Analytica Chimica Acta 463 (2002) 143 154
Automated control and optimisation of ion mobility
spectrometry responses using a sheath-flow inlet
a a b b c,"
D. Young , G.A. Eiceman , J. Breach , A.H. Brittain , C.L.P. Thomas
a
Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, NM 88003, USA
b
Graseby Dynamics Ltd., Park Avenue, Bushey, Watford, Herts. WD2 2BW, UK
c
Department of Instrumentation and Analytical Science, Institute of Science and Technology, University of Manchester,
UMIST, P.O. Box 88, Manchester M60 1QD, UK
Received 21 February 2002; received in revised form 29 April 2002; accepted 8 May 2002
Abstract
The stability and speed of the operation of sheath-flow inlet high temperature ion mobility spectrometer were studied
over sampled mass fluxes in the range 0 50 ng s-1 for dichloromethane and ethyl acetate. The response to step-changes in
sheath-flow in the region of 10 cm min-1 stabilised within 1.4 s, although, the recovery of a response on shutting down the
sheath-flow could be longer if adsorptive memory effects had built up within the inlet at high concentrations. The effect of
mass flux as opposed to sampled concentration was highlighted and the importance of the role of mixing in the reaction region
in controlling ion spectrometric responses emphasised. The incorporation of a sheath-flow inlet into an automated feed-back
control system was demonstrated with the system observed to maintain, without operator intervention, a linear response over
an order of magnitude increase in the mass flux compared to the same instrument without the sheath-flow inlet fitted. The
overall maximum sensitivity of the system was not significantly altered, and the drift time was not affected while the precision
of the responses to test atmospheres was comparable to other ion mobility spectrometric systems with a RSD in the responses
of less than 7%. © 2002 Elsevier Science B.V. All rights reserved.
Keywords: Ion mobility spectrometry; Dichloromethane; Ethyl acetate; Sampling; Sheath-flow inlet
1. Introduction several tens or thousands of units currently deployed
by NATO; and for the detection of narcotics and explo-
Ion mobility spectrometry (IMS) is an instrumen- sives [2,3], with IMS-based units now commonly de-
ployed at most major airports, and used by government
tal technique for the detection and identification of
law enforcement agencies. Recently, August 2001,
gas-phase chemical species [1]. The method is based
on the ionisation of analyte vapours and the characteri- an air quality monitor incorporating GC IMS tech-
sation of their mobility in a weak electric field at ambi- nology has been installed on the International Space
Station following successful tests of the prototype on
ent pressure. The inherent simplicity of the technology
space shuttle missions [4]. In laboratory-based work,
and low power requirements have led to IMS being the
IMS responses have been proven for a wide-range of
basis of several field analysers for specific applications
organic compounds [5], biological analytes [6,7] and
including: chemical warfare agent monitoring [1], with
inorganic chemicals [8], as well as analysing samples
in liquid [9] and solid phases [10,11]. Other appli-
"
Corresponding author. Tel.: [phone];
cations have been demonstrated in environmental
fax: [phone].
E-mail address: [e-mail] (C.L.P. Thomas). and process monitoring [12,13], as a chromatography
0003-2670/02/$  see front matter © 2002 Elsevier Science B.V. All rights reserved.
PII: S0003-2670(02)00380-X
144 D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154
detector [1], and for physical chemistry studies ions present such that no further response can be gen-
[14,15]. erated even if more analyte is introduced. This is due
The basis of IMS is the ionisation of the analyte. A to the time-dependent nature of reactant ion formation;
standard instrument is equipped with a emitting ra- the reactant ions are created in a complex multi-step
63
dioactive source, usually Ni. This causes ionisation process which is relatively slow compared to the one
of the supporting gas, typically air or nitrogen, leading or two stage processes leading to the creation of prod-
to the formation of reactant ions such as (H2O)nH+ uct ions. In other words, the reactant ions react faster
and (H2O)nO2- in a multi-step reaction [1]. These than they are produced. Even within the dynamic re-
reactant ions undergo collision induced reactions with sponse range of the technique, it has long been recog-
analytes or impurities in the supporting gas, for exam- nised that responses are rarely linear with analyte
ple, proton transfer, electron capture, or cluster forma- concentration over a significant range. It is acknowl-
tion. This leads to the formation of product ions based edged that most IMS calibrations have a region where
on the analytes present, which are then separated in linear response can be approximated which extends
the electric field and characterised by their mobility. over only one or two orders of magnitude of concentra-
These processes define some of the major strengths tion, and that responses are more accurately described
and weaknesses of the technique. A major strength is as exponential.
