating separating SV(t)
and constant Vc voltages
h is the distance between two DMS electrodes
Therefore, the effective electric field in the analytical gap is
E
tV
th
effeff
()
=
()
/.
To calculate the resulting drifting velocity of ions in the perpendicular
direction 〈ϑ⊥〉, the integration of Equation 6.5 within one period (T) of RF
voltage is needed:
JJ J
^^
=
()
=
()
ò
tT
td
t
T
1
0
(6.6)
For separating particular ions, the values of SV and Vc should be tuned
to specic combinations, when the average velocity of targeted ions in
the transfer direction should be zero〈ϑ⊥〉 = 0. In this case, the resulting ion
zigzag movement along the analytical axis will have average zero tilt. For
example, such trajectory is depicted as a medium ions’ zigzag trajectory in
the right column of Figure 6.3. This condition corresponds to the situation
where an ion’s shift under the effect of strong-pulsed asymmetric wave-
form SV(t) electric eld is compensated due to the continuous effect of the
weak DC compensation voltage, and, therefore, ions move along the axis of
the analytical gap.
Using this approach, Equation 6.7 may be dened and used for predict-
ing the peak position for specic ion species (with given alpha functions)
in DMS spectra:
V
Ef
th
E
c
s
»
()
+
()
a
a1
(6.7)
where
f(t) is function describing the waveform of an alternating electric eld
Es is maximum magnitude of the alternating electric eld
Equation 6.7 shows that ions with a high value of alpha have higher values of
compensation voltage. The peak position of the DMS spectrum is increased
by increasing the resulting product of the following three parameters
[α ∗ Es ∗ f(t)]: alpha parameter ∗ strength of electric eld ∗RF pulse waveform.
According to Equation 6.7, the value of the compensation voltage directly
reects the magnitude of the differential mobility of ion species connected
with the alpha parameter.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
106 Diagnostic Devices with Microuidics
6.3.7 Peak Width in Planar DMS Spectra
The mathematical procedure for calculating peak width is presented in a
review28 and shown in the following:
FWHM hQ
wlKK
h
effeff
»*
** =12
tmin
(6.8)
where
KT
KE
tdt
eff
^=
()
()
ò
1
is the average (effective)value of the coefcient
of mobility for a given ion species and RF waveform.
This equation predicts the peak width for arbitrary ion species in DMS
spectra, if the ion’s alpha function is known. In the process of the mathemati-
cal calculation, the planar design of the analytical gap was used. Calculated
values of FWHM (full width half maximum) were then compared with exper-
imentally measured values of FWHM in different sized DMS sensors (with
different length, width, height). For conrmation, additional measurements
of the FWHM for the same ion species were obtained in experiments with
the same sensor design but with a variation of the gas consumption rates [Q].
Comparison between the experimental and calculated FWHM shows close
agreement, within ~10% error.
Equation 6.8 expresses four fundamental properties of planar DMS
sensors:
1. First, the ion residence time in the analytical gap is a universal param-
eter that provides comprehensive information about DMS sensor
performance. Peak width inversely depends (hyperbolically:Y = A/τ)
upon ion residence time in the analytical gap of sensor.
2. Second, the peak width depends directly on the volumetric ow rate
of transport gas and height of analytical gap.
3. Third, increasing the length and width of the gap helps to narrow
peak width.
4. Finally, the effective coefcient of mobility for a given ion species
can also affect observed peak widths: ions with higher effective coef-
cient of mobility have narrower FWHM.
6.3.8 Optimization of the DMS Design: Gap Dimension Effect
As shown at the beginning of this section, the performance of the planar
DMS sensor depends upon: (1) design and size of the analytical gap, (2)
alpha parameters α(E) of tested ion species, and (3) the strength and wave-
form of applied separating asymmetric waveform electric eld Es ∗ f(t).
Variation of these parameters affects DMS sensor output, which can be
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
107Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
manifested as changing of peak position, peak width, and peak intensity.
For validation and demonstration of a relationship between DMS outputs
and experimental parameters, a systematic investigation was conducted to
explore the effect of each dimension of analytical gap on DMS performance.
Protocol for the experimental investigation included the measurements of
the same samples in 12 version of planar DMS sensors with the same gap
height h = 0.5 mm but different gap widths w : 0.5 , 1 , 2 , 3 mm and lengths
l : 5 , 10 , 15 mm. Transport gas was regulated for ve ow rates (Q): 100,
200, 300, 400, 500 cc/min. The general conclusion from the obtained set of
data in these experiments was that the ion residence time in the analytical gap
is a universal parameter that provides comprehensive information about its perfor-
mance. Equations 6.7 and 6.8, obtained on the bases of mathematical calcula-
tion, also support this fundamental conclusion. Equation 6.7 indicates that
peak positions of ions in a planar design (with given gap height, h) depend
only on the strength of electric elds and ion alpha parameters and do not
contain Q.
To prove this important fundamental statement, results of direct exper-
iments to measure the effect of ion residence times on DMS output sig-
nals were investigated. Figure 6.6 shows four experimental DMS spectra
obtained for a 16-compound mixture acquired in same sensor with
dimensions 1 × 10 × 30 mm, while applying the same separation RF eld
Normalized peak intensity
Ion residence time (ms)
Abundances
161284
4
9
14
19
24
29
34
39
44
49
Relative int. ×40
FWHMx10
Peak capacity
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0
1.0
0.8
0.6
0.4
0.2
0.0 3020100–10–20
1.0
0.8
0.6
0.4
0.2
0.0 3020100–10–20
Compensation voltage
30
16.3 ms
10.3 ms
6.5 ms
5 ms
20100–10–20
3020100–10–20
FIGURE 6.6
Demonstration of the effect of varying residence time on DMS output.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
108 Diagnostic Devices with Microuidics
132 Townsends (Td). The only difference was ow rate, which provided
the possibility of achieving different ion residence time in analytical gap
ranging from 5 to 16.3 ms. A set of these experimental spectra are presented
on the left panel.29 In the right panel, the effect of changing residence time
on peak width, peak intensity, and peak capacity is shown. Experiments
explored 16 chemicals with masses ranging from 90.1 up to 649.1 Da: Alanine,
Serine, Proline, Valine, Melamine, Methionine, Histidine, MDA, Minoxidil,
Desipramine, Nordiazepam, Penbutolol, Propafenone, Angiotensin I (2+),
Angiotensin II (2+), Angiotensin I (3+). For convenience, to compare the
peak width and total spread of peak spectra, data are presented in normal-
ized intensity view.
Following Equation 6.7 (which does not include term Q), the position
of peaks Vc should be independent of changing ow rate (Q). According
to Equation 6.8, reducing the residence time of ions in analytical gap (or
increasing the ow rate) should strongly increase peak width, which
clearly can be seen on experimental spectra. Calculated from experimental
spectra, the effect of residence time on peak capacity (PC), FWHM, and
intensity of peaks are presented in the right panel and show an expected
result; increasing residence time of ions leads to increasing peak capacity
of sensor. However, decreasing the residence time increases intensity and
width of all peaks.
In summary, planar DMS sensor design optimization depends upon the
particular application.
• The ion residence time in the analytical gap is a universal parameter
that provides comprehensive information about the performance of
the planar DMS sensor.
• The general relationship between sensor residence time and peak
height shows that longer residence time reduces intensity but
enhances selectivity, by narrowing peak width. Therefore, there is
always a trade-off between peak height and peak width.
• Modifying analytical gap dimensions allows regulation of a DMS
sensor performance to be optimized for specic applications.
Examples of optimization toward this end include minimizing the
ow consumption, enhancing selectivity, or optimizing system for
maximum sensitivity, or minimizing the power consumption.
The general principles for the effect of sensor sizes on DMS sensor perfor-
mance are as follows:
• Increasing the analytical gap length improves resolution of the DMS
sensor due to increasing residence time, but also decreases the coef-
cient of transmission ions through the sensor.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
109Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
• Increasing the analytical channel width reduces peak width but also
requires higher-consumption gas ow rate to return to the same
peak intensity.
• Decreasing the analytical gap height reduces gas consumption but
requires a higher-frequency RF generator, and overall sensor resolu-
tion is also decreased.
6.4 Categorization of DMS Sensor Operation
in Integrated Systems
6.4.1 Motivation for Integration of DMS with
Other Analytical Instruments
The examples of operation of the DMS sensor presented earlier demonstrate
the analytical potential of DMS system for detection and identication of
individual chemicals. However, in applications where there is a need for fast
screening of complex samples in the eld, the resolving power of any stand-
alone ionization-type instrument (including the powerful atmospheric pres-
sure ionization mass spectrometric systems [API-MS]) becomes insufcient.
This is primarily due to the presence of a high number of possible chemical
interferents in environmental probes, which leads to additional complicat-
ing ions formation processes and, consequently, to changing of the sensor
response. For example, the presence of traces of impurities in the ionization
chamber or in the analytical channel can stimulate formation of (1) new ion-
molecular complexes due to elastic interactions of targeted ion with impu-
rity molecules, (2) initiation of fragmentation, and (3) neutralization due to
charge exchange in gas-phase reactions with impurity molecules. Therefore,
depending upon the nature of the targeted ion species and analysis condi-
tions, the interaction of targeted ions with other gas-phase species can lead
to undesirable consequences, resulting in the quenching or even loss of the
response from the ion of interest or masking of the targeted ions response
under enhanced intensity of chemical noise. One effective way to improve
the quality of analysis of complex mixtures is the coupling of ionization-type
instruments with other analytical techniques.
In analytical sciences, it is well known that integration of two or more
chemical characterization methods can provide enhanced performance of
chemical analysis. In integrated DMS systems, the analytical performance
improvement is typically the result of the following factors30,31: (1) the ana-
lytical space of the combined system is increased due to the complementary
chemical and physical characterization of the ion species, and (2) in hybrid
systems, the total analytical burden can be distributed between the integrated
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
110 Diagnostic Devices with Microuidics
methods, which helps to reduce the required level of performance of each of
the integrated methods.
