- Email: [email protected]
Qualitative analysis of trace constituents by ion mobility increment spectrometer
Talanta 61 (2003) 369 /375 www.elsevier.com/locate/talanta
Qualitative analysis of trace constituents by ion mobility increment spectrometer I.A. Buryakov The Design and Technological Institute of Instrument Engineering for Geophysics and Ecology (IDE), The Siberian Branch of RAS, Pr. Ak. Koptyuga 3, Block 6, 630090 Novosibirsk, Russia Received 23 September 2002; received in revised form 28 April 2003; accepted 29 April 2003
Abstract Ion mobility increment spectrometry (IMIS) is a high sensitive selective ionization technology for detection and identification of ultra-trace constituents, including toxic compounds, CW-agents, drugs and explosives in ambient air or liquid sample. Like an ion mobility spectrometry (IMS), this technology rests on sampling air containing a mixture of trace constituents, its ionization, spatial separation of produced ions and separated ions detection. Unlike IMS, ions of different types in IMIS are separated by ion mobility increment, a . Value a , is a function of the parameters: electric field strength and form, atmospheric pressure. To exclude the influence of these parameters on an a , the method of explosives identification by a standard compound was suggested. As a standard compound iodine was used. The relationship among the mobility coefficient increments equal to the relationship among the compensation voltage ai / aiodine /Ui /Uiodine is determined, where i are ions of 1,3-dinitrobenzene, 1,3,5-trinitrobenzene, p -mononitrotoluene, 2,4-dinitrotoluene and 2,4,6-trinitrotoluene This relationship is practically independent of the above mentioned parameters in the range 25 B/E /N B/90 Td. The limits of the relative error of this relationship are determined both from spectra of individual compounds and nitrocompound-iodine mixtures. # 2003 Elsevier B.V. All rights reserved. Keywords: Explosives; Ion mobility spectrometry; Atmospheric pressure ionization
1. Introduction Devices that combine high sensitivity, selectivity and capacity of compound identification are required to perform qualitative analysis of trace
E-mail address: [email protected] (I.A. Buryakov).
constituents. An ion mobility increment spectrometer (IMIS) can be used to detect and identify ultra-trace constituents: toxic compounds, CWagents, drugs and explosives in ambient air or liquid sample [1 /6]. Like an ion mobility spectrometer (IMS), the operation of IMIS rests on sampling air containing a mixture of trace constituents, its ionization, spatial separation of
0039-9140/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0039-9140(03)00305-9
370
I.A. Buryakov / Talanta 61 (2003) 369 /375
produced ions and separated ions detection. The names like ion non-linear drift spectrometer [5], high-field asymmetric waveform IMS [7], field ion spectrometer [8], radio-frequency-based ion-mobility analyzer [9] are used as a synonym for IMIS. IMIS differs from IMS in that ions of different types are separated in IMIS by ion mobility increment that depends on electric field strength [10]. Ion drift velocity, V , caused by an action of electric field (E ) is [11]: V K(E)E; K(E=N)K0 (1a(E=N)) 2n X E K0 1 a2n ; N n1
(1)
where K0 (cm2 (V cm) 1) is the mobility coefficient in a weak field (E /N B/6 ‘townsends’ or Td, where 1 Td /1 /1017 V cm2), N is neutral gas density, a (E /N ) is a normalized function which describes the electric field dependence of the mobility */the mobility coefficient increment, a2n are decomposition coefficients, n ]/1 is an integer. Under the action of periodic alternating asymmetric waveform field, Ed(t) /Edf (t ) (Ed is an amplitude, f(t) is a form of the field), that meets the conditions [12]:
g
tT
f (t)dt0; t
1 T
g
(2)
tT
f
2n1
(t)dthf
2n1
i"0;
t
ions executing fast oscillatory motions with period, T , drift with velocity Vi characteristic for each i-th component proportional to ai (E /N ) (broken brackets mean averaging over a period). Should the ion drift of the i -th component be compensated by constant electric field, Eci , the ion velocity will be zero: hVi ihK0 (1ai )Ed f (t)Eci i0
(3)
The compensation field of the i-th component is expressed by substituting decomposition of coefficient, a , from Eq. (1) into Eq. (3) and using condition (Eq. (2)) and approximation (Ed/ Ec)n :/End /nEn1 Ec, with jEdj/jEcj [10]: d
X 2n Ed 2n1 ai2n hf i Eci : Ed N n1 2n X E 1 (2n1)ai2n d hf 2n1 i : N n1
=
(4)
The ions for which Vi /0 are transported to a collector with a carrier-gas, the other ions leave a separation region and recombine. With Ec changing, spectrum of a mixture of ions of all types is recorded. A separation region can be of planar or axial symmetry depending on which electrodes, made of two plane-parallel plates [4,10] or two coaxial cylinders [1,3,5], are used. Electric field strength is equal to: E U=d;
E(r)U=(r ln(r2 =r1 ));
(5)
respectively, where U is the potential difference across the electrodes, d is a distance between the plane /parallel plates, r is a radial coordinate, r1 and r2 are radiuses of the inner and external cylinders, respectively. From Eqs. (4) and (5) one can see that compensation voltage, Uci , is one-of-a-kind for each ion type and is determined by coefficients, a2ni , that depend on temperature, T , drift gas composition (e.g. the presence of water vapors (CH2O) and etc.). Voltage, Uci is a function of the amplitude of asymmetric waveform dispersion voltage, Ud, voltage waveform, f(t), density, N , distance, d , or radiuses, r1 and r2. Problems in comparing the obtained experimental results arise if devices with different d or r1, r2 are used, since one and the same ion types are recorded at spectra with different Uc values (e.g. TNT ions at [5] and [13]). Fluctuations of T , CH2O, Ud, f (t ) and N cause fluctuations of Uc, and this complicates the analyzed compound identification. To enhance reproducibility and certainty of analyte identification of a spectrometric device any standard compound can be used for a calibration [14]. As a standard compound any available easy to handle compounds can be used. For example, in [15] with the IMS in positive mode 2,4-lutidine was used as a standard chemical compound. Trace constituents of 1,3-dinitrobenzene, 1,3,5trinitrobenzene, p-mononitrotoluene, 2,4-dinitro-
I.A. Buryakov / Talanta 61 (2003) 369 /375
toluene and 2,4,6-trinitrotoluene in air were studied. The iodine was used as a standard compound. The a (E /N ) dependencies of a mobility coefficient increment for negative ions of these compounds are determined. An agreement between the relationship among the a (E /N ) dependencies for different ion types and the relationship among their compensation voltages is shown. The relationship among the compensation voltages can be used for the qualitative analysis of the trace constituents. Standard compensation voltage deviations are determined both for individual compounds and nitrocompound-iodine mixtures.