the high sensitivity of IMS due to the large number of These issues were discussed in a previous study
collisions at atmospheric pressure which allows effi- which addressed the problem of limited dynamic
cient ionisation of even trace levels of analyte vapours. range by varying the transfer efficiency of incoming
Limitations caused by these processes include the analyte into the ionisation region of an ion mobility
non-linear nature of the response and a limited dy- spectrometer by changing the level of a sheath-flow.
namic range that ends in detector saturation. Saturation Experiments were performed with dichloromethane
occurs when charge is taken from all of the reactant and ethyl acetate, and both analyte concentration and
Fig. 1. The effect of sheath-flow on the response of an ion mobility spectrometer to dichloromethane at a concentration of 300 mg m-3
(mass flux = 15 ng s-1). The intensity of the reactant ion peak, denoted by solid circles, reduces as the sheath-flow is increased while
the accompanying reactant ion peak, denoted by solid squares, increases. This effect may be used to protect an ion mobility spectrometer
from overloading and to extend the range of concentrations over which the device may be used.
D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154 145
sheath-flow were varied. It was observed that varying to the concentration dependence of the phenomenon.
the sheath-flow could significantly change the re- At that time, important issues relating to the stability
sponse to a fixed level of analyte, and that this effect and speed of the phenomena were not characterised,
appeared to operate rapidly and reproducibly. Ex- indeed the importance of demonstrating the effect in
trapolation from the data collected suggested that the a control loop and was highlighted as the next stage
linear dynamic range of the analytes studied could be in research programme.
increased significantly, perhaps to over three orders This study, therefore, sought to establish the stabil-
of magnitude of concentration. [16]. Variation of the ity and speed of a sheath-flow inlet and investigate the
sheath-flow to this inlet configuration was demon- effect of sample flow on the operation of the device.
strated to vary instrument sensitivity, and could even The feasibility of incorporating a sheath-flow inlet
quench responses altogether. Fig. 1 illustrates the into a system to enable automatic inlet protection was
observed changes in instrument response to negative also tested.
ions as the sheath-flow was varied between 20 and
80 cm3 min-1, with drift flow of 100 cm3 min-1 with
2. Experimental
0.30 g m-3 (15 ng s-1) of dichloromethane entering
the system. Only a minor product ion peak response
2.1. Instrumentation and apparatus
was apparent when the sheath-flow was 20 cm3 min-1,
but this was the dominant feature in the spectrum at
80 cm3 min-1. Correspondingly, the reactant ion peak The experimental arrangement used in this work
(RIP) was seen to be large at the lower sheath-flows is shown in Fig. 2. Air was used for all flows in
but was driven almost to complete depletion as the this system and was supplied from a cylinder source
sheath-flow was increased. This effect was observed (standard grade, BOC Ltd., Surrey, UK), via a pro-
to operate in the positive ion detection mode, with prietary gas purification and pressure regulation sys-
ethyl acetate as analyte. It was also noted that the way tem (Chrompack gas-clean system, Chrompack UK
in which the monomer and dimer product ion peak Ltd., London, UK). This included charcoal filters and
heights changed with sheath-flow appeared similar 5 Å molecular sieve traps and allowed independent
Fig. 2. A Schematic diagram of the experimental arrangement showing the gas connections and flow controls used in this work. Key to
labels: (A) purified air inlet; (B) pressure regulator; (C) needle-valve; (D) test atmosphere generator; (E) 4-way valve; (F) exponential
mixing and dilution flask; (G) sheath-flow inlet; (H) ion mobility spectrometer; (I) automated sheath-flow control valve running from
processed ion mobility spectrometer signal; and (J) exhaust line. The shaded area denotes the heated zone.