Hyphenated or hybrid design approaches provide additional exibility
to improve manufacturability and the cost of the total system. Due to the
enhanced performance of the integrated system, design constraints and
performance of the individual component subsystems can often be relaxed
without a sacrice in the performance of the overall system. In other words,
integration of pre-separation and detection methods can help to enhance the
ability of atmospheric pressure ionization-type analyzers, and make possible
their use in more complex and harsh environments.
6.4.2 List of Products Where DMS Serves as System Engine
Once demonstration of the analytical potential of planar DMS systems
was established, additional efforts started to commercialize DMS/FAIMS
methodology. Currently, there are a number of DMS devices serving as
highly sensitive analytical tools, which use the DMS sensor as an analyti-
cal system’s engine, as presented in Figure 6.7. A goal of this chapter is to
provide information about instrumentation and application developments
for the planar geometry DMS system. In the following sections, we will
focus only on systems that operate on the basis of planar DMS sensors with
microscale dimensions. As was discussed in Table 6.1, the microchannel
DMS design has a number advantages and greater potential to improve
microAnalyzer
GC-DMS
NASA Air quality
monitoring the ISS
Breath analysis
AB SCIEX
SelexION
TM
Technolo gy
ThermoFisher
EGISTM defender
(explosives and narcotics)
Varian: CP4900 MicroGC
(mercaptans, sulfur, etc.)
Chemring
JUNOTM
(C
WA, TICs/TIMs)
DMS sensor
and electronics
First draper DMS sensor
DMS products
FIGURE 6.7
Illustration of developed systems around the DMS sensor with a microchannel analytical gap.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
111Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
analytical performance of mobility-based instruments critically needed for
biomedical applications.
Formally, all these systems can be categorized into three groups according
to the function that the DMS sensor performs in each presented system:
1. Simple gas analyzers, where DMS operates as a single DMS spectral
detector for analysis of simple samples.
2. Hyphenated systems, where the DMS used as spectral detector after
any pre-fractioning techniques, for example, chromatography,
temperature programming samples evaporating (or fractionat-
ing) systems, selective pre-concentrators with follow-on ash
thermo-desorption.
3. Tandem systems with a DMS interface mounted in front of high-speed
operated instruments (e.g., mass spectrometer or IMS). In this
geometry, the DMS serves for selection and directs only targeted
groups of ions in the analyzer. This approach has a special term:
“Plasma chromatography.”32 In regular GC chromatography, the
fractionation process of gaseous samples is time-consuming (usu-
ally 1–10 min), because molecule separation occurs due to multiple
adsorption–desorption events on the internal surfaces of GC col-
umns, and depends on the adsorption properties of analyte mol-
ecules. In plasma chromatography, analyte ion separation occurs
in gas-phase media, where ion interaction with bulk gas molecules
occurs with very high collision frequency (>109 s−1); therefore, DMS
version of plasma chromatography is capable of executing ultrafast
separation of gaseous mixtures.
6.4.3 Operation of a Single DMS as a Spectral
Detector in a Gas Analyzer
These type of analyzers provide fast measurements (second(s)) and can be
used for analysis of simple gas mixtures. In this family of instruments, chem-
ical identication occurs on the basis of peak position in DMS linear spectra
(DMS response intensity = f(Vc) obtained with xed RF voltage). To achieve
more reliable chemical identication, another multispectral method of
chemical identication was suggested.33 This advanced method for chemical
identication offered simultaneously synchronized scans for strong asym-
metric waveform dispersion voltage (SV) with weak DC voltage (Vc), and
simultaneously recorded series of DMS spectra for different values of dis-
persion voltages. Integrated software helps to generate 2D dispersion plots
from these data, which more thoroughly illustrates the process of shifting ion
peak position as a function of separation voltage (SV). This method provides
a panoramic view of the behavior of all ion species and reveals the specic
nonlinear behavior for each individual ion species, facilitating enhanced
chemical identication. In Figure 6.8, 2D traces for methyl salicylate (MSal)
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
112 Diagnostic Devices with Microuidics
ions are shown. To generate this particular 2D image, the Vc voltage is con-
tinuously scanned every second (between +10 up to −30 V), and the RF volt-
age is synchronized with Vc and scanned from 500 up to 1500 V with a step
of 10 V. Analysis of these 2D traces tells us that this chemical is able to form
simultaneously positive (left panel) and negative (right panel) ion species.
Corresponding linear spectra (shown in the following) contain two peaks
that correspond to RIP (background) and analyte (MSal) peaks. RIP ion traces
show signicantly stronger dependence from separation voltage, in contrast
to MSl ions. The presented example shows that in dispersion plots, differ-
ent behavior for individual ion species can be used as a source of additional
information for ion species identication.
Figure 6.7 shows the Juno™ handheld analyzer, developed by Chemring
Detection Systems (CDS) (2004). This device weighs ~2 lb, and has dimen-
sions: 7.64 L × 3.95 W × 2.2 D (in.). It was designed for detection and identi-
cation of the traces of TIC and chemical warfare agents (CWA): VX, GA, GB,
GD, GF, HD, L, HN3, AC, and CK in the eld and in battleeld conditions, at
levels less than IDLH (immediately dangerous to life or health).
Negative ions
RIP–
MS–
600
800
1000
Positive ions
RIP+
MS+
RF (V)
1200
1400
0.20
0.16
0.12
–30 –30
OH
O
O
–20
RF = 850 V
–20–10 –10
00
10
10
Vc
Vc
FIGURE 6.8
Dispersion plots and linear DMS spectra for both polarity methyl salicylate ions. In this par-
ticular case, placing the horizontal cursor in the position RF = 850 V therefore presented linear
spectra correspond to 850 V.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
113Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
6.4.4 DMS Operation as a Spectral Detector for Fractionating Systems
In particular geometries, the DMS sensor can operate with separating techniques.
A typical example for such a combination is use of the DMS sensor as a chro-
matographic spectral detector. This geometry provides two advantages:
pretreatment (fractionation) of complex sample probes, and simultaneous
chromatographic characterization of each fraction of a mixture. Therefore,
each fraction of a mixture is ionized and characterized independently in
DMS. Such two-step characterization of chemicals of interest provides 2D
characterization of analyte ions: the retention time of original molecules in
the GC column and the ion’s alpha parameter (which is connected with the
position of the analyte peak in DMS spectrum). Such 2D characterization
of individual chemicals enables more condent chemical identication and
improves the accuracy of quantitative measurements. The following systems
from Figure 6.6 belong to this category of instruments:
• The Varian CP-4900 analyzer was developed in 2004 for sensitive and
precise monitoring of sulfur components (odorants) in gas samples.
• The ThermoFisher™ EGIS Defender was designed to operate with
fast GC pre-separation and provide rapid (<20s) screening of pas-
sengers’ bags for explosives and drugs in airports.
• The Draper Lab/Sionex microAnalyzer (2007) was developed for
continuous, sensitive, and precise monitoring of the traces of indi-
vidual volatile organic compounds (VOCs) and was targeted for
many applications: breath samples analysis, monitoring environ-
mental samples, petrochemical and pharmaceutical industry, and
so on.
• Sionex Corporation built a version 1.0 of the microAnalyzer (2009)
for NASA as a continuous air quality monitoring (AQM) and ana-
lyzer for the International Space Station. A version 2.0 of the micro-
Analyzer was built by Draper with deployment occurring in 2013.
These systems have been successfully operating in the International
Space Station since 2009.34,35
6.4.5 DMS Operation as an Ambient Pressure Ion Pre-
Filter for Sophisticated Analytical Instruments
In this application, the DMS sensor operates as a tunable ion pre-lter for
selection of the targeted ion species and exclusion of non-targeted ion spe-
cies prior to mass spectrometry. Ion triaging prior to MS analysis helps to
decrease the complexity of the ion population introduced into the mass ana-
lyzer and leads to improving the quality of MS measurements, resulting in:
enhanced selectivity in MS measurements,36–38 a reduction in the chemical
noise in measured mass spectra,39 extending the linear dynamic range in
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
114 Diagnostic Devices with Microuidics
MS measurements,40 and at least a 10-fold improvement in the limit of MS
detection. Examples of commercial planar DMS ion lters with ambient mass
spectrometric systems are described in the following:
• AB SCIEX adopted the DMS technology and commercialized it in
2011 under the name AB Sciex SelexION™ Technology.41 The sys-
tem has been successfully used for challenging mass spectrometric
applications related to monitoring and quantitative measurements
of the chemical of interest in complex matrices. The technology
was shown to be particularly useful for separating isobaric and
isomeric ion species analysis in the pharmaceutical, food safety,
environmental testing, proteomics, metabolomics, and other appli-
cations areas.
• Other groups have also combined commercial stand-alone DMS
systems with potentially eldable MS systems: Mini1042 and NSTec
Experimental.43 These proof-of-concept experiments showed that
DMS pre-ltering in front of compact/mobile ambient mass spec-
trometers provides signicant value. It is typically the case that in
the process of miniaturization the MS performance is reduced, and
the analytical power of stand-alone MS systems can become insuf-
cient to resolve quasi-isobaric components, such as those com-
monly encountered in background spectra or when mobile API
MS devices operate in harsh environmental conditions. In several
publications42–44 it is shown that in applications when detection
and quantitative measurement in eld conditions are needed, even
low-resolution portable mass spectrometers with DMS pre- ltering
can provide a quality of analysis similar to stand-alone, high-
performance desktop-type mass spectrometers. Therefore, adding
DMS pre-ltering aids in restoring the analytical capability of com-
pact mass spectrometers, which is lost during miniaturization. In
addition to analytical advantages, long-term test results show that
DMS pre-ltering improves the robustness of any API MS. Testing
DMS-MS systems in harsh environment conditions28 showed that
the presence of a DMS interface in front of an API MS substantially
increases the duration of service time before cleaning is required.