371
and r2 /1.8 cm in diameter, the chamber is purged with a carrier-gas (dry air with a water vapor concentration 5/100 ppm, flow rate Qtr /20 /50 cm3 s 1, temperature tg /259/1 8C, N /(2.439/ 0.07) /1019 cm 3), an electrode (9/200 V), for transporting ions from the ionization chamber into the separation chamber, an ion collector, an electrometer (width of a zero line of the electrometer noise Wz /2 /1014 A), a compensation voltage generator. A dispersion voltage generator used is developed by IDE. The generator has the following parameters: Voltage form (Fig. 1)
2. Experimental
f (t) (sin[p(tmT )=t]2t=pT)=(12t=pT); with mT 5t5(mT t); f (t)
2.1. Instrumentation
(2t=pT)=(12t=pT); with (mT t)5 t5 (m1)T;
Block diagram of IMIS is given on Fig. 1. The analytical unit of the device is developed by the Mine Safety Appliances Company [8] and comprises a heated ionization chamber (sampling flow rate, Qs /1.2 cm3 s 1, b-source 63Ni, temperature, ti /150 8C), an ion separation chamber formed by two electrodes which are coaxial cylinders, r1 /1.4
(m ]/0 is an integer); high-voltage pulse duration, t/2.7 ms, period T /6.8 /7.8 ms, dispersion voltage amplitude, Ud //0.7 to /5 kV. We used an oscillograph to control the form and amplitude of Ud(t) voltage. Absolute error in compensation voltage measurement is dUc /9/0.05 V; relative measuring error of the compensation voltage is dUc /9/1%, asymmetric waveform voltage amplitude is dUd /9/7% and time intervals is dt /9/ 7%.
(6)
2.2. Chemicals
Fig. 1. Block diagram of experimental unit. (1) ionization chamber, (2) ion separation chamber, (3) electrode for transporting ions, (4) dispersion voltage generator, (5) compensation voltage generator, (6) ion collector, (7) electrometer; Qs, air / vapor mixture sample; Qin, purified air inlet flow; Qtr, carriergas; Qout, outlet flow.
Samples of 1,3-dinitrobenzene (DNB), 1,3,5trinitrobenzene (TNB), p-mononitrotoluene (MNT), 2,4-dinitrotoluene (DNT) and 2,4,6-trinitrotoluene (TNT), crystalline iodine in highest purity were obtained from the Research Institute of Special Technics and Communication (Novosibirsk, Russia). To obtain air /vapor mixtures of nitrocompounds and/or iodine, we mixed the flows of purified air and purified air passed through quartz tubes (¥ 0.3 cm, 20 cm in length) the inside of which were covered with tested compounds. Table 1 shows the main types of ion-molecule reactions into which the molecules of tested compounds enter being ionized by b-source at
I.A. Buryakov / Talanta 61 (2003) 369 /375
372
Table 1 The main types of ion-molecule reactions, mass /charge ratio (m /z ) and types of ions produced Compound
Reaction
m /z
Ion
Refs.
p-MNT DNT TNT
136 181 226
(M/H)
[16,17]
Proton abstraction
TNB DNB
Associative electron capture
213 168
M
[18,19] [20]
Iodinea
Dissociative electron capture
127
M
[21]
M is molecule or atom of a compound, H is hydrogen atom. a b-Ionization in N2.
atmospheric pressure, mass /charge ratio (m /z ) and types of ions produced (M is molecule or atom of a compound, H is hydrogen atom). In the last column one can find references to the papers where these ions were identified by mass-spectrometry.
3. Results and discussion 3.1. Dependence of Uc on the separation conditions and IMIS parameters The IMIS spectrum is the compensation voltage dependence of the ion current (I(Uc)), each ion type is recorded as an ion peak. Fig. 2 shows a spectrum for a mixture of vapors: iodine, MNT, DNB, DNT, TNB and TNT in air taken with
Fig. 2. Spectrum of a vapors mixture of iodine, MNT, DNB, DNT, TNB, TNT in air. Ud /3900 V.