146 D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154
pressure regulation of three gas lines, which were the system. This method has been used widely in ana-
analyte flow, sheath-flow and drift flow. lytical work [17], including IMS studies [18]. The
The analyte flow was split into two lines, with both apparatus consisted of a 500 cm3 glass round bottom
flow rates set by Porter digital flow controllers (Var- flask fitted with a Dreschel head. This glassware was
ian Associates, California, USA). The default setting acid washed, deactivated with dichlorodimethylsilane
was 3 cm3 min-1. Flows were checked regularly be- (CAS [75 78 5], Aldrich Chemical Co., 99% purity),
ć%
fore experiments were run, and the average flow rate and stored at 200 C in a vacuum oven before use.
was found to be 3.03 cm3 min-1 (Ã = 0.07, n = 35). During use, it was kept hot by a heating mantle and
These two flows were passed through separate cham- insulated with aluminium foil, to minimise analyte
bers of a test atmosphere generator. adsorption. Surface temperature measurements were
The test atmosphere generator was a cylindrical made with a digital thermometer, which showed that
ć% ć%
stainless steel block with four individually sealed the flask was at an average of 100.4 C(Ã = 2.7 C,
chambers fitted with Swagelok connections (Manch- n = 20) during the dichloromethane experiments. For
ester Valve & Fitting Co. Ltd., Warrington, UK). One ethyl acetate experiments the flask temperature was
ć%
of these chambers was kept empty to allow blank re- increased, and the mean was found to be 124.1 C
ć%
sponses to be recorded and another held a permeation (Ã = 2.1 C, n = 16).
source of the chemical under study. Throughout this The default situation was for the apparatus to be
work, the vapour generator was kept thermostatted at running with the blank flow through the flask, with no
ć%
50 Ä… 0.5 C by a proportional derivative temperature analyte present in the system. To start an experiment
control unit, a 240 V/30 W cartridge heater placed the analyte flow was switched into the flask, and IMS
centrally within the block, and a type K thermocouple data acquisition was begun at the same time. Thus, the
(all from RS Components Ltd., Stockport, UK). acquisition times of IMS spectra allowed them to be
The analytes used in these studies were ethyl ac- correlated to analyte concentrations.
etate and dichloromethane, as they provide positive The transfer lines between the switching valve and
and negative mode responses, respectively, and ethyl the mixing flask, and the flask and the sheath-flow
acetate gives a monomer dimer response. Ethyl ac- inlet, were kept to as small a volume as possible to
etate (CAS [141 78 6]) was purchased from Fisons minimise dead volumes and analyte adsorption in the
Scientific Equipment (Loughborough, UK) and was system. To this end, polyamide coated methyl deacti-
residue analysis grade (quoted <2 ppm residue after vated capillary column (i.d. = 0.53 mm, Chrompack
evaporation, <0.05% water). Dichloromethane (CAS UK Ltd.) was used. This column was connected to
[75 09 2]) was purchased from Aldrich Chemical the capillary in the sheath-flow inlet (see later) with
Co. (Dorset, UK) and was HPLC grade (99.9% pu- a stainless steel zero dead volume union (Valco, pur-
rity quoted). The release rates of the permeation chased from Jones Chromatography Ltd., Glamorgan,
sources were determined by a series of weight loss UK). The transfer lines and the sheath-flow inlet were
measurements on the sources spread over a period of heated to further minimise analyte adsorption. This
storage in the test atmosphere generator of approx- was done by wrapping a heating cord (Fisher Scien-
imately 1000 h, and were found to be 0.912 mg h-1 tific, UK) around these regions and insulating with
(Ã = 0.001 mg, n = 7) for dichloromethane, and aluminium foil. The temperature was controlled by a
0.245 mg h-1 (Ã = 0.002 mg, n = 5) for ethyl acetate. proportional derivative temperature controller and a
The gas streams from the test atmosphere gener- type K thermocouple (both from RS Components Ltd.)
ć%
ator were fed through a 4-way valve that enabled and was set to match that of the mixing flask, 100 C
ć%
rapid switching of the analyte and blank lines into for dichloromethane experiments and 125 C for ethyl
and out of the experiment. Analyte concentration acetate experiments.
was varied by using an exponential mixing and di- The design of the sheath-flow inlet has previously
lution flask, as has been previously described [16]. been described [17], and is shown in Fig. 3. The inner-
This enabled concentration relationships to be effi- most (analyte) tube was an aluminium clad deactivated
ciently studied in a reproducible way, whilst avoid- capillary column, with o.d. 0.73 mm, and i.d. 0.53 mm
ing perturbation of the flow dynamics of the inlet (Fisher Scientific UK, Loughborough, UK). The sec-
D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154 147
Fig. 3. A line drawing with sections along Y to Y1 and X to X1 showing the dimensions and layout of the sheath-flow inlet used in this
study. The PTFE and stainless steel spacers have been omitted for clarity.