The foregoing presented a brief review of the chronology of a list of differ-
ent prototypes, built in the last decade, and shows that DMS technology has
signicant analytical potential for detection and identication of traces of
targeted chemicals in complex samples. One unquestioned advantage of this
technology is its ability to operate with other analytical systems, when it is
used for triage of ion species before analysis. This depends upon the par-
ticular application, condition of operation, and requirements that the DMS
system can be optimized in various ways.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
115Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
6.5 Advantages of Tandem Systems for Specific Applications
6.5.1 The GC-DMS System: Protecting Crew Health and
Safety Aboard the International Space Station (ISS)
Crewmembers remain in the semiclosed environment of the International
Space Station (ISS) for as long as 6 months, and during this time they rely on
the ISS revitalization systems to keep the air and water safe for human con-
sumption. It is vital to the health and safety of the crew that the air and water
be periodically monitored and assessed.
The gas chromatograph-mass spectrometry (GC-MS) is the standard
ground laboratory instrument for analysis of air contaminants. Unfortunately,
it is too large, power hungry, and maintenance-intensive for use on the ISS.
Replacing the MS with a DMS dramatically changes the resource issues of
the GC-MS, thus allowing a small, low-power instrument that is relatively
maintenance-free to monitor the contaminants onboard the ISS. The size and
appearance of this system is presented in Figure 6.9, which was taken during
a TV interview with an astronaut during a mission on the ISS. The reduc-
tion in size and maintenance resources is due to the fact that the DMS oper-
ates at atmospheric pressure as opposed to vacuum conditions required for
MS. This was the motivation to develop a new air quality monitor for NASA
Draper DMS
microAnalyzer V2.0
FIGURE 6.9
Air Quality Monitor (AQM) aboard the International Space Station (ISS).
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
116 Diagnostic Devices with Microuidics
needs on the basis of Sionex’s existing GC-DMS system. Currently, the air
quality onboard ISS is monitored every 73 h by the Draper microAnalyzer
(referred to as the Air Quality Monitor—AQM), and the data are downlinked
weekly for assessment by the NASA toxicologist. Numerous compounds
have been detected on the ISS over its 15-year life, but only a handful of com-
pounds meet the criteria to be monitored by real-time instrumentation. This
list of compounds (Table 6.1) was developed by applying these simple crite-
ria: (1) compounds frequently detected in spacecraft at concentrations above
trace levels (e.g., ethanol, acetone), (2) compounds with signicant toxicity at
low concentrations, even though they are detected infrequently on spacecraft
(e.g., benzene), and (3) compounds that can affect the response of continu-
ously (autonomic) operating the environmental control systems (e.g., silox-
anes, 2-propanol). The AQM monitors the list of compounds in Table 6.2, but
more compounds could be targeted, if necessary, and the instrument is capa-
ble of detecting nontarget compounds as well. The NASA toxicologist sums
the health effects for each compound, based upon the concentrations derived
from the AQM and uses this number (total T-value) to determine the accept-
ability of the air quality on the ISS during missions. In nominal conditions,
the main contributor to the T-value, not detected by AQM, is carbon dioxide.
A health effects graph produced from AQM data is shown in Figure 6.10.
Additional value in having the technology present onboard the AQM
was demonstrated in 2015, when an independently operated environmen-
tal control sensor system suddenly alarmed indicating a massive ammonia
leak into the ISS. The crew was evacuated to a safe haven, sealed from the
LAB module. It was thought that the alarm was false, but the options to con-
rm a false alarm were either to send the crew into the potentially contami-
nated LAB module to take samples or remotely activate the AQM from the
ground and review the recorded data. The AQM was remotely activated, a
method for detecting ammonia was uploaded from ISS ground control, and
TABLE 6.2
List of Targeted Components for AQM
Target Compounds
2-Butanone Hexanal
2-Propanol Hexane
Acetaldehyde Hexamethylcyclotrisiloxane
Acetone Methanol
Acrolein m-p Xylenes
Benzene n_Butanol
Dichloroethane Octamethylcyclotetrasiloxane
Decamethylcyclopentasiloxane Toluene
Ethanol Trimethylsilanol
Ethyl acetate Ammonia
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
117Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
the microAnalyzer conrmed that there was the normal amount of ammonia
in the LAB atmosphere. This avoided the need to potentially put the crew in
harm’s way, which would have been required if they had to break the seal on
the safe haven to enter the LAB.
A second important operational use of the AQM in 2015 was to help trou-
bleshoot unusually high conductivity values in the water processing unit.
Comparison of AQM response on headspace vapors with water processor
data showed that higher than normal levels of ethanol were probably the
cause of the high conductivity in the water processing unit.
6.5.2 DMS Operation as a GC Detector for
Measurements in Harsh Environments
To demonstrate the enhanced analytical power of a GC-DMS system, experi-
mental data related to the detection of trace amounts of CWA simulants in
clean samples and in samples containing potential interferents are shown.
Figure 6.11 shows the response of a microAnalyzer (GC-DMS) for an air
mixture that contains traces of four phosphonates, which commonly serve
1.1
1.0
0.9
0.2
0.1
T value by health effects
0.0
October 1–30
Year-to-date
Gastrointestinal toxicity
Hepatotoxicant
Total
Headache
Skin flushing
Visual disturbances
Ototoxicant
Immunotoxicant
Respiratory system injury
Reproductive toxicant
CNS depression
Carcinogen
Mucosal imitant
FIGURE 6.10
A health effects graph produced from AQM data.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
118 Diagnostic Devices with Microuidics
as simulants for nerve agents. Presented in the top panel (a) is a 2D chro-
matogram illustrating the DMS scanning response versus GC retention
time for a clean sample. It contains four well-resolved spots related to each
of the four simulants (A, B, C, D). Each spot is characterized as a specic
combination of retention time (GC separation) and compensation voltage
(DMS separation). This 2D chemical characterization in the GC-DMS system
(b
)
(a
)
AB
CD
150
Time (S) (includes preconcentration time)
AC
D
B
With DMS peak
selection
(d)
100500
0.50
0.55
0.60
0.65
0.70
0.75
0.80
(c)
Without DMS peak
selection
50 100 1500
1.2
1.4
1.6
FIGURE 6.11
GC chromatograms obtained in a microAnalyzer (GC-DMS) for gas mixtures of four organo-
phosphate chemicals commonly used as simulants of nerve agents. (a) 2D chromatogram for
mixture sample with four chemicals (A,B,C,D). (b) 2D chromatogram for the same sample
with adding interferents (AFFF and diesel). (c) Regular chromatograms extracted from raw 2D
experimental data (experimental data (b)). (d) Extracted chromatogram from experimental data
(b) by using DMS peak selection. (From Nazarov, E.G. et al., Planar differential mobility spec-
trometry as a powerful tool for gas phase ion separation and detection (Spectroscopy Solution:
Premier Learning for Analytical Chemists.))
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
119Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
enhances the identication power of GC measurements in comparison with
any regular total current measured detectors such as ame ionization, ther-
mal conduction, and electron capture. The estimated values of LODs for
each of the four chemicals were 65 ppt for A, 424 ppt for B, 20 ppt for C, and
107 ppt for D.
To model the challenging environmental conditions, additional vapors of
interferents were added to the same analyte mixture. The interferents used
were diesel fuel vapors and vapors of aqueous lm forming foam (AFFF),
a shipboard re-ghting material known to cause interferences with IMS-
based CWA sensor systems. A 2D GC-DMS chromatogram of this mixture
is presented in panel (b) of Figure 6.11. As evidenced from the chromato-
gram, the addition of the interfering materials substantially increased the
chemical noise, with the appearance of numerous new peaks that overlap
with the analyte peaks. As a result, detection and identication of the target
chemicals become problematic. A regular chromatographic view for total
ion current measurements is presented in panel (c). Due to the high chemi-
cal noise, the peaks for A and B components cannot be resolved from inter-
ferences, and only identiable peaks can be seen for C and D components.
The nal chromatogram presented in panel (d) was obtained using a regime
of DMS operation with peak selection, wherein the RF and Vc parameters
are optimized for the detection of each of the four individual components.
As a result, the components A and B are now resolved from the interfering
background signal and it is possible to detect, identify, and quantify all four
targeted chemicals.
6.5.3 Example of Chemical Noise Reduction in DMS-MS
Experimental spectra that illustrate the advantages of integrating DMS to MS
are discussed subsequently and shown in Figure 6.12a through c as reported
in several publications.38,39,46
In Figure 6.12a, two mass spectra are presented for monitoring 2′-deox-
ycytidine (100 pg/μL) in mouse urine samples. 2′-Deoxycytidine (MW =
228 g/mol) is a radiation exposure biomarker and can be used in biodo-
simetry to determine the level of radiation exposure of individual live
subjects.39 The top spectrum was obtained without DMS ion pre-ltering
(where RF voltage is turned off) and the bottom spectrum with DMS lter-
ing (when RF = 3200 V). The rst spectrum has signicant chemical noise,
which is typical for a complex sample such as urine. The spectrum contains
a small peak (with an S/N of around 3–5) related to the targeted biomarker,
2′-deoxycytidine, which is the protonated molecular ions peak MH+ with
m/z = 229 Da. DMS pre-ltration becomes very simple. The recorded mass
spectrum contains only one major peak with expected m/z of 229 Da. Most
importantly, due to suppression of chemical noise, the S/N is now at least
50, which shows that DMS pre-ltration improved the value of the LOD for
this assay by 10- to 15-fold.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
120 Diagnostic Devices with Microuidics
6.5.4 Separation of Isobaric Ion Species in DMS-MS
Figure 6.12b shows DMS spectra46 for ve isobaric components with close
molecular masses (MW~ 308), but different structures: Benoxinate, MW =
308.4158; Phenylbutazone, MW = 308.3743; Bestatine MW = 308.37; Warfarin,
MW = 308.3278, and Quinoxyfen, MW = 308.135. In mass spectrometric anal-
ysis with electrospray ionization, all of these chemicals form protonated ion
species (MH+) with close m/z (~309.25 ± 0.15 Da). Therefore, to resolve these
components, one would need a mass spectrometer with enhanced resolution.