Ud /3900 V. The Uci values corresponding to the peak maxima are observed when relationship Eq. (4) holds for each ion type. The Uci value is a function of Ud, f(t ) and N . Experimental curves of Uci versus Ud for the ions of tested compounds are given on Fig. 3. Positive Uc for negative ions with negative Ud testifies that a (E /N ) /0 and mobility coefficient, K , grows with voltage, Ud. At the initial stage of this dependence Uci increases as the square of amplitude, Ud, then with the increase in Ud the extent to which Uci depends on Ud decreases and changes to linear. The rate of growth increases in a series: TNT B/TNB B/ DNT B/DNB B/MNT B/iodine. With the high values of Ud, Uci passes into saturation. For the ions of MNT and iodine the Uci (Ud) dependence is limited by the range: Ud //0.7 to /3.9 kV, since
Fig. 3. Experimental of Uci versus Ud for the ions of tested compounds.
I.A. Buryakov / Talanta 61 (2003) 369 /375
with Ud B//3.9 kV the current of these ion types is not detected. It is likely that a high amplitude of ion oscillations under the action of the dispersion field due to a high mobility coefficient value is responsible for this. The key varied parameter of voltage waveform, f(t), (Eq. (6)) that affects the Uci value is the t/T relationship. Under changes of the parameter in the range t /T /0.35 /0.4 the Uci value for the tested ion types has decreased to 3% from the maximum value observed with t/T /0.36. In changing the transport flow rate within the range Qtr /20/50 cm3 s 1 and the tested compound concentration in the range Ci /1 /(10 1 to 10 3)/Ci sat, where Ci sat, refers to the saturated vapor concentration, the variation of Uci did not exceed the measuring error bounds. Thus the Uci value for different compound, with other conditions being equal, is determined by the a (E /N ) dependence. 3.2. Determination of the a (E /N ) dependencies Coefficients a2 and a4 (Table 2) were calculated solving an inverse problem of Eq. (4) and using experimental data (Fig. 3) expressed in terms of strength by an equation: U /E /r /ln(r2/r1) with r /(r1/r2)/2. The values of coefficients f2n / 0.23, 0.13 and f2n1 /0.12, 0.11 with n/1, 2 and t/T /0.36 were determined by numeric calculation with an error of 9/2%. Such an error is due to error of the time interval measurement, dt . The values of a2n were calculated by means of polynomial approximation using a least-squares method (polynomial regression).
373
Fig. 4. Curves of mobility coefficient increment a versus parameter E /N .
Numeric values of standard deviations (sU) of theoretical curves, Uc(Ud), constructed by coefficients, a2n , from the theoretical points (Fig. 3) are given in the last column of Table 2. Fig. 4 shows theoretical curves, a (E /N ), build up by coefficients, a2n , with Eq. (1). The analogues behaviour of curves, a (E /N ) and Uc(Ud), (Fig. 3) testifies that over a wide range of the variations in the E /N parameter a and Uc are connected by a simple relation. This is displayed in Fig. 5 which gives the values for estimated coefficients, a , versus experimental relations, Ec/Ed (equivalent to Uc/Ud) for all tested ion types. A regression line is shown by a straight line. Over the turndown of parameter, E /N /25 /90 Td, points fall well on a straight line with a slope S /8.35 /1/0.12 and
Table 2 Numerical values of coefficients a2n and mean-square deviations sU of theoretical curves Uc(Ud) from experimental points Uc(Ud) Compound
a2 /10 6, Td 2
a4 /10 10, Td4
sU , V
Iodine p-MNT DNB DNT TNB TNT
11.9 10.0 8.2 7.7 5.7 4.4
/7.4 /6.0 /4.3 /4.2 /3.0 /2.7
0.07 0.12 0.17 0.16 0.12 0.09
Fig. 5. Values of estimated coefficients a versus experimental relations Ec/Ed (or Uc/Ud) and regression line a/8.35/Ec/Ed.
374
I.A. Buryakov / Talanta 61 (2003) 369 /375
standard deviation, sS /9/0.00041 (the number of points is B /86). From Eq. (4) S /1/f3, hence Eq. (4) with account of Eq. (5) for all tested compounds over the range of values E /N /25/90 Td can be represented as Uc a(Ed =N)Ud hf 3 i:
(7)
With E /N /90 Td proportionality between a and Ec/Ed is violated, and we have a strong point deviation from the regression line (shown as arrows).
3.3. Identification of nitrocompounds by the iodine peak location at a drift spectrum From Eq. (7) the relationship among the compensation voltages of the tested compound peaks in the range E /N /25/90 Td is equal to the relationship among the increments of their mobility coefficients: Uci =Ucj ai =aj :
(8)
This is apparent from Fig. 6 which gives experimental values for the relationship among the compensation voltages: kUi(Ud) /Uci (Ud)/Uc iodine(Ud) (points) and curves: kai (E /N )/ai (E /N )/ aiodine(E /N ) (continuous lines), i /ions of MNT,
Fig. 6. Increment coefficients reduced to increment coefficient of iodine: experimental values kUi (Ud) /Uci (Ud)/Uc iodine(Ud) */figures; dependencies kai (E /N ) /ai (E /N )/aiodine(E / N ) */continuous lines, i -ions of MNT, DNB, DNT, TNB and TNT.