ond (sheath-flow) tube was stainless steel o.d. = tions. These spacers were tested to ensure that they
(1/8) in. (3.18 mm), i.d. = 1.35 mm (Dockweiler UK remained stable when flow passed over them, whilst
Ltd., Wrexham, UK). The sheath-flow was introduced not obstructing flow.
into this tube via an (1/8) in. stainless steel T-union The sheath-flow was set and monitored by a mass
(Manchester Valve & Fitting Co. Ltd., Warrington, flow controller (model UFC-1100GI, 200 cm3 min-1
UK), which sealed against the capillary at the far end full scale, UNIT Instruments Ltd., Dublin, Eire).
with a graphite reducing ferrule. The outermost (vent) An input signal to this device between 0 and +5V
tube was stainless steel o.d. = (5/16) in. (7.94 mm), would give a proportional flow output between 0
i.d. = 6.11 mm (Dockweiler UK Ltd.). Vent gas exited and 200 cm3 min-1. An output voltage (between
from this tube via a (5/16) in. stainless steel T-union 0 and +5 V) was also generated by the gas leav-
(Manchester Valve & Fitting Co. Ltd.), which sealed ing the device, and provided an indication of the
against the (1/8) in. tube at the far end via a graphite current sheath-flow rate. Calibration of this output
reducing ferrule. Once constructed, the pneumatic signal against replicate measurements with a bubble
integrity of this unit was validated by pressure testing. flow-meter gave a linear relationship with R2 > 0.99.
The three tubes ended together, approximately in The input signal voltage to the flow controller was
the same plane at the edge of the ionisation region supplied by an analogue output line from a PC in-
of the IMS cell. Initial testing had suggested that the terface card, with a capacitive isolation chip (model
symmetry of the sheath tube within the vent tube was ISO122P, Farnell Electronic Components Ltd., Leeds,
important to proper operation, whilst the symmetry of UK) between the two devices to protect them both.
the capillary within the sheath tube did not appear to This meant that the required flow could be controlled
be critical. Nevertheless, the sheath tube was held ax- from the PC. The output from this flow controller was
ially inside the vent tube with a PTFE spacer, which connected to the (1/8) in. T-union on the sheath-flow
consisted of a collar section around the sheath tube inlet by a short length of (1/8) in. stainless steel tube,
and three narrow legs of equal length which rested to provide the sheath-flow into the system.
against the inner walls of the vent tube. The capillary The IMS cell used was a Graseby Dynamics Ltd.
was supported inside the sheath tube by a single strand high temperature unit (HiT-IMS), the specifications
of wire which wrapped around the capillary and then and operating parameters of which are listed in
63
pushed against the sheath tube walls in several direc- Table 1. The 10 mCi ionisation source was Ni plated
148 D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154
Table 1
in-house in Pascal (Turbo Pascal, Version 6.0, Borland
Ion mobility spectrometer operating parameters
International Inc., California, USA), which controlled
acquisition of data from the IMS cell, signal averag-
Ion mobility spectrometer
Type A high temperature ceramic
ing, and the saving of data files.
drift tube with external field
Automated changes in sheath-flow were achieved
defining electrodes
by using a Pascal program to control the mass flow
Inlet Direct inlet to the reaction
controller to create a feedback loop from the IMS
region
cell response. The interface card acquired the IMS re-
Drift gas Air
Drift length (cm) 4.25
sponses, which were evaluated within the Pascal pro-
Drift field (V cm-1) 200
gram. On the basis of this, the program defined the
63
Ionisation source (MBq) Ni 370
voltage level to be sent from the card to the mass flow
Cell temperature (K) 423
controller, which then altered the sheath-flow rate and
IMS signal processing
thus, also the IMS response. This approach enabled
Scan frequency (Hz) 50
the implementation of algorithms to control the out-
Data points per scan 1024
put voltage used as the input to the sheath-flow mass
Signal averaging 250 scans
Gating pulse duration ( s) 180 flow controller.
Acquisition frequency (kHz) 52
2.3. Experimental studies
onto the surfaces of a nickel plated brass tube 7.4 mm
Three experimental campaigns were undertaken to
in diameter and 7.0 mm long. The cell drift flow was
demonstrate the concept of automated sheath-flow
controlled by a needle valve and measured by a ro-
control of IMS:
tameter, with default setting of 100 Ä… 5cm3 min-1.