At the same time, due to the fact that these chemicals have distinguishable
structures, and hence ion mobilities, DMS is able to resolve these structures
easily, as shown in Figure 6.12b. The additional parameter of DMS compen-
sation voltage allows one to identify these isobaric components, even with
FIGURE 6.12
Sample spectra obtained with a DMS pre-lter in front of an API MS: (a) reducing chemical
noise in mass spectra. (Continued)
450
450
400
400
350
Chemical noise reduction
350
m/z
(a)
Detection radiation of biomarkers in DMS-MS
2΄ -deoxycytidine, 100 pg/μL in urine samples
300
300
250
250
200
200
cytidine DM S-filtered MS (3250 VRF, CV –4.7 V. 100 pg/μL)
150100
0
1000
2000
3000
HO
HO
O
O
N
N
NH2
4000
5000
0
2
4
6
8
10
12
μ2H+
μH+
150
With DMS filtration
100
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
121Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
a low-resolution mass spectrometer (e.g., a eld portable MS), which has
resolving power of one mass unit, and possibly lower.
6.5.5 Selection and Identification of Isomeric Ion Species in DMS-MS
Figure 6.12c shows a DMS spectra obtained in a gas mixture that contains
two stereoisomeric components: ephedrine and pseudoephedrine.38 Both of
these chemicals have the same molecular formula, C10H5NO, identical molec-
ular weight, 165.11536, and similar molecular structures. The only difference
between the two chemicals is the spatial arrangement of –OH groups in their
FIGURE 6.12 (Continued )
Examples that provide DMS pre-lter in front of an API MS: (b) separation of isobaric compo-
nents, and (c) separation of isomeric components.
Separation of isomeric molecules with same
MW = 165.11536
(c)
1.0
0.8
0.6
0.4
0.2
0.0
–2 0 24681012
Compensation voltage
Normalized signal
Pseudoephedrine Ephedrine (C10H15NO)
CH3
CH3
CH3
H3CNH H
N
OH
OH
0
BenoxinateQuinoxyfen
PhenylbutazoneBestatin
Warfarin
–10–20
Normalized signal
CV (V)
–30
Separation of isobaric ions
All ions (MH)+ have similar mass, m/z=309 Da
–40–50
(b
)
0.0
0.2
0.4
0.6
0.8
1.0
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
122 Diagnostic Devices with Microuidics
molecular structures. Regardless of their similarities, these two chemicals
have very different physiological effects: ephedrine is used in traditional
medicine for the treatment of asthma and bronchitis, while pseudoephed-
rine is used for nasal congestion treatment. Therefore, it is important in the
process of synthesis of these compounds to be capable of online control and
identication of the desired nal product. Even expensive, high-performance
MS instruments cannot by themselves differentiate these compounds. But, as
shown in Figure 6.12c, peak positions of ephedrine and pseudoephedrine
ions are different and well resolved in DMS spectra. Therefore, by using a
combination of DMS pre-ltration with MS, differentiation of these is iso-
meric compounds becomes possible.
6.5.6 DMS Operation as Interface in Front of IMS: DMS-IMS2 System
Motivation for operating in the DMS-IMS2 mode can be inferred from
Figure 6.3, where a comparison between operating principles and out-
puts of both (IMS and DMS) ion-mobility-based sensors is given. Equation
6.3 shows that in general, the ion’s comprehensive coefcient mobility
K(E) = K(0) [1 + α(E)] is a function of electric eld strength and consists of two
terms: (1) K(0) , which is the absolute value of ion coefcient mobility at low
electric eld conditions, and (2) α(E), the alpha function that connects to the
effect of a strong electric eld on ion mobility (see Equation 6.4). A compari-
son between the operating principles of IMS and DMS reveals that these two
systems operate in different regimes. In IMS, ion separation occurs in a low
(E ~ 200 V/cm) DC electric eld, and values of K(0) are obtained by measure-
ment of ion arrival times. In DMS, ion spatial separation occurs in oscillating
(between 1 and 35 kV/cm) asymmetric waveform electric eld conditions.
The resulting trajectories depend upon differences between analyzed ion’s
coefcient of mobility, which are different in high and low voltage portions
of the separating RF voltage. Differences in methodologies of ion separation
and experimental conditions lead to differences in IMS and DMS outputs:
• Output of an IMS contains solely information about low eld coef-
cient mobility K(0).
• Output of a DMS includes information about an ion’s alpha coef-
cient α(E).
Therefore, integration of IMS and DMS spectra provides value in obtain-
ing different portions of comprehensive coefcient mobility K(E), which is
described in Equations 6.3 and 6.4. Therefore, an integrated DMS-IMS sys-
tem enables a 2D characterization of ion species, which contains enriched
mobility-related information. This consequently has an enhanced level of ion
species’ identication in comparison to a stand-alone mobility instrument.
This system was developed in collaboration with Hamilton Sundstrand
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
123Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
Corp. The objective was to develop a fast-operated tandem system, which
minimizes the false alarm rate for detection and identication traces of tar-
geted hazardous chemicals.
A schematic of T-tandem DMS-IMS2 instrument consists of one planar DMS
sensor and two independently operated cylindrical time-of-ight IMS(s)
that are coupled to the output of the DMS. A block diagram and design
of a tandem system are shown in Figure 6.13a. DMS itself can operate as a
I
2
U
2
Q
4
(a)
I
1
Q
1
Q
3
RF
“–”IMS
“+”IMS
U
1
DMS
Q
2
V
c
(b)
RIP
5.0
IMS drift time (ms)
Positive ions
Positive ions
4.54.03.53.02.52.0
–12
–10
–8
–6
–4
–2
0
2
4
Monomer
Cluster
DMS compensation voltage
(V
c
)
DMMP
(c)
Ko= 1.86 cm2s–1V–1
IMS pos. spectrum
DMS pos. spectrum
1.0
–20
–15
–10
–5
0
1.52.0 3.03.5 4.0
Toluene
RIP
H+(H2O)n
DMS compensation voltage (V)
IMS drift time (ms)
2.5
FIGURE 6.13
Tandem DMS-IMS2 system and examples of its output. (a) Schematic of T-tandem (DMS-IMS2)
system. (b) Example of 2D pattern (Vc vs. td) recorded in T-tandem system for detection traces
(45 ppb) of toluene vapors in nitrogen. (c) 2D pattern for positive dimethyl methyl phosphonate
(DMMP) ions.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
124 Diagnostic Devices with Microuidics
dual polarity DMS sensor, or as a lter for specic ion species, which after
selection in DMS, are directed into the IMS cells. Drift time is measured and
recorded on IMS detectors. This T-design conguration is capable of analyz-
ing positive and negative ions in parallel.
Figure 6.13b and c presents 2D plots that were obtained in regime of opera-
tion “Compensation voltage vs. IMS Drift Time.”In this regime, the DMS
compensation voltage is periodically (t ~ 1 s) sweeping. At the same time,
the IMS shutter runs continuously with a signicantly faster period, which
allows recording of IMS spectra every 0.01 s. Special software developed for
DMS-IMS2 system generates in real-time scale these 2D patterns (Vc vs. td) and
displays for both polarity ions patterns on screen. In such a representation,
any individual chemicals ions are revealed as specic spots in 2D surfaces
with a particular Vc and td. Therefore, identication of a particular chemical
can be realized by determining its response spot coordinates: td—drift time
and Vc—compensation voltage. If need be, the software is able to separately
display any IMS or DMS spectra too.
As expected, a 2D pattern shows the enhanced analytical space of a DMS-
IMS system in comparison with a single sensor output. For example, in
Figure6.13b, the linear IMS spectrum (horizontal position) shows a single
(with td = 2.5 ms) unresolved peak for RIP and toluene response, but an inte-
grated 2D pattern shows two very-well resolved spots with the same td =
2.5 ms, but different Vc. = 4 and 8 V. This example shows the power of tandem
measurements, which provide information about a comprehensive value of
mobility; adding the DMS portion of mobility-related information (or alpha
parameter) helps to solve the problem of chemical separation/identication.
There are also possibilities for the opposite case, when IMS data introduce
additional information in comparison to DMS.
Presented in Figure 6.13c is another example of a DMMP response in a
T-tandem system. A 2D pattern for positive ions shows three spots corre-
sponding to three ion species: RIP, monomer, and cluster peaks. In this case,
both methods are certainly able to resolve all three peaks separately. But
the 2D approach provides enhanced separation power in contrast to linear
spectra. This is an expected result, and in this case we can explain this phe-
nomenon on the basis of simple geometric considerations. In general, the
resolving power of any type of spectral measurement can be presented as
R = l/FWHMav, where l is distance between two spots and FWHMav is aver-
age width of a response spot (or peak). In 2D cases, the distance between
two spots can be determined as lab
=+
22
, where a and b are distances
between two peaks in linear spectra, which can be obtained from spectra
obtained in DMS and IMS. This means that the value of l is longer than each
of the segments of a or b. At the same time, the average value of FWHMav
when a spot is symmetric (e.g., circle in this case) can be considered the same.
Therefore, the resolving power R of a 2D system is in principle higher than
can be obtained in a one-dimensional system. It means that for a DMS-IMS2,
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
125Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
there is a higher separation performance and an ability to operate with more
complex samples.
So, based on the presented results and considerations, the following may
be claimed.
The key benets of the DMS-IMS2 are as follows:
• High sensitivity for most TICs and TIMs. For instance, the typical
sensitivity for CWAs and organophosphate compounds is in the low
ppb-ppt range and comparable with stand-alone gas analyzers. This
is achieved due to the high-efciency (up to ~80%) transmission ion
species between two sensors.
• High speed of operation. Overall speed for the detection of single
targeted components may be ~1 s.
• Bipolar ion spectra. Both positive and negative ions are separated
and measured simultaneously without instrumental switching.
• Enhanced separation power (peak capacity), small physical size, and
moderate power consumption allow it to be used in the eld or in
handheld congurations.