Table 3 Average values of coefficients kai , kUi and mean-square deviations skU of calculated Uci from experimental Uci values Ratio
kai
kUi
s kU , V
MNT/iodine DNB/iodine DNT/iodine TNB/iodine TNT/iodine
0.83 0.72 0.66 0.48 0.37
0.85 0.73 0.67 0.49 0.38
0.07 0.1 0.15 0.12 0.08
DNB, DNT, TNB and TNT. One can see from the Figure that with E /N /25 Td coefficients are kai :/kUi and are hardly dependent on the dispersion voltage value. Table 3 gives average values of coefficients, kai , kUi , and standard deviations, skU , of calculated Uci from experimental Uci values. The fact that coefficients, kUi , are characteristic of the given pair of the compounds only and kUi does not depend on the amplitude and waveform of the dispersion voltage, atmospheric pressure substantially simplifies the identification of the drift spectrum peaks with a standard compound, eg. iodine. Having determined Uci of the iodine peak and knowing the kUi values for iodine, one can identify nitrocompound peaks independent of the given separation conditions and IMIS parameters. Since the dependence of the kUi coefficients on the separation region parameters (d or r1, r2; Qtr) is lacking, one can compare interlaboratory data. The limits of relative error in determination of coefficients, kUi , depend on how voltages, Uci and Uc iodine, are determined: by the spectra of the individual compounds or nitrocompound-iodine mixtures. Ion peaks had the same average values of Uci on recording spectra both of individual compounds and mixtures. Standard deviation of voltage, Uci , on recording individual compound spectra was sUc1 /9/1% within one day and sUc10 /9/2% within 10 days (the number of measurements of each compound was b /15/ day). When Uci was adjusted to the standard pressure (760 mm Hg), standard deviation, sUc1 :/sUc10, of the relationship among compensation voltages, kUi /Uci /Uc iodine, on detecting individual compounds was skU1 /9/1.4%. The standard deviation of the relationship among
I.A. Buryakov / Talanta 61 (2003) 369 /375
compensation voltages, kUi, on recording spectra of nitrocompound-iodine mixture was skU1 /9/ 0.5% within 1 day and skU10 /9/0.7% within 10 days.
[2]
[3]
4. Conclusion The position (voltage Uci ) of a definite ion type peak on a drift spectrum is a function of its mobility coefficient increment, ai , which in its turn depends on amplitude, Ud, dispersion voltage waveform, f(t), separation region parameters (d or r1, r2), atmospheric pressure value, N . The relationship among the compensation voltages of the tested compounds in the range E /N /25 /90 Td is equal to the relationship among their mobility coefficient increments, Uci /Ucj /ai /aj, and is almost independent of the above-mentioned factors. For the lack of this dependence qualitative analysis of trace constituents with a standard compound is simplified, and comparison of interlaboratory data is possible. Analysis of a sample mixed with a standard compound is preferred, since upon recording a drift spectrum of such a mixture standard deviations of the Uci /Uc iodine relationship are twice less than upon recording individual spectra of a sample and a standard compound (1.4 and 0.7%, respectively).
[4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
[14] [15] [16] [17] [18] [19]
References [1] B. Carnahan, S. Day, V. Kouznetsov, M. Matyjaszczyk, A. Tarassov, Proceeding of the 41th Annual ISA Analysis
[20] [21]
375
Division Symposium, Framingham, MA, 21 /24 April, 1996, p. 85. R.G. Ewing, Detection of chemical warfare materials with a high-field asymmetric waveform ion mobility spectrometer, Proceeding of the 8th International Conference on IMS, 1999, Buxton, UK. B. Ells, D.A. Barnett, K. Froese, R.W. Purves, S. Hrudey, R. Guevremont, Anal. Chem. 71 (1999) 4747. G.A. Eiceman, E.G. Nazarov, B. Tadjikov, R.A. Miller, Field Anal. Chem. Tech. 4 (2000) 297. I.A. Buryakov, Yu.N. Kolomiets, V.B. Luppu, J. Anal. Chem. 56 (2001) 336. Buryakov, Yu.N. Kolomietz, A.V. Bolotov, A.I. Vasin, Yu.N. Vlasov, J. Anal. Chem. 57 (2002) 606. R.W. Purves, R. Guevremont, S. Day, C.W. Pipich, M.S. Matyjaszczyk, Rev. Sci. Instrum. 69 (1999) 4094. B.L. Carnahan, A.S. Tarasov, US Patent 5,420,424, 1995. R.A. Miller, G.A. Eiceman, E.G. Nazarov, A.T. King, Sensors Actuators, B 67 (2000) 300. I.A. Buryakov, E.V. Krylov, E.G. Nazarov, U.Kh. Rasulev, Int. J. Mass. Spec. Ion. Proc. 128 (1993) 143. E.A. Mason, E.W. McDaniel, Transport Properties of Ions in Gas, Wiley, New York, 1988, p. 141. M.P. Gorshkov, USSR Patent 966583, 1982. B. Carnahan, S. Day, V. Kouznetsov, A. Tarassov, Proceeding of the 4th International Workshop on Ion Mobility Spectrometry, August 6 /9, 1995, Cambridge, UK. G.E. Spangler, Anal. Chem. 65 (1993) 3010. A.P. Snyder, W.M. Maswadeh, G.A. Eiceman, Y.-F. Wang, S.E. Bell, Anal. Chem. Acta 316 (1995) 1. G.E. Spangler, J.P. Carrico, D.N. Campbell, J. Test. Evalution, JTEVA 13 (1985) 234. K.A. Daum, D.A. Atkinson, R.G. Ewing, Int. J. IMS 4 (2001) 179. G.R. Asbury, J. Klasmeier, H.H. Hill, Talanta 50 (2000) 1291. K.A. Daum, D.A. Atkinson, R.G. Ewing, W.B. Knighton, E.P. Grimsrud, Talanta 54 (2001) 299. P. Kebarle, S. Chowdhury, Chem. Rev. 87 (1987) 531. G.W. Griffin, I. Dzidic, D.I. Carroll, R.N. Stillwell, E.C. Horning, Anal. Chem. 45 (1973) 1207.