The drift gas was pre-heated before entering the cell
" a study of the effect of analyte flux;
by passing it through copper tubing which had been
" characterisation of the stability and speed of a
coiled around the body of the heated cell, below
sheath-flow inlet running under automated control;
the insulation. The sheath-flow inlet was connected
and
into the front inlet of the instrument by making a
" the assessment of the performance of an automated
direct Swagelok connection between this and the
sheath-flow system against different analyte fluxes
sheath-flow inlet vent tube ((5/16) in.). Before sealing
by varying the concentration of the sampled gas.
the connection, it was ensured that the inlet reached
the ionisation source.
2.3.1. Analyte flux phenomena
Four experiments were run across a range of
2.2. Data acquisition and feedback system
ethyl acetate concentrations spanning the range
0.05 0.69 g m-3, with the flow of analyte adjusted to
The PC used was equipped with a 386 processor give approximately the same mass flux of material
and 7.5MB RAM, and an interface card equipped with into the reaction region of the instrument. This study
analogue and digital input and output channels (ADC enabled the sensitivity of the instrument to changes
42, Blue Chip Technology, Clwyd, UK). The card was in the analyte flow rate to be assessed (see Table 2).
configured for grounded analogue inputs in the range
Ä…2.5 V, providing resolution of 1.25 mV, and all dig- 2.3.2. Speed of response tests
ital channels were set as inputs. One of the analogue The speed with which the sheath-flow effect op-
input channels was used for the instrument signal, and erated was evaluated by using computer control
one of the digital channels was used to monitor the of sheath-flow rate. Programs were written which
trigger pulse. Also, one of the analogue output lines changed the sheath-flow at defined points during a set
was used to provide the signal voltage to the mass flow of acquisitions. These experiments were undertaken
controller to define the sheath-flow. The operation of with four averages per spectrum, as this was found to
the interface card was controlled by a program written reduce acquisition time to 0.7 s per averaged spectrum
D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154 149
Table 2
Summary of analyte flux rate study
Reference
[1] [2] [3] [4]
[Ethylacetate] (g m-3) 0.05 0.09 0.3 0.69
Analyte flow (cm3 min-1) 20 10 3 1.3
Analyte flux (ng s-1) 16.7 15.0 15.0 15.0
RIP intensity (V) 0.388 0.406 0.318 0.190
Protonated ethyl acetate product ion peak intensity (V) 0.507 0.509 0.555 0.371
Proton bound ethyl acetate dimer (V) 0.099 0.093 0.076 0.192
whilst still producing acceptable signal to noise qual- from its current level, Ft (cm3 min-1), to a new one,
ity in the captured mobility spectra. Speed of re- Ft +1 (cm3 min-1). This initial height was taken from
sponse experiments were performed with both ethyl an ion mobility spectrum obtained in the absence of
acetate and dichloromethane, at drift flows of 100 and analyte, and assigned as a variable to be used in sub-
300 cm3 min-1 at a constant concentration and mass sequent calculations. Alternative thresholds could be
flux of analyte, the experimental parameters used are a constant value chosen from previous calibrations,
summarised in Table 3. Each experiment was per- and this was used when the product ion peak was
formed with steps both up and down in sheath-flow. the trigger feature, or comparison of the current peak
height with that from the previous acquisition. The
2.3.3. The effect of analyte flux on the system most commonly used algorithm was a step-change in
under automatic control sheath-flow, where each time the threshold was ex-
The critical feature in the quality of operation of this ceeded the sheath-flow was lowered by a constant
system is the control algorithm. There are numerous amount, for example:
ways to implement such a function and it was beyond
the scope of this study to conduct an exhaustive exam- If IRIP <(0.7 × IRIP,0),
ination of feedback control algorithms for sheath-flow
then Ft+1 = (Ft+1 - 10) (1)
IMS. Instead, efforts were focused on using simple
coding to investigate the concept. The spectral fea- The majority of experiments with the feedback system
ture used to control the sheath-flow triggering was the were performed with an analyte flow of 3 cm3 min-1
RIP intensity, IRIP (V), although other features were and a drift flow of 100 cm3 min-1, though a drift flow
examined, such as the product ion peak intensity for of 300 cm3 min-1 was also investigated. These studies
the dichloromethane responses. A threshold value of used the step-change algorithm described earlier by
70% of initial RIP height, IRIP,0 (V), corresponding to Eq. (1) whilst performing exponential concentration
the lower limit of the linear response region was se- experiments. Both dichloromethane and ethyl acetate
lected to act as a trigger for changing the sheath-flow were considered.