6.6 Biodefense Applications: Aerosolized
Pathogens as Potential Bioweapons
Many documented cases of intentional pathogen release have been recorded
in modern history. Examples of agents used in chemical and bioterror attacks
include Salmonella,47 Bacillus anthracis spores,48,49 and sarin nerve gas.50 The
biological toxin ricin was distributed in bioterror attacks on the U.S. Capitol.51
Threats may not be limited to intentional release of aerosolized pathogens,
with at least one conrmed accidental release to a community in the last
several decades. In 1979, weapons-grade B. anthracis was released from a
military facility in Sverdlovsk, former Soviet Union, resulting in the largest
outbreak of inhalational anthrax in the twentieth century.52
The capability of B. anthracis to form highly resistant spores makes it a
prime candidate for an easily released biological weapon. Anthrax spores are
among the category A pathogens listed by the Centers for Disease Control
and Prevention,53 and can be aerosolized to target human inhalation. Early
detection technologies are required for fast and reliable characterization of
threat agents and for quickly exposing a hoax attack, if necessary. There
remains a need for a robust multipurpose microanalyzer capable of both
specic and sensitive environmental pathogen detection. Over the past sev-
eral decades, scientists have adapted molecular biology techniques to detect
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
126 Diagnostic Devices with Microuidics
B. anthracis using DNA-based, antibody-based, and mass spectrometry anal-
ysis approaches. These tests vary greatly in sensitivity, response time, cost,
availability, and complexity of use. With the identication of species-specic
primers,54 rapid polymerase chain reaction (PCR) has identied specic
Bacillus species from both environmental55 and clinical samples.56 A novel
detection method uses DNA-aptamers conjugated to magnetic electroche-
miluminescent beads to bind and detect Sterne strain B. anthracis spores.57
Sequencing on microchips containing gel-immobilized oligonucleotides
has identied B. anthracis by single-nucleotide polymorphism (SNP) anal-
ysis,58 and several commercial PCR kits/platforms are available that differ
in sensitivity depending on sample type and preparation.59 The most recent
is based on rapid-cycle, real-time PCR, developed as a collaborative effort
and dubbed the “Mayo-Roche Rapid Anthrax Test,”60 which yields results in
~35 min, but may be difcult to deploy in the eld.
Antibody-based methods traditionally use uorescent-conjugated anti-
bodies to spore-coat proteins to detect low levels of Bacillus spores. Phillips
and Martin (1983) showed that it is possible to detect Bacillus spores with
specicity using uorescein-conjugated polyclonal antibodies directed
toward the spore coat. However, it was found that multiple anthrax sero-
types exist among B. anthracis strains, rendering specic detection with this
method difcult.61 More recently, monoclonal and polyclonal antibodies have
been produced against Bacillus epitopes; these distinguish moderately well
between B. anthracis and B. subtilis, but less effectively between B. anthracis
and B. cereus spores. Additionally, variability exists in the specicity of anti-
bodies between spore coat and vegetative cell epitopes.62 Nevertheless, sev-
eral novel antibody-based assays have been developed to identify Bacillus
species. The electrochemiluminescent immunoassay (ECLIA) is based on
a redox reaction between ruthenium (II)-trisbipyridyl Ru[(bpy)3]2+ labeled
antibody and the excess of tripropylamine, which generates photons.63
A magnetic particle uorogenic immunoassay (MPFIA) technique employs
antibody-coated magnetic beads as solid phase in suspension for bacterial
capture and concentration in a 96-well microplate format.16 Both the ECLIA
and MPFIA are fast, but still require almost double the time of rapid PCR-
based tests. Antibodies have also been immobilized onto silicon chips or
membranes for higher-throughput screening of environmental samples.
One signicant limitation of these methods involves the specicity of the
antibodies selected for use. However, uorescent-labeled phage antibodies
have recently been produced, and show promise as Bacillus species-specic
markers64; antibody-based methods for Bacillus detection may further
improve in the future.
Numerous large-scale benchtop chemical and analytical detectors have
been explored to rapidly identify Bacillus spores. Most gas chromatograph
(GC) detectors, such as the widely used ame ionization detector (FID),
produce a signal indicating the presence of a compound eluted from the col-
umn; however, this signal lacks the information required for unambiguous
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
127Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
compound identication. An expedient and simple method for identi-
cation of unknown analytes requires a detector to provide an orthogonal
set of information for each chromatographic peak. The mass spectrometer
(MS) is generally considered one of the most denitive detectors for com-
pound identication, as it generates a ngerprint pattern of fragment ions
for each GC eluent. Mass spectrometric information is often sufcient for
sample identication through comparison to compound libraries, and has
been used to identify species of bacteria.65–69 Bacterial cell extracts them-
selves have been shown to produce reproducible spectra comprised mainly
of phospholipids, glycolipids, and proteins.70 As such, this is a very sen-
sitive method for identifying Bacillus species, and even unique biomark-
ers have been identied between closely related B. cereus strains.71 The
so-called tandem MS method has yielded a wealth of specically identi-
ed protein biomarkers for B. cereus using bioinformatic approaches.72
Analysis by matrix-assisted laser desorption/ionization mass spectrometry
(MALDI-MS) has also shown that very low mass biomarkers between 2 and
4 kDa distinguish B. anthracis from other closely related Bacillus species.73
While this result was obtained using a very specic carrier matrix, it dem-
onstrates that species-specic markers can exist if sample preparation is
optimized. However, minor variations in sample/matrix preparations for
MALDI-MS can produce signicant changes in observed spectra.74 Finally,
MALDI-MS has been shown to distinguish bacteria in aerosolized samples,
in a continuous fashion.75
While GCs are continuously being miniaturized and reduced in cost,76
mass spectrometers are still very expensive, and their size remains relatively
large, making them difcult to deploy in the eld. The DMS produces spec-
tra that can differentiate between compounds that co-elute in GC-MS, often
yielding an improved ability to identify samples. For MALDI-MS, a statisti-
cal model has demonstrated the ability to distinguish between roughly 10
species similar to B. subtilis when the spectral masses are grouped in 1.5 Da
ranges.77 This is due to roughly the same number of proteins per unit-mass
interval. Recent data also suggest a 75% correct identication rate using
MALDI-MS with no false positives.78 However, the DMS technology may
easily distinguish between even larger numbers of species, as the spectra
may be more easily deconvolved than those of MS due to differing ion
mobilities.
Pyrolysis can be used to convert the spore sample into its component sub-
stances through the use of heat. Pyrolysis can be employed for identication
and quantization of samples through the analysis of both the parent and the
product ions. Detection of bacterial spore biomarkers by pyrolysis has been
demonstrated using gas chromatography-ion mobility spectrometry and mass
spectrometry,49,50 and is a viable method of sample handling for DMS analysis.
Bacterial spore detection has been demonstrated utilizing a DMS system
with a front-end pyrolysis experimental setup consisting of a CDS Pyroprobe
1000 (CDS Analytical, Inc., Oxford, PA) connected to the inlet of an HP 5890
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
128 Diagnostic Devices with Microuidics
gas chromatograph (GC) (Agilent Technologies, Palo Alto, CA). The GC was
equipped with a 0.5 m deactivated fused silica column. A prototype DMS was
connected to the detector outlet of the GC. The interface temperature of the
pyrolyzer was set at 110°C, the GC inlet was set to 150°C, the GC oven was
held constant at 200°C, and the GC detector heating block was set to 150°C.
A slurry of Bacillus spores suspended in water was pyrolyzed by ramping the
temperature up to 650°C at a rate of 0.01°C/ms, and then holding this tem-
perature for 99.99 s. The spectra of the pyrolyzed spores corresponding to
the detected positive and negative ions were recorded on a laptop computer
connected to the DMS unit.
For each of the three species, B. subtilis, B. cereus, and B. thuringiensis, 100
experiments for each of three concentrations (900 experiments in total) were
conducted. The concentrations used were 2 × 107 spores/mL (80,000 spores/
experiment), 2.5 × 106 spores/mL (10,000 spores/experiment), and 1.25 ×
106 spores/mL (5,000 spores/experiment). The data were then analyzed by
ProteomeQuest® (Correlogic Systems Inc.), a proprietary pattern recognition
software package that combines genetic algorithm elements rst described
by Holland51 with cluster analysis elements described by Kohonen,52 as pre-
viously described.53–57
The data from each species were randomly divided into three categories:
a training set (50 spectra of each species), a testing set (150 spectra of each
species), and a validation set (~100 spectra of each species). The training and
testing sets consisted of les whose species’ identities were known by the
computer. The validation set, withheld from the modeling process, was then
scored by the model to give an independent measure of the accuracy of the
model on blinded data. The accuracy was calculated from the results of the
independent validation set using the following equation: Accuracy = (True
Positives + True Negatives)/(Total Number of Samples). The les were rst
compared in binary groups consisting of a single species at all three concen-
trations compared to a second species at all three concentrations, to create
models that differentiated one species from another. The models giving the
highest validation accuracies were as follows: B. subtilis versus B. thuringi-
ensis 98.5%; B. subtilis versus B. cereus 92.0%; B. cereus versus B. thuringiensis
69.2%. The latter two species proved slightly more difcult to distinguish,
which is not surprising as these two species are genetically very similar.
The biomarkers found across many models are displayed in Figure 6.14.
Panel (a) shows the biomarkers found in 40 models that allowed discrimi-
nation of B. subtilis and B. thuringiensis. Note that there is one biomarker
that was selected in many of the models, which indicates that it is important
in the discrimination of these two species. Panel (b) shows a similar plot for
B. subtilis and B. cereus, and again we see that the same biomarker appears
in many of these models as well. When comparing the models of B. cereus
and B. thuringiensis in panel (c), no biomarkers appear as frequently across
all models, which is consistent with these two species being difcult to
separate.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
129Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
To verify that B. cereus and B. thuringiensis tend to be more difcult to sepa-
rate from each other than from B. subtilis due to their similarity, several binary
models that distinguish B. subtilis from a pool of B. cereus and B. thuringiensis
les were created. Again the 5 k, 10 k, and 80 k les for each species were
combined and randomized prior to modeling. The model with the highest
validation accuracy was 92.3%. The high classication obtained here shows
that B. cereus and B. thuringiensis have biomarkers common to each other but
different from B. subtilis.
As B. cereus and B. thuringiensis are the most difcult to classify, we mod-
eled these two species at each concentration individually to determine if
there is a concentration limit below which the species become indistinguish-
able. The models offering the highest accuracy were 60.8% at 5 k concentra-
tion, 64% at 10 k concentration, and 88% at 80 k concentration. Therefore,
classication is more successful for these two closely related species when
more spores are present.
