- Email: [email protected]
Determination of ammonia in ethylene using ion mobility spectrometry
Talanta ELSEVIER
Talanta 45 (1997) 19 23
Determination of ammonia in ethylene using ion mobility spectrometry John H. Cross a.,, Thomas F. Limero ~, James L. Lane b, Fusheng Wang c KRUG Life Sciences, Suite 120, 1290 Hercules Drive, Houston, TX 77058, USA b Shell Deer Park Chemical Plant, Highway 225, Deer Park, T J( 77536, USA Shell Oil Products Company, Westhollow Technology Center, 3333 Highway 6, Houston, TX 77082, USA
Received 9 October 1996; received in revised form 18 March 1997; accepted 18 March 1997
Abstract
A simple procedure to analyze ammonia in ethylene by ion mobility spectrometry is described. The spectrometer is operated with a silane polymer membrane, 63Ni ion source, H + (H20),, reactant ion, and nitrogen drift and source gas. Ethylene containing parts per billion (ppb) (v/v) concentrations of ammonia is pulled across the membrane and diffuses into the spectrometer. Preconcentration or preseparation is unnecessary, because the ethylene in the spectrometer has no noticeable effect on the analytical results. Ethylene does not polymerize in the radioactive source. Ethylene's flammability is negated by the nitrogen inside the spectrometer. Response to ammonia concentrations between 200 ppb and 1.5 ppm is near linear, and a detection limit of 25 ppb is calculated. © 1997 Elsevier Science B.V. Keywords: Ammonia; Analysis; Detection; Determination; Ethylene; Ion mobility spectrometry (IMS)
I. Introduction
Ion mobility spectrometry (IMS) has the sensitivity, selectivity, robustness and operational simplicity necessary for performing analyses in the field. Field monitoring of drugs, chemical warfare agents, and explosives among others are reviewed in a recently published m o n o g r a p h [1], which also provides background information on IMS. The qualities that make IMS attractive for these field analyses are also important considerations for
* Corresponding author. Tel,: + 1 281 4838800; fax: + 1 281 4833058; e-mail: [email protected].
on-line chemical process analysis, and a patent has been issued for an IMS-based process monitor for N H 3 [2]. The patent is licensed to Molecular Analytics, formerly Environmental Technology Group, which has made a process ammonia analyzer commercially available. The major difference between field analyses and process analyses is the sample medium. In field analyses, the samples consist of air drawn from the atmosphere. In process analyses, the samples contain significant fractions of organic chemicals drawn from process equipment such as reactors, pipelines, or stacks. The most specific reports in the literature about using IMS for media other than air regard liquid
0039-9140/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0039-91 40(97 )00095-7
20
J.H. Cross et al./Talanta 45 (1997) 19 23
chromatography or supercritical fluid chromatography (SFC). For IMS with liquids, the usual, simple radioactive ionization source was abandoned in favor of a more complex corona discharge ionization source [3]. When IMS is used as a SFC detector, the mobility spectra can show drift time effects when the mobile phase is CO2 or CO2 with modifiers [4]. Discussions of media effects on IMS can be found in Eiceman and Karpas [5] and St. Louis and Hill [6]. Ion mobility spectrometers operating in the positive mode detect compounds by