Table 3
Speed of response experiments parameters
Reference
[5] [6] [7] [8] [9]
Drift flow (cm3 min-1) 100 100 100 100 100
Minimum sheath-flow (cm3 min-1) 1010101010
Maximum sheath-flow (cm3 min-1)8080705025
Step size (cm3 min-1) 2070104015
150 D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154
3. Results and discussion is helpful to differentiate between the concentration
of the sampled gas that is presented as analyte flow
and the effective concentration of the analyte species
3.1. Analyte flux phenomena studies
in the reaction region following the mixing processes
engendered by the sheath-flow inlet and its interaction
Fig. 4 compares spectra obtained from essentially
with the drift flow. It is important, therefore, that in
the same mass fluxes of ethyl acetate at four different
this instrument responses are related to mass fluxes
analyte flow rates and hence, concentrations. The drift
and not only to concentration.
flow and sheath-flows were kept constant throughout
these experiments at 100 and 50 cm3 min-1, respec-
3.2. Stability and speed of sheath-flow responses
tively Although, there was a 15-fold difference in
analyte concentration across the range of experiments
the differences between the responses were minor. It Typical responses obtained from the step-change
did appear that the response for an analyte flow of speed of response experiments for dichloromethane
20 cm3 min-1 was more pronounced than the others at mass flux of 15 ng s-1 (300 mg m-3) are shown
with a greater dimer to monomer ratio. Indicating in Fig. 5. They show the speed at which the RIP
a change in the mixing regime as the analyte flow signal changed with reducing sheath-flow. Similar
approaches that of the sheath-flow. Similar behaviour behaviours but in the opposite direction were seen
was seen with dichloromethane, at three different an- for RIP recovery as sheath-flow was stepped down,
alyte flows studied. Again there was little difference and experiments with ethyl acetate yielded the same
between the three responses, despite an eight-fold characteristics. The data obtained from all of these
difference in analyte concentration. Different exper- experiments showed that the response to changes in
iments with different sheath-flows provided similar sheath-flow in the region of 10 cm min-1 stabilised
behaviours. These results indicate that in this config- within 1.4 s and that the time to stabilise did not ap-
uration sheath-flow was more important in dictating pear to be dependent on the analyte, and therefore, the
the instrument response than analyte flow. Thus, it detection polarity. Operating the system at high mass
Fig. 4. A comparison of ethyl acetate spectra obtained for mass flux inputs of approximately 15 ng s 1 to the detector, but with different
analyte flow rates annotated on the figure, and concentrations, see Table 2. Note that the tailing between the monomer and dimer peaks
at approximately 5.25 and 6.25 ms, respectively is due to thermal decomposition of the dimer to the monomer form in the drift tube,
promoted by the elevated temperatures used in this study. The conditions used are summarised in Table 1.
D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154 151
Fig. 5. A plot showing the speed of the changes in the intensity of the reactant ion peak (RIP) against changes in the level of sheath-flow
for dichloromethane at mass flux of 15 ng s-1 (300 mg m-3). The circles indicate the RIP intensity while the solid line is the sheath-flow.
fluxes with relatively high concentrations resulted in 3.3. Automated control of responses
the build up of adsorption artefacts in the interface.
In such instances, the recovery of the response on During operation of the automated system the me-
shutting down the sheath-flow could be significantly chanics of the feedback loop were observed to work
longer than 1.4 s, see Fig. 6 for an example with satisfactorily. Inspection of the data showed that
ethylacetate at 20 ng s-1, especially if large changes changes in sheath-flow occurred at the times when
in the sheath-flow were invoked. the peak height threshold conditions were properly
Fig. 6. Speed of change and stabilisation of reactant ion peak height on sheath-flow steps down from 25, 50 and 80 to 10 cm3 min-1. This
experiment was run with ethylacetate at a mass flux of 20 ng s-1 (0.40 g m-3), and data points were taken every 0.7 s. Note how much
longer the recovery from saturation takes for the 80 10 cm3 min-1 step-change. This is due to adsorptive memory effects in the interface.