A set of three-way comparisons were also performed to classify all three
groups from one another in a single model. For these models only the 80 k
data were used, since we determined that below that concentration B. cereus
Model number
10
20
(a) (b)
(c)
30
00
10,000 20,000 30,000
Feature number
0
10
20
Model number
30
10,000 20,000 30,000
Feature number
0
10
20
Model number
30
10,000 20,000 30,000
Feature number
FIGURE 6.14
Distribution of features across 40 models. The models are: (a) B. subtilis versus B. thuringiensis,
(b) B. subtilis versus B. cereus, (c) B. cereus versus B. thuringiensis.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
130 Diagnostic Devices with Microuidics
and B. thuringiensis are more difcult to distinguish. For each species, the
spectra were randomly assigned to a training set of 25, a testing set of 50, and a
validation set of 25. An overall accuracy of 77.3% was obtained in one model.
Representative spectra from the three species at 5000-spore concentration
are shown in Figure 6.15. Biomarkers resulting from the three-way model are
–0.2
–0.35
–0.35
–0.5
–0.5
–0.65
–0.65
–0.2
–0.35
–0.5
–0.65
–40
Scan number Scan number Scan number
V
c
V
c
–0.2
–0.2
0.4
1.6
1
–0.2
0.4
1.6
1
–0.4
1
1.6
60
40
20
1
60
40
20
1
60
40
20
1
–30–20 –100 10
–40–30 –20–10 010
–40–30 –20–10 010
–40
60
40
20
1
60
40
20
1
60
40
(a)
(b
)
(c)
20
1
–30 –20 –10 010
–40 –30 –20 –10 010
–40 –30 –20 –10 010
–0.2
V
c
V
c
V
c
V
c
FIGURE 6.15
Representative DMS spectra of 80,000 spores undergoing pyrolysis at 650°C for 99.99 s.
Positive ion spectrum (left), negative ion spectrum (right). (a) B. subtilis, (b) B. cereus, and (c) B.
thuringiensis.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
131Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
indicated with circles. The raw data are shown here, but the biomarkers were
selected based on their relative ratio after normalization between zero and
one. The data from these experiments look very similar by eye, yet the pat-
tern recognition algorithms were able to nd biomarkers present in sufcient
quantities to reliably distinguish the species from one another.
A softer ionization technique, such as Matrix-Assisted Laser
Desorption/Ionization (MALDI), may also be used as a sample introduction
technique for DMS detection, as has been described earlier. In this method,
spore samples are pretreated, such as with coronal plasma discharge or other
chemical means, so that they will yield a high number of biomarkers. The
pretreated spores are then complexed with a chemical matrix, usually an acid
solution. The matrix–spore complex is excited by a laser using an energy
level sufcient to excite the matrix but not the pretreated spore itself. The
matrix and the spore then split apart, yielding electrostatic charged moieties
that are introduced in the mass spectrometer to yield spectral biomarkers.
Preliminary data taken with this front-end shows the ability to detect sample
(Figure 6.16). There is no effect on the background spectra (panel a) due to the
laser ring on a clean plate (panel b), which indicates that all signal detected
in panel (c) results from the sample complexed with the matrix material.
Both, pyrolysis and MALDI introduction techniques should yield many
useful biomarkers for spore samples, and it is possible that sample handling
protocols may maximize the number of unique biomarkers produced.
6.7 Medical Applications
6.7.1 Breath Analysis
There are dozens of volatile organic compounds present in exhaled human
breath, many of which show promise for diagnosis and management of dis-
eases, but with little available technical or clinical research and development
to date.79–81 Many volatile gases are produced by disease conditions, and can
often be smelled by physicians on the patient’s breath. Examples include
ketones in conditions such as starvation and ketoacidosis, feculent amines
generated following bowel obstruction, and bacterial by-products related to
anaerobic infections. Another example of breath analysis is the use of exhaled
breath samples by police departments to measure blood alcohol levels of
automobile drivers. Numerous diagnostic tests measure exhaled hydrogen
after ingestion of a specic sugar or starch load to signal the presence of
lactose deciency, bacterial overgrowth of the small bowel, malabsorption,
or decits in pancreatic function resulting from cystic brosis.82–84 One of the
most common breath tests involves an ofce-based diagnostic for H. pylori,
requiring the patient to rst eat 14C labeled urea, after which 14C labeled CO2
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
132 Diagnostic Devices with Microuidics
is detected in exhaled breath. There are other examples of radioactive labeled
metabolites used in additional gastroenterology tests. Exhaled nitric oxide
has been measured as a marker of inammation in the lungs.85–87 Analysis of
exhaled breath is being explored as a rapid toxicology test for carbon mon-
oxide and methanol.
6.7.2 Cardiovascular Applications
Cardiovascular applications of breath diagnostic testing are also emerging.
For instance, two volatile hydrocarbons, ethane and pentane, are produced
–0.20
–0.10
10
Compensation voltage (V)
Time (s)
0–10–20–30
10
Time (s)
0–10–20–30
0.20
0.10
–0.10
–0.20
0
0
20
40
60
(b)
(c)
–0.20
–0.10
0
0.10
0.20
Time (s)
0
20
(a)
40
60
10–10–20–30 0
0
20
40
60
0
0.10
0.20
FIGURE 6.16
AP-MALDI-DMS spectra. (a) Background signal, equipment on, but laser not ring. (b) Laser
ring on clean plate. (c) Laser ring on plate spotted with 100 fM desmosine in α-cyano-4-
hydroxycinnamic acid matrix.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
133Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
by the peroxidation of linoleic and linolenic acid, polyunsaturated fatty
acids found in cellular membranes. These chemicals are oxidized dur-
ing tissue ischemia and reperfusion injury.88–94 Breath pentane has been
found to be elevated in proportion to ischemia and inammation in heart
disease,95–98 in cigarette smoking,99 in ischemic bowel disease,100 in cirrho-
sis,101 in rheumatoid arthritis,102 and even in schizophrenia.103 Pentane may
serve as a marker for reperfusion injury and could be used at the bedside
to meter the rate of infusion of thrombolytic drugs or the percent of supple-
mental oxygen.
Phillips et al. demonstrated that the majority of 37 normal volunteers
exhaled the same level of pentane as ambient air, some more and some less,
with a normal distribution.104 Cailleux established that a normal level of pen-
tane is less than 10 pmol/L.105
Numerous literature reports have used gas chromatography or mass
spectrometry, or both. Past breath analysis studies have been hampered
by samples saturated with water vapor, variable ambient levels of gases
being measured, ambient pentane dissolved in body fat, and co-elution of
isoprene.106 When exhaled gas is used to measure a process in the lung, exclu-
sion of upper airway dead space gas may be necessary, but perhaps not when
we are looking at diseases in other organs. Other measurement challenges
have been related to earlier techniques of concentrating and detecting gases,
using GC and MS, but may not remain problems with PFAIMS.
6.7.3 Diagnosis and Monitoring of Tuberculosis
Tuberculosis, one of the world’s most prevalent killers, may be a diagnos-
tic target particularly well suited for noninvasive diagnosis using DMS or
IMS. Tuberculosis is caused by the obligate human pathogen, Mycobacterium
tuberculosis. This bacterium currently infects~1 in 4 people worldwide. Each
year, ~9 million cases of chronic TB infections become active, leading to both
increased transmission and high morbidity. About 26% of those with active
disease will die, making TB the eighth leading cause of death worldwide.
The WHO lists TB as the fourth leading cause of disability adjusted life years
(DALYs), a leading index for loss of per capita income and reduction in gross
national product. Worldwide incidence of TB continues to increase dramati-
cally, due in part to the rising prevalence of HIV/AIDS, which is associated
with a negative catastrophic synergy with TB.
Treatment of TB is extensive and requires a minimum of three medica-
tions, each of which has important considerations and limitations to use.
Inappropriate use of the medications has driven the emergence of multidrug-
resistant isolates (MDR-TB) that have an extraordinarily poor prognosis and
signicantly increases the cost of therapy. Diagnosis and identication of
chronic cases, which are then treated using a lower cost, more effective drug
regimen is the most critical step in the management of TB. Unfortunately,
denitive diagnosis requires laboratory culture of TB, a process that averages
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
134 Diagnostic Devices with Microuidics
15–19 days. For this reason, empiric use of anti-tuberculosis medications is
widespread, resulting in further selection of MDR-TB isolates.
DMS technology can contribute to the control of the epidemic and the treat-
ment of individual TB patients by capitalizing on the intersection between
the instrumentation technology and modern bioinformatics. The respiratory
form of TB produces a range of volatile organic compounds that are believed
to be liberated into the exhaled breath of those with both chronic and active
infections. Until very recently, ion separation technologies used on biomark-
ers lacked the sophistication to deconvolve these volatile organic signatures
in a reproducible manner. Emerging proteomic algorithms coupled with
DMS provide a method with which the delity and clarity of signatures
may rst be separated and subsequently identied by analyzing proteomic
signatures.
6.7.4 Rapid Detection of Invasive Aspergillosis
Invasive aspergillosis (IA) is a leading cause of morbidity and death in
patients with compromised immunity, particularly patients with hemato-
logic malignancy and recipients of hematopoietic stem cell or solid organ
transplantation. One of the major barriers to optimal care of these patients
is the difculty of identifying IA in its early stages—symptoms and radio-
logic ndings are nonspecic, and existing microbiologic methods are insuf-
ciently sensitive or specic to be reliable indicators of disease. Patients often
require lung biopsies, which are technically difcult and risky in patients
with compromised immunity and abnormal coagulation parameters.
In an immunocompromised host, IA can spread rapidly, with progressive
pneumonia and even disseminated disease. Rapid, accurate, noninvasive
identication of IA would improve clinical outcomes in these patients and
guide efforts to prescribe antifungal therapy more precisely to patients who
require treatment and spare those who do not—antifungal drugs are toxic,
costly, and have numerous complicated interactions with other medications.