Qualitative analysis of trace constituents by ion mobility increment spectrometer I.A. Buryakov The Design and Technological Institute of Instrument Engineering for Geophysics and Ecology (IDE), The Siberian Branch of RAS, Pr. Ak. Koptyuga 3, Block 6, 630090 Novosibirsk, Russia Received 23 September 2002; received in revised form 28 April 2003; accepted 29 April 2003
Abstract Ion mobility increment spectrometry (IMIS) is a high sensitive selective ionization technology for detection and identification of ultra-trace constituents, including toxic compounds, CW-agents, drugs and explosives in ambient air or liquid sample. Like an ion mobility spectrometry (IMS), this technology rests on sampling air containing a mixture of trace constituents, its ionization, spatial separation of produced ions and separated ions detection. Unlike IMS, ions of different types in IMIS are separated by ion mobility increment, a . Value a , is a function of the parameters: electric field strength and form, atmospheric pressure. To exclude the influence of these parameters on an a , the method of explosives identification by a standard compound was suggested. As a standard compound iodine was used. The relationship among the mobility coefficient increments equal to the relationship among the compensation voltage ai / aiodine /Ui /Uiodine is determined, where i are ions of 1,3-dinitrobenzene, 1,3,5-trinitrobenzene, p -mononitrotoluene, 2,4-dinitrotoluene and 2,4,6-trinitrotoluene This relationship is practically independent of the above mentioned parameters in the range 25 B/E /N B/90 Td. The limits of the relative error of this relationship are determined both from spectra of individual compounds and nitrocompound-iodine mixtures. # 2003 Elsevier B.V. All rights reserved. Keywords: Explosives; Ion mobility spectrometry; Atmospheric pressure ionization
1. Introduction Devices that combine high sensitivity, selectivity and capacity of compound identification are required to perform qualitative analysis of trace
E-mail address: [email protected] (I.A. Buryakov).
constituents. An ion mobility increment spectrometer (IMIS) can be used to detect and identify ultra-trace constituents: toxic compounds, CWagents, drugs and explosives in ambient air or liquid sample [1 /6]. Like an ion mobility spectrometer (IMS), the operation of IMIS rests on sampling air containing a mixture of trace constituents, its ionization, spatial separation of
0039-9140/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0039-9140(03)00305-9
370
I.A. Buryakov / Talanta 61 (2003) 369 /375
produced ions and separated ions detection. The names like ion non-linear drift spectrometer [5], high-field asymmetric waveform IMS [7], field ion spectrometer [8], radio-frequency-based ion-mobility analyzer [9] are used as a synonym for IMIS. IMIS differs from IMS in that ions of different types are separated in IMIS by ion mobility increment that depends on electric field strength [10]. Ion drift velocity, V , caused by an action of electric field (E ) is [11]: V K(E)E; K(E=N)K0 (1a(E=N)) 2n X E K0 1 a2n ; N n1
(1)
where K0 (cm2 (V cm) 1) is the mobility coefficient in a weak field (E /N B/6 ‘townsends’ or Td, where 1 Td /1 /1017 V cm2), N is neutral gas density, a (E /N ) is a normalized function which describes the electric field dependence of the mobility */the mobility coefficient increment, a2n are decomposition coefficients, n ]/1 is an integer. Under the action of periodic alternating asymmetric waveform field, Ed(t) /Edf (t ) (Ed is an amplitude, f(t) is a form of the field), that meets the conditions [12]:
g
tT
f (t)dt0; t
1 T
g
(2)
tT
f
2n1
(t)dthf
2n1
i"0;
t
ions executing fast oscillatory motions with period, T , drift with velocity Vi characteristic for each i-th component proportional to ai (E /N ) (broken brackets mean averaging over a period). Should the ion drift of the i -th component be compensated by constant electric field, Eci , the ion velocity will be zero: hVi ihK0 (1ai )Ed f (t)Eci i0
(3)
The compensation field of the i-th component is expressed by substituting decomposition of coefficient, a , from Eq. (1) into Eq. (3) and using condition (Eq. (2)) and approximation (Ed/ Ec)n :/End /nEn1 Ec, with jEdj/jEcj [10]: d
X 2n Ed 2n1 ai2n hf i Eci : Ed N n1 2n X E 1 (2n1)ai2n d hf 2n1 i : N n1
=
(4)
The ions for which Vi /0 are transported to a collector with a carrier-gas, the other ions leave a separation region and recombine. With Ec changing, spectrum of a mixture of ions of all types is recorded. A separation region can be of planar or axial symmetry depending on which electrodes, made of two plane-parallel plates [4,10] or two coaxial cylinders [1,3,5], are used. Electric field strength is equal to: E U=d;
E(r)U=(r ln(r2 =r1 ));
(5)
respectively, where U is the potential difference across the electrodes, d is a distance between the plane /parallel plates, r is a radial coordinate, r1 and r2 are radiuses of the inner and external cylinders, respectively. From Eqs. (4) and (5) one can see that compensation voltage, Uci , is one-of-a-kind for each ion type and is determined by coefficients, a2ni , that depend on temperature, T , drift gas composition (e.g. the presence of water vapors (CH2O) and etc.). Voltage, Uci is a function of the amplitude of asymmetric waveform dispersion voltage, Ud, voltage waveform, f(t), density, N , distance, d , or radiuses, r1 and r2. Problems in comparing the obtained experimental results arise if devices with different d or r1, r2 are used, since one and the same ion types are recorded at spectra with different Uc values (e.g. TNT ions at [5] and [13]). Fluctuations of T , CH2O, Ud, f (t ) and N cause fluctuations of Uc, and this complicates the analyzed compound identification. To enhance reproducibility and certainty of analyte identification of a spectrometric device any standard compound can be used for a calibration [14]. As a standard compound any available easy to handle compounds can be used. For example, in [15] with the IMS in positive mode 2,4-lutidine was used as a standard chemical compound. Trace constituents of 1,3-dinitrobenzene, 1,3,5trinitrobenzene, p-mononitrotoluene, 2,4-dinitro-
I.A. Buryakov / Talanta 61 (2003) 369 /375
toluene and 2,4,6-trinitrotoluene in air were studied. The iodine was used as a standard compound. The a (E /N ) dependencies of a mobility coefficient increment for negative ions of these compounds are determined. An agreement between the relationship among the a (E /N ) dependencies for different ion types and the relationship among their compensation voltages is shown. The relationship among the compensation voltages can be used for the qualitative analysis of the trace constituents. Standard compensation voltage deviations are determined both for individual compounds and nitrocompound-iodine mixtures.