152 D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154
Fig. 7. A plot of RIP intensity against sheath-flow under automated control during an exponential increase in ethylacetate mass flux. The
solid line indicates the sheath-flow level while the joined circles denoted the RIP intensity.
10 cm3 min-1, with a programme duty cycle of 3.75 s.
fulfilled, and that these changes were affected rapidly
This plot also illustrates how the sensitivity was suc-
and accurately. Operation of the system was found to
cessively reduced as the sheath-flow was lowered,
be robust and reproducible. Initial testing proved that
the algorithms could operate on any threshold level re- and each level was correspondingly maintained for
a larger flux range. Fig. 8 shows the quality of the
quired, but the default was 70% of initial RIP height.
ion mobility spectra obtained during this study. In
Fig. 7 shows how the RIP height varied with
comparison the same ion mobility spectrometer, chal-
ethyl acetate flux when the algorithm stepped the
lenged with the an identical concentration profile but
sheath-flow from 70 to 30 cm3 min-1 in steps of
Fig. 8. Ion mobility spectra from the automated sheath-flow experiment for ethyl acetate at various sheath-flows and analyte fluxes. The
analyte fluxes were 0.9, 2.4 and 5.5 ng s-1 for the 30, 40 and 70 cm3 min-1 sheath-flows, respectively. The spectra have been offset by
approximately 0.5 V for clarity.
D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154 153
without a sheath-flow inlet, saturated at a ethylacetate ment responses. Further there is little known about
mass flux of 5 ng s-1. the variability introduced into ion mobility spectro-
metric experiments due to differences in gas flow and
mixing patterns within the instrument. Meanwhile,
4. Conclusion
custom and practice have led to concentration being
consistently used as the defining parameter in IMS
At maximum flow the sheath-flow inlet system vapour analysis studies. This study suggests that the
was found to provide sensitivities slightly greater mass flux of analyte is a better defining parameter
than those obtained in conventional IMS instruments, and should be used instead of concentration.
whilst reducing the sheath-flow rate could lower the The response speed of the inlet was found to be sat-
instrument sensitivity to <1% of the maximum value. isfactory. Changes on the time-scale of 1 or 2 s suggest
Using the sheath-flow effect extended the linear re- applications in GC IMS systems to allow control of
sponse range (defined as RIP maintained at >70% response during the elution of a chromatographic run,
maximum height) for ethyl acetate by over an order of and would also allow a portable instrument to cope
magnitude. The system stabilised to a new sensitivity with most situations encountered in environmental or
in typically <2 s, and replicate studies (n = 3) sug- process control applications. A crucial feature in main-
gested RSDs in RIP heights of <7%. Such precision taining this speed of operation will be the specification
compares favourably with conventional IMS systems. of the flow controller used to provide the sheath-flow,
Response levels showed no significant dependence on as responses appeared to stabilise almost as soon as
analyte flow rate. the sheath-flow was steady. Another feature of this in-
Operation of the sheath-flow inlet was successfully let configuration was the recovery behaviour of the
automated. A computer program was produced that instrument, returning from full saturation in ca. 90 s
continuously monitored peak heights and changed without interrupting sample introduction. Provided the
the sheath-flow level when responses strayed outside instrument or sample input paths had not become con-
pre-defined parameters. This allowed: automatic pro- taminated then the system would be ready to perform
tection against saturation; continued operation in the further analysis within this period.
linear response region, by reducing sheath-flow when Originally, the sheath-flow inlet was designed as
responses got too large; and, optimisation of sensi- a GC interface for IMS. This study sought to char-
tivity by increasing the sheath-flow when responses acterise and exploit the features of such a system to
were low. protect an IMS from saturation, and enhance the range
The analyte source flow studies showed that a of concentrations it could be used for in quantitative
range of sample flows may be used in the system de- work. The results obtained demonstrate that with au-
scribed, and, the utility of defining IMS responses in tomated flow control the sheath-flow inlet effectively
terms of mass fluxes rather than concentrations was protects an instrument from saturation, and increases
highlighted. It is interesting to note that IMS flow pa- the linear response range.