Aspergillus species are metabolically versatile, capable of producing sec-
ondary metabolites. It has been shown that Aspergillus fumigatus and other
Aspergilli release unique sesquiterpene metabolite signatures unique to each
species. In a proof-of-concept study analyzing breath samples from patients
with suspected IA using thermal desorption GC-MS, a unique A. fumiga-
tus sesquiterpene metabolite signature distinguished patients with IA from
patients with other infections with 94% sensitivity and 93% specicity.107
Detection of these metabolites has been translated onto a GC-DMS instru-
ment, with identication of these metabolites within 30 min of breath sam-
pling at the bedside (Figure 6.17).
Infections caused by other mold species are clinically similar to IA but
are often optimally treated with other antifungal agents. There are no spe-
cic diagnostic modalities for these infections short of an invasive biopsy
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
135Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
procedure. In in vitro studies, human pathogens such as the Mucorales,
Fusarium, and Scedosporium have distinct secondary metabolite sesquiterpene
signatures that distinguish them from Aspergillus species; these signatures
may ultimately be harnessed for the noninvasive identication of infections
caused by these fungal species and differentiation of these infections from
IA.108–110
6.8 Conclusions
In the foregoing, DMS and FAIMS technologies have been described as a
means to provide rapid, inexpensive, and unambiguous detection of ana-
lytes ranging from chemical species present in industrial processes or the
environment to a host of biomarkers associated with various human dis-
eases. The ability to obtain early detection of trace levels of signatures of
human diseases in a noninvasive manner has the potential to revolution-
ize clinical diagnostics, and the data obtained for aspergillus detection and
in a range of other applications show promise toward a novel approach to
point-of-care diagnostics in settings far from advanced clinical care centers.
As the technology continues to advance, its utilization as a clinical diagnostic
is envisioned in doctor’s ofces and the home and in a wide range of devel-
oping world applications.
Baseline
Lung CT scan showing
suspicious nodule
1 week into antifungal
treatment
FIGURE 6.17
GC-DMS analysis of a breath sample from a patient with biopsy-proven invasive aspergillosis.
Breath sample at baseline contains sesquiterpene secondary metabolites (circled in red), which
disappeared after a week of effective antifungal therapy.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
136 Diagnostic Devices with Microuidics
References
1. Eiceman G. and Z. Karpas. 2005. Ion Mobility Spectrometry, 2nd edn., CRC Press,
Taylor & Francis LLC, Boca Raton, FL.
2. Buryakov I.A., E.V. Krylov et al. 1993. A new method of separation of multi-
atomic ions by mobility at atmospheric pressure using a high-frequency ampli-
tude-asymmetric strong electric eld. International Journal of Mass Spectrometry
and Ion Processes 128:143–148.
3. Makinen M., M. Nousiainen et al. 2011. Ion spectrometric detection technologies
for ultra-traces of explosives: A review. Mass Spectrometry Reviews 30:940–973.
4. Bruins A.P. 1994. Atmospheric-pressure-ionization mass spectrometry:
I. Instrumentation and ionization techniques. TrAC: Trends in Analytical
Chemistry 13:37–43.
5. Gorshkov M.P. 1982. Inventor’s certicate of USSR No. 966583, G01N27/62.
6. Avakov A.S., I.A. Buryakov et al. 1987. Mobility of multi-atomic ion species in
air and new possibility of their separation. In: Proceedings of the VII-th All-Union
Conference on Low Temperature Plasma Physics, Tashkent, Uzbek Republic.
7. Rasulev U.Kh., E.G. Nazarov et al. 1991. Identication of amine microimpurities
in air-gas mixtures by means of surface ionization with a subsequent ion mobil-
ity separation of desorbing ions. In: CIP 91 [Colloque international sur les procedes
plasma] No8, Antibes-Juanles-Pins, France, October 6, 1991, Vol. 47, pp. 409–411.
8. Buryakov I.A., E.V. Krylov et al. 1991. Analysis of ionic compositions of solu-
tions using an ion gas analyzer. In: Malakhov V.V., ed. Chemical Analysis of
Environment, Nauka, Novosibirsk, Russia, pp. 113–127.
9. Buryakov I.A., E.V. Krylov et al. 1991. Drift-spectrometer for trace amounts of
amines in atmospheric air, Preprint #44. Arifov Institute of Electronics, Tashkent,
Uzbekistan. Uzbek Academy of Sciences, p. 17.
10. Buryakov I.A., E.V. Krylov et al. 1991. Separation of ions according to mobility
in a strong AC electric eld. Soviet Technical Physics Letters 17:446–447.
11. Buryakov I.A., E.V. Krylov et al. 1993. Drift spectrometer for the control of amine
traces in the atmosphere. Journla of Analytical Chemistry 48:156–165.
12. Carnahan B., S. Day et al. 1995. Development and applications of transverse eld
compensation ion mobility spectrometer. In: Proceeding of Fourth International
Workshop on IMS, Cambridge, U.K.
13. Purves R.W. and R. Guevremont. 1999. Electrospray ionization high-eld asym-
metric waveform ion mobility spectrometry-mass spectrometry. Analytical
Chemistry 71:2346–2357.
14. Guevremont R. and R.W. Purves. 1999. Atmospheric pressure ion focusing in a
high-eld asymmetric waveform ion mobility spectrometer. Review of Scientic
Instruments 70:1370–1383.
15. Guevremont R. 2004. High-eld asymmetric waveform ion mobility spectrom-
etry: A new tool for mass spectrometry. Journal of Chromatography A 1058:3–19.
16. Nazarov E.G. and U.Kh. Rasulev. 1991. A surface ionization detector for amines,
hydrazines, and their derivatives (SID). Science—Technical Advances. All-Union
Institute of Scientic Information (VIMI), Moscow, Russia, Vol. 2, pp. 53–57.
17. Shvartsburg A.A., F. Li, K. Tang et al. 2006. High-resolution eld asymmetric
waveform ion mobility spectrometry using new planar geometry analyzers.
Analytical Chemistry 78:3706–3714.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
137Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
18. Guevremont R. 2001. High-eld asymmetric waveform ion mobility spectrom-
etry: A new tool for mass spectrometry. Journal of Chromatography A 1058:3–19.
19. Shvartsburg A.A., K. Tang et al. 2004. Understanding and designing eld
asymmetric waveform ion mobility spectrometry separations in gas mixtures.
Analytical Chemistry 76:7366–7374.
20. Prieto M., T.A. Prox, and R.A. Yost. 2010. Design and evaluation of a novel hemi-
spherical FAIMS cell. International Journal of Mass Spectrometry 298:41–44.
21. Marilyn P. and R.A. Yost. 2011. Spherical FAIMS: Comparison of curved elec-
trode geometries. International Journal for Ion Mobility Spectrometry 14:61–69.
22. Miller R.A., G.A. Eiceman et al. 2000. A novel micromashined high-eld
asymmetric waveform-ion mobility spectrometer. Sensors and Actuators B
67:300–306.
23. Miller, R.A., E.G. Nazarov et al. 2001. A MEMS radio-frequency ion mobil-
ity spectrometer for chemical vapor detection. Sensors and Actuators A
91:301–312.
24. Eiceman G.A., E.G. Nazarov et al. 2002. A micro-machined planar eld asym-
metric IMS as a gas chromatographic detector. Analyst 127:466–471.
25. Nazarov E.G., R.A. Miller et al. 2002. Monitoring of trace amounts of sulfur
compounds in air/hydrocarbon gas streams using a differential mobility spec-
trometer. International Journal for IMS 5(3):76–81.
26. Wilks A., M. Hart et al. 2012. Characterization of a miniature, ultra-high-eld,
ion mobility spectrometer. International Journal for Ion Mobility Spectrometry
15:199–222.
27. Mason E.A. and E.W. McDaniel. 1988. Transport Properties of Ions in Gases, Wiley,
New York.
28. Schneider B.B., E.G. Nazarov et al. 2015. Differential mobility spectrometry/
mass-spectrometry history, theory, design optimization, simulations, and appli-
cations. Mass Spectrometry Reviews 9999:1–51.
29. Schneider B.B., E.G. Nazarov et al. 2012. Peak capacity in differential mobility
spectrometry: Effects of transport gas and gas modiers. International Journal for
Ion Mobility Spectrometry 15:141–150.
30. Anderson A.G., K.A. Markoski et al. 2008. DMS-IMS2, GC-DMS, DMS-MS:
DMS hybrid devices combining orthogonal principles of separation for chal-
lenging applications. Proceedings of SPIE 6954:69540H (Apr.17).
31. Coy S.L., E.V. Krylov et al. 2008. DMS-preltered mass spectrometry for the
detection of biomarkers. Proceedings of SPIE 6954:695411 (Apr.17).
32. Carr T.W., ed. 1984. Plasma Chromatography, Plenum Press, New York.
33. Petinarides J., M.T. Grifn et al. 2005. Implementation of a new technology for point
detection. In: Gardner P.J., ed. Chemical and Biological Sensing VI, The International
Society for Optical Engineering, Vol. 5, pp. 65–74, Bellingham, WA. Proceedings of
SPIE: Volume 5795.
34. Limero Th., E. Reese et al. 2011. Preparation of a gas chromatograph-differential
mobility spectrometer to measure target volatile organic compounds on the
international space station. International Journal for Ion Mobility Spectrometry
14(2–3):81–91.
35. Limero Th., E. Reese et al. 2012. Results from the air quality monitor (gas-
chromatograph-differential mobility spectrometer) experiment on board the
international space station. International Journal for Ion Mobility Spectrometry
15:189–198.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
138 Diagnostic Devices with Microuidics
36. Levin D.S., P.A. Vouros et al. 2006. Characterization of gas-phase molecu-
lar interactions on differential mobility ion behavior utilizing an electrospray
ionization-differential mobility-mass spectrometer system. Analytical Chemistry
78(1):96–106.
37. Levin D.S., R.A. Miller et al. 2006. Rapid separation and quantitative analysis of
peptides using a new nanoelectrospray-differential mobility spectrometer-mass
spectrometer system. Analytical Chemistry 78(15):5443–5452.
38. Schneider B.B., T.R. Covey et al. Planar differential mobility spectrometer as a
pre-lter for atmospheric pressure ionization mass spectrometry. International
Journal of Mass Spectrometry 298:45–54.