371
and r2 /1.8 cm in diameter, the chamber is purged with a carrier-gas (dry air with a water vapor concentration 5/100 ppm, flow rate Qtr /20 /50 cm3 s 1, temperature tg /259/1 8C, N /(2.439/ 0.07) /1019 cm 3), an electrode (9/200 V), for transporting ions from the ionization chamber into the separation chamber, an ion collector, an electrometer (width of a zero line of the electrometer noise Wz /2 /1014 A), a compensation voltage generator. A dispersion voltage generator used is developed by IDE. The generator has the following parameters: Voltage form (Fig. 1)
2. Experimental
f (t) (sin[p(tmT )=t]2t=pT)=(12t=pT); with mT 5t5(mT t); f (t)
2.1. Instrumentation
(2t=pT)=(12t=pT); with (mT t)5 t5 (m1)T;
Block diagram of IMIS is given on Fig. 1. The analytical unit of the device is developed by the Mine Safety Appliances Company [8] and comprises a heated ionization chamber (sampling flow rate, Qs /1.2 cm3 s 1, b-source 63Ni, temperature, ti /150 8C), an ion separation chamber formed by two electrodes which are coaxial cylinders, r1 /1.4
(m ]/0 is an integer); high-voltage pulse duration, t/2.7 ms, period T /6.8 /7.8 ms, dispersion voltage amplitude, Ud //0.7 to /5 kV. We used an oscillograph to control the form and amplitude of Ud(t) voltage. Absolute error in compensation voltage measurement is dUc /9/0.05 V; relative measuring error of the compensation voltage is dUc /9/1%, asymmetric waveform voltage amplitude is dUd /9/7% and time intervals is dt /9/ 7%.
(6)
2.2. Chemicals
Fig. 1. Block diagram of experimental unit. (1) ionization chamber, (2) ion separation chamber, (3) electrode for transporting ions, (4) dispersion voltage generator, (5) compensation voltage generator, (6) ion collector, (7) electrometer; Qs, air / vapor mixture sample; Qin, purified air inlet flow; Qtr, carriergas; Qout, outlet flow.
Samples of 1,3-dinitrobenzene (DNB), 1,3,5trinitrobenzene (TNB), p-mononitrotoluene (MNT), 2,4-dinitrotoluene (DNT) and 2,4,6-trinitrotoluene (TNT), crystalline iodine in highest purity were obtained from the Research Institute of Special Technics and Communication (Novosibirsk, Russia). To obtain air /vapor mixtures of nitrocompounds and/or iodine, we mixed the flows of purified air and purified air passed through quartz tubes (¥ 0.3 cm, 20 cm in length) the inside of which were covered with tested compounds. Table 1 shows the main types of ion-molecule reactions into which the molecules of tested compounds enter being ionized by b-source at
I.A. Buryakov / Talanta 61 (2003) 369 /375
372
Table 1 The main types of ion-molecule reactions, mass /charge ratio (m /z ) and types of ions produced Compound
Reaction
m /z
Ion
Refs.
p-MNT DNT TNT
136 181 226
(M/H)
[16,17]
Proton abstraction
TNB DNB
Associative electron capture
213 168
M
[18,19] [20]
Iodinea
Dissociative electron capture
127
M
[21]
M is molecule or atom of a compound, H is hydrogen atom. a b-Ionization in N2.
atmospheric pressure, mass /charge ratio (m /z ) and types of ions produced (M is molecule or atom of a compound, H is hydrogen atom). In the last column one can find references to the papers where these ions were identified by mass-spectrometry.
3. Results and discussion 3.1. Dependence of Uc on the separation conditions and IMIS parameters The IMIS spectrum is the compensation voltage dependence of the ion current (I(Uc)), each ion type is recorded as an ion peak. Fig. 2 shows a spectrum for a mixture of vapors: iodine, MNT, DNB, DNT, TNB and TNT in air taken with
Fig. 2. Spectrum of a vapors mixture of iodine, MNT, DNB, DNT, TNB, TNT in air. Ud /3900 V.