rameters have never been standardised, and are rarely The next stages in the application of this approach
the same between different workers or instruments. are likely to focus on sheath-flow GC IMS appli-
Indeed little work has been presented that systemati- cations. A mixture of analytes contains a range of
cally evaluates the effect on ion mobility spectrometer reactivity constants, [16], and this results in situations
responses of altering gas flows. This work has shown where for similar on-column masses one component
that the nature of the response of an IMS is strongly yields an indiscernible response whilst another sat-
influenced by the mixing that takes place within the urates the instrument. Controlling the sheath-flow
reaction region of the instrument. Indeed, it is also would enable such separations to be run with the gas
possible to configure an instrument so that widely flow parameters optimised for the individual analytes
differing concentrations yield essentially the same re- as they eluted.
sponses. A study of the literature of IMS reveals that Finally, the recovery of quantitative data from a dy-
little has been reported on the effects that gas flows namically controlled sheath-flow systems requires a
and inlet and reaction region designs have on instru- multivariate calibration that relates flow parameters
154 D. Young et al. / Analytica Chimica Acta 463 (2002) 143 154
and instrument response to analyte flux and hence, on Ion Mobility Spectrometry, Hilton Head Island, South
Carolina, USA, 1998.
sampled concentration. Initial work in this area has
[5] S. Bell, E.G. Nazarov, G.A. Eiceman, W.F. Wang, E.R.
begun and suggests that the sheath-flow effect can be
Miranda, Classification of ion mobility spectra using neural
modelled by a sensitivity parameter that is indepen-
networks, in: Proceedings of the 7th International Conference
dent of the analyte under study and applies to all sit-
on Ion Mobility Spectrometry, Hilton Head Island, South
uations. The development of these concepts is also a Carolina, USA, 1998.
[6] A.P. Snyder, D.B. Shoff, G.A. Eiceman, D.A. Blyth, J.A.
logical continuation of the work reported in this study.
Parsons, Anal. Chem. 63 (1991) 526 529.
[7] Y. Liu, S.J. Valentine, A.E. Counterman, C.S. Hoaglund, D.E.
Clemmer, Anal. Chem. 69 (1997) 728A 735A.
Acknowledgements
[8] G.A. Eiceman, C.S. Leasure, V.J. Vandiver, Anal. Chem. 58
(1986) 76 80.
The authors gratefully acknowledge the support [9] A.R.M. Przybylko, C.L.P. Thomas, P.J. Anstice, P.R. Fielden,
J. Brokenshire, F. Irons, Anal. Chim. Acta 311 (1995) 77 83.
from the EPSRC and Graseby Dynamics under the
[10] C. Wu, W.F. Siems, G.R. Asbury, H.H. Hill Jr., Anal. Chem.
CASE award scheme that was provided to David
70 (1998) 4929 4938.
Young during the course of this research.
[11] G.A. Eiceman, D.A. Blyth, D.B. Shoff, A.P. Snyder, Anal.
Chem. 62 (1990) 1374 1379.
[12] Q. Meng, Z. Karpas, G.A. Eiceman, Int. J. Environ. Anal.
References
Chem. 61 (1995) 81 94.
[13] G.A. Eiceman, S. Sowa, S. Lin, S.E. Bell, J. Hazard. Mater.
[1] G.A. Eiceman, Z. Karpas, Ion Mobility Spectrometry, CRC 43 (1995) 13 30.
Press, Boca Raton, FL, 1993. [14] K. Giles, E.P. Grimsrud, J. Phys. Chem. 96 (1992) 6680
[2] Z. Karpas, Forensic science applications of ion mobility 6687.
spectrometry, Forensic Sci. Rev. 1 (1989) 104 119. [15] R.G. Ewing, G.A. Eiceman, J.A. Stone, Int. J. Mass Specrosc.
[3] L.L. Danylewych-May, C. Cumming, Explosive and taggant 193 (1999) 57 68.
detection with ionscan, in: J. Yinon (Ed.), Advances in [16] D. Young, C.L.P. Thomas, J. Breach, A.H. Brittain, G.A.
Analysis and Detection of Explosives, Kluwer Academic Eiceman, Anal. Chim. Acta 381 (1999) 69 83.
Publishers, Norwell, MA, 1993, pp. 385 401. [17] A. Zlatkis (Ed.), Electron Capture: Theory and Practice in
[4] T. Limero, J. Trowbridge, E. Reese, J. Breach, J. James, Chromatography, Elsevier, Amsterdam, 1981.
Evaluation of inflight results from the volatile organic [18] W.A. Munro, C.L.P. Thomas, M.L. Langford, Anal. Chim.
analyser, in: Proceedings of the 7th International Conference Acta 375 (1998) 49 63.