39. Coy S.L., E.V. Krylov et al. 2010. Detection of radiation-exposure biomarkers
by differential mobility pre-ltered mass spectrometry (DMS-MS). International
Journal of Mass Spectrometry 291:108–117.
40. Hall A.B., S.L. Coy et al. 2013. Extending the dynamic range of the ion trap by dif-
ferential mobility ltration. Journal of the American Society for Mass Spectrometry
24:1428–1436.
41. SCIEX. n.d. SelexION® ION Mobility Technology. https://sciex.com/products/
ion-mobility-spectrometry/selexion-technology. Accessed March 20, 2017.
42. Tadjimukhamedov, F.Kh., A.U. Jackson et al. 2010. Evaluation of differential
mobility spectrometer coupled to a miniature mass spectrometer. Journal of the
American Society for Mass Spectrometry 21:1477–1481.
43. Manard M.J., R. Trainhama et al. 2010. Differential mobility spectrometry/mass
spectrometry: The design of a new, eld-portable mass spectrometer for real-
time chemical analysis. International Journal of Mass Spectrometry 295:138–144.
44. Nazarov E.G. 2012. Differential ion mobility mass spectrometry—The next
5 years—Status and history of development. E-book “Ion Mobility Mass
Spectrometry—The Next 5 Years.” http://www.owlstonenanotech.com/ultra-
faims/imsms-news/ebook-next-5-years. Accessed December 1, 2016.
45. Nazarov E.G., T.A. Postlethwaite et al. Planar differential mobility spectrometry
as a powerful tool for gas phase ion separation and detection (Spectroscopy
Solution: Premier Learning for Analytical Chemists). http://www.spec-
troscopy-solutions.org/Information/Archive/Featured-Articles/2220-/
Planar-Differential-Mobility-Spectrometry-as-a-Powerful-Tool-for-Gas-Phase-
Ion-Separation-and-Detection. Accessed December 1, 2016.
46. Schneider B.B., T.R. Covey et al. 2010. Control of chemical effects in the sepa-
ration process of a differential mobility mass spectrometer system. European
Journal of Mass Spectrometry 16(1):57–71.
47. Torok, T.J., R.V. Tauxe et al. 1997. A large community outbreak of salmonel-
losis caused by intentional contamination of restaurant salad bars. JAMA
278(5):389–395.
48. Keim, P., K.L. Smith et al. 2001. Molecular investigation of the Aum
Shinrikyo anthrax release in Kameido, Japan. Journal of Clinical Microbiology
39(12):4566–4567.
49. Jernigan, J.A., D.S. Stephens et al. 2001. Bioterrorism-related inhalational
anthrax: The rst 10 cases reported in the United States. Emerging Infectious
Diseases 7(6):933–944.
50. Okumura, T., N. Takasu et al. 1996. Report on 640 victims of the Tokyo subway
sarin attack. Annals of Emergency Medicine 28(2):129–135.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
139Planar Differential Mobility Spectrometry for Clinical Breath Diagnostics
51. Fernandez, M. and J. Weisman. 2004. Incident illustrates lapses in security net.
Washington Post, February 4, 2004.
52. Meselson, M., J. Guillemin et al. 1994. The Sverdlovsk anthrax outbreak of 1979.
Science 266(5188):1202–1208.
53. Darling, R.G., C.L. Catlett et al. 2002. Threats in bioterrorism. I: CDC category A
agents. Emergency Medicine Clinics of North America 20(2):273–309.
54. Lee, M.A., G. Brightwell et al. 1999. Fluorescent detection techniques for real-
time multiplex strand specic detection of Bacillus anthracis using rapid PCR.
Journal of Applied Microbiology 87(2):218–223.
55. Makino, S.I., H.I. Cheun et al. 2001. Detection of anthrax spores from the air by
real-time PCR. Letters in Applied Microbiology 33(3):237–240.
56. Turnbull, P.C., R.A. Hutson et al. 1992. Bacillus anthracis but not always anthrax.
Journal of Applied Bacteriology 72(1):21–28.
57. Bruno, J.G. and J.L. Kiel. 1999. In vitro selection of DNA aptamers to anthrax
spores with electrochemiluminescence detection. Biosensors and Bioelectronics
14(5):457–464.
58. Strizhkov, B.N., A.L. Drobyshev et al. 2000. PCR amplication on a microar-
ray of gel-immobilized oligonucleotides: Detection of bacterial toxin- and drug-
resistant genes and their mutations. Biotechniques 29(4):844–848, 850–842, 854
passim.
59. Higgins, J.A., M.S. Ibrahim et al. 1999. Sensitive and rapid identica-
tion of biological threat agents. Annals of the New York Academy of Sciences
894:130–148.
60. Uhl, J.R., C.A. Bell et al. 2002. Application of rapid-cycle real-time polymerase
chain reaction for the detection of microbial pathogens: The Mayo-Roche Rapid
Anthrax Test. Mayo Clinic Proceedings 77(7):673–680.
61. Phillips, A.P. and K.L. Martin. 1988. Investigation of spore surface antigens in
the genus Bacillus by the use of polyclonal antibodies in immunouorescence
tests. Journal of Applied Microbiology 64(1):47–55.
62. Longchamp, P. and T. Leighton. 1999. Molecular recognition specic-
ity of Bacillus anthracis spore antibodies. Journal of Applied Microbiology
87(2):246–249.
63. Yu, H. 1998. Comparative studies of magnetic particle-based solid phase uo-
rogenic and electrochemiluminescent immunoassay. Journal of Immunological
Methods 218(1–2):1–8.
64. Zhou, B., P. Wirsching et al. 2002. Human antibodies against spores of the
genus Bacillus: A model study for detection of and protection against anthrax
and the bioterrorist threat. Proceedings of the National Academy of Sciences USA
99(8):5241–5246.
65. Shute, L.A., C.S. Gutteridge et al. 1984. Curie-point pyrolysis mass spectrom-
etry applied to characterization and identication of selected Bacillus species.
Journal of General Microbiology 130(2):343–355.
66. Wang, Z., K. Dunlop et al. 2002. Mass spectrometric methods for generation
of protein mass database used for bacterial identication. Analytical Chemistry
74(13):3174–3182.
67. Demirev, P.A., Y.P. Ho et al. 1999. Microorganism identication by
mass spectrometry and protein database searches. Analytical Chemistry
71(14):2732–2738.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017
140 Diagnostic Devices with Microuidics
68. Krishnamurthy, T., M.T. Davis et al. 1999. Liquid chromatography/microspray
mass spectrometry for bacterial investigations. Rapid Communications in Mass
Spectrometry 13(1):39–49.
69. Fox, A., G.E. Black et al. 1993. Determination of carbohydrate proles of Bacillus
anthracis and Bacillus cereus including identication of O-methyl methyl-
pentoses by using gas chromatography-mass spectrometry. Journal of Clinical
Microbiology 31(4):887–894.
70. Vaidyanathan, S., D.B. Kell et al. 2002. Flow-injection electrospray ionization
mass spectrometry of crude cell extracts for high-throughput bacterial identi-
cation. Journal of the American Society for Mass Spectrometry 13(2):118–128.
71. Hathout, Y., P.A. Demirev et al. 1999. Identication of Bacillus spores by
matrix-assisted laser desorption ionization-mass spectrometry. Applied and
Environmental Microbiology 65(10):4313–4319.
72. Demirev, P.A., J. Ramirez et al. 2001. Tandem mass spectrometry of intact pro-
teins for characterization of biomarkers from Bacillus cereus T spores. Analytical
Chemistry 73(23):5725–5731.
73. Elhanany, E., R. Barak et al. 2001. Detection of specic Bacillus anthracis spore
biomarkers by matrix-assisted laser desorption/ionization time-of-ight mass
spectrometry. Rapid Communications in Mass Spectrometry 15(22):2110–2116.
74. Wang, Z., L. Russon et al. 1998. Investigation of spectral reproducibility in direct
analysis of bacteria proteins by matrix-assisted laser desorption/ionization
time-of-ight mass spectrometry. Rapid Communications in Mass Spectrometry
12(8):456–464.
75. Stowers, M.A., A.L. van Wuijckhuijse et al. 2000. Application of matrix-assisted
laser desorption/ionization to on-line aerosol time-of-ight mass spectrometry.
Rapid Communications in Mass Spectrometry 14(10):829–833.
76. Farquharson, S. and W.W. Smith. 2004. Differentiating bacterial spores from
hoax materials by Raman spectroscopy, Proceedings of SPIE, 5269:9–15.
77. Pineda, F.J., J.S. Lin et al. 2000. Testing the signicance of microorganism
identication by mass spectrometry and proteome database search. Analytical
Chemistry 72(16):3739–3744.
78. Jarman, K.H., S.T. Cebula et al. 2000. An algorithm for automated bacterial
identication using matrix-assisted laser desorption/ionization mass spectrom-
etry. Analytical Chemistry 72(6):1217–1223.
79. Phillips M. 1992. Breath tests in medicine. Scientic American July:74–79.
80. Zarling E.J. and M. Clapper. 1987. Technique for gas chromatographic mea-
surement of volatile alkanes from single breath samples. Clinical Chemistry
33:140–141.
81. Phillips M. and J. Greenberg. 1992. Ion-trap detection of volatile organic com-
pounds in alveolar breath. Clinical Chemistry 38:60–65.
82. Perman J.A., R.G. Barr et al. 1978. Sucrose malabsorption in children:
Noninvasive diagnosis by interval breath hydrogen determination. Journal of
Pediatrics 93:17–22.
83. Davidson G.P., T.A. Robb et al. 1984. Bacterial contamination of the small intes-
tine as an important cause of chronic diarrhea and abdominal pain: Diagnosis
by breath hydrogen test. Pediatrics 74:229–235.
84. Pauwels S., R. Fiasse et al. 1985. A simplied method of measuring breath
hydrogen by end-expiratory sampling for diagnosis of lactose malabsorption.
Acta Clinica Belgica 40:174–178.
Downloaded by [Jeffrey Borenstein] at 09:51 28 July 2017