Ud /3900 V. The Uci values corresponding to the peak maxima are observed when relationship Eq. (4) holds for each ion type. The Uci value is a function of Ud, f(t ) and N . Experimental curves of Uci versus Ud for the ions of tested compounds are given on Fig. 3. Positive Uc for negative ions with negative Ud testifies that a (E /N ) /0 and mobility coefficient, K , grows with voltage, Ud. At the initial stage of this dependence Uci increases as the square of amplitude, Ud, then with the increase in Ud the extent to which Uci depends on Ud decreases and changes to linear. The rate of growth increases in a series: TNT B/TNB B/ DNT B/DNB B/MNT B/iodine. With the high values of Ud, Uci passes into saturation. For the ions of MNT and iodine the Uci (Ud) dependence is limited by the range: Ud //0.7 to /3.9 kV, since
Fig. 3. Experimental of Uci versus Ud for the ions of tested compounds.
I.A. Buryakov / Talanta 61 (2003) 369 /375
with Ud B//3.9 kV the current of these ion types is not detected. It is likely that a high amplitude of ion oscillations under the action of the dispersion field due to a high mobility coefficient value is responsible for this. The key varied parameter of voltage waveform, f(t), (Eq. (6)) that affects the Uci value is the t/T relationship. Under changes of the parameter in the range t /T /0.35 /0.4 the Uci value for the tested ion types has decreased to 3% from the maximum value observed with t/T /0.36. In changing the transport flow rate within the range Qtr /20/50 cm3 s 1 and the tested compound concentration in the range Ci /1 /(10 1 to 10 3)/Ci sat, where Ci sat, refers to the saturated vapor concentration, the variation of Uci did not exceed the measuring error bounds. Thus the Uci value for different compound, with other conditions being equal, is determined by the a (E /N ) dependence. 3.2. Determination of the a (E /N ) dependencies Coefficients a2 and a4 (Table 2) were calculated solving an inverse problem of Eq. (4) and using experimental data (Fig. 3) expressed in terms of strength by an equation: U /E /r /ln(r2/r1) with r /(r1/r2)/2. The values of coefficients f2n / 0.23, 0.13 and f2n1 /0.12, 0.11 with n/1, 2 and t/T /0.36 were determined by numeric calculation with an error of 9/2%. Such an error is due to error of the time interval measurement, dt . The values of a2n were calculated by means of polynomial approximation using a least-squares method (polynomial regression).
373
Fig. 4. Curves of mobility coefficient increment a versus parameter E /N .
Numeric values of standard deviations (sU) of theoretical curves, Uc(Ud), constructed by coefficients, a2n , from the theoretical points (Fig. 3) are given in the last column of Table 2. Fig. 4 shows theoretical curves, a (E /N ), build up by coefficients, a2n , with Eq. (1). The analogues behaviour of curves, a (E /N ) and Uc(Ud), (Fig. 3) testifies that over a wide range of the variations in the E /N parameter a and Uc are connected by a simple relation. This is displayed in Fig. 5 which gives the values for estimated coefficients, a , versus experimental relations, Ec/Ed (equivalent to Uc/Ud) for all tested ion types. A regression line is shown by a straight line. Over the turndown of parameter, E /N /25 /90 Td, points fall well on a straight line with a slope S /8.35 /1/0.12 and
Table 2 Numerical values of coefficients a2n and mean-square deviations sU of theoretical curves Uc(Ud) from experimental points Uc(Ud) Compound
a2 /10 6, Td 2
a4 /10 10, Td4
sU , V
Iodine p-MNT DNB DNT TNB TNT
11.9 10.0 8.2 7.7 5.7 4.4
/7.4 /6.0 /4.3 /4.2 /3.0 /2.7
0.07 0.12 0.17 0.16 0.12 0.09
Fig. 5. Values of estimated coefficients a versus experimental relations Ec/Ed (or Uc/Ud) and regression line a/8.35/Ec/Ed.
374
I.A. Buryakov / Talanta 61 (2003) 369 /375
standard deviation, sS /9/0.00041 (the number of points is B /86). From Eq. (4) S /1/f3, hence Eq. (4) with account of Eq. (5) for all tested compounds over the range of values E /N /25/90 Td can be represented as Uc a(Ed =N)Ud hf 3 i:
(7)
With E /N /90 Td proportionality between a and Ec/Ed is violated, and we have a strong point deviation from the regression line (shown as arrows).
3.3. Identification of nitrocompounds by the iodine peak location at a drift spectrum From Eq. (7) the relationship among the compensation voltages of the tested compound peaks in the range E /N /25/90 Td is equal to the relationship among the increments of their mobility coefficients: Uci =Ucj ai =aj :
(8)
This is apparent from Fig. 6 which gives experimental values for the relationship among the compensation voltages: kUi(Ud) /Uci (Ud)/Uc iodine(Ud) (points) and curves: kai (E /N )/ai (E /N )/ aiodine(E /N ) (continuous lines), i /ions of MNT,
Fig. 6. Increment coefficients reduced to increment coefficient of iodine: experimental values kUi (Ud) /Uci (Ud)/Uc iodine(Ud) */figures; dependencies kai (E /N ) /ai (E /N )/aiodine(E / N ) */continuous lines, i -ions of MNT, DNB, DNT, TNB and TNT.
Table 3 Average values of coefficients kai , kUi and mean-square deviations skU of calculated Uci from experimental Uci values Ratio
kai
kUi
s kU , V
MNT/iodine DNB/iodine DNT/iodine TNB/iodine TNT/iodine
0.83 0.72 0.66 0.48 0.37
0.85 0.73 0.67 0.49 0.38
0.07 0.1 0.15 0.12 0.08
DNB, DNT, TNB and TNT. One can see from the Figure that with E /N /25 Td coefficients are kai :/kUi and are hardly dependent on the dispersion voltage value. Table 3 gives average values of coefficients, kai , kUi , and standard deviations, skU , of calculated Uci from experimental Uci values. The fact that coefficients, kUi , are characteristic of the given pair of the compounds only and kUi does not depend on the amplitude and waveform of the dispersion voltage, atmospheric pressure substantially simplifies the identification of the drift spectrum peaks with a standard compound, eg. iodine. Having determined Uci of the iodine peak and knowing the kUi values for iodine, one can identify nitrocompound peaks independent of the given separation conditions and IMIS parameters. Since the dependence of the kUi coefficients on the separation region parameters (d or r1, r2; Qtr) is lacking, one can compare interlaboratory data. The limits of relative error in determination of coefficients, kUi , depend on how voltages, Uci and Uc iodine, are determined: by the spectra of the individual compounds or nitrocompound-iodine mixtures. Ion peaks had the same average values of Uci on recording spectra both of individual compounds and mixtures. Standard deviation of voltage, Uci , on recording individual compound spectra was sUc1 /9/1% within one day and sUc10 /9/2% within 10 days (the number of measurements of each compound was b /15/ day). When Uci was adjusted to the standard pressure (760 mm Hg), standard deviation, sUc1 :/sUc10, of the relationship among compensation voltages, kUi /Uci /Uc iodine, on detecting individual compounds was skU1 /9/1.4%. The standard deviation of the relationship among
I.A. Buryakov / Talanta 61 (2003) 369 /375
compensation voltages, kUi, on recording spectra of nitrocompound-iodine mixture was skU1 /9/ 0.5% within 1 day and skU10 /9/0.7% within 10 days.
[2]
[3]
4. Conclusion The position (voltage Uci ) of a definite ion type peak on a drift spectrum is a function of its mobility coefficient increment, ai , which in its turn depends on amplitude, Ud, dispersion voltage waveform, f(t), separation region parameters (d or r1, r2), atmospheric pressure value, N . The relationship among the compensation voltages of the tested compounds in the range E /N /25 /90 Td is equal to the relationship among their mobility coefficient increments, Uci /Ucj /ai /aj, and is almost independent of the above-mentioned factors. For the lack of this dependence qualitative analysis of trace constituents with a standard compound is simplified, and comparison of interlaboratory data is possible. Analysis of a sample mixed with a standard compound is preferred, since upon recording a drift spectrum of such a mixture standard deviations of the Uci /Uc iodine relationship are twice less than upon recording individual spectra of a sample and a standard compound (1.4 and 0.7%, respectively).
[4] [5] [6] [7] [8] [9] [10] [11] [12] [13]
[14] [15] [16] [17] [18] [19]
References [1] B. Carnahan, S. Day, V. Kouznetsov, M. Matyjaszczyk, A. Tarassov, Proceeding of the 41th Annual ISA Analysis
[20] [21]
375
Division Symposium, Framingham, MA, 21 /24 April, 1996, p. 85. R.G. Ewing, Detection of chemical warfare materials with a high-field asymmetric waveform ion mobility spectrometer, Proceeding of the 8th International Conference on IMS, 1999, Buxton, UK. B. Ells, D.A. Barnett, K. Froese, R.W. Purves, S. Hrudey, R. Guevremont, Anal. Chem. 71 (1999) 4747. G.A. Eiceman, E.G. Nazarov, B. Tadjikov, R.A. Miller, Field Anal. Chem. Tech. 4 (2000) 297. I.A. Buryakov, Yu.N. Kolomiets, V.B. Luppu, J. Anal. Chem. 56 (2001) 336. Buryakov, Yu.N. Kolomietz, A.V. Bolotov, A.I. Vasin, Yu.N. Vlasov, J. Anal. Chem. 57 (2002) 606. R.W. Purves, R. Guevremont, S. Day, C.W. Pipich, M.S. Matyjaszczyk, Rev. Sci. Instrum. 69 (1999) 4094. B.L. Carnahan, A.S. Tarasov, US Patent 5,420,424, 1995. R.A. Miller, G.A. Eiceman, E.G. Nazarov, A.T. King, Sensors Actuators, B 67 (2000) 300. I.A. Buryakov, E.V. Krylov, E.G. Nazarov, U.Kh. Rasulev, Int. J. Mass. Spec. Ion. Proc. 128 (1993) 143. E.A. Mason, E.W. McDaniel, Transport Properties of Ions in Gas, Wiley, New York, 1988, p. 141. M.P. Gorshkov, USSR Patent 966583, 1982. B. Carnahan, S. Day, V. Kouznetsov, A. Tarassov, Proceeding of the 4th International Workshop on Ion Mobility Spectrometry, August 6 /9, 1995, Cambridge, UK. G.E. Spangler, Anal. Chem. 65 (1993) 3010. A.P. Snyder, W.M. Maswadeh, G.A. Eiceman, Y.-F. Wang, S.E. Bell, Anal. Chem. Acta 316 (1995) 1. G.E. Spangler, J.P. Carrico, D.N. Campbell, J. Test. Evalution, JTEVA 13 (1985) 234. K.A. Daum, D.A. Atkinson, R.G. Ewing, Int. J. IMS 4 (2001) 179. G.R. Asbury, J. Klasmeier, H.H. Hill, Talanta 50 (2000) 1291. K.A. Daum, D.A. Atkinson, R.G. Ewing, W.B. Knighton, E.P. Grimsrud, Talanta 54 (2001) 299. P. Kebarle, S. Chowdhury, Chem. Rev. 87 (1987) 531. G.W. Griffin, I. Dzidic, D.I. Carroll, R.N. Stillwell, E.C. Horning, Anal. Chem. 45 (1973) 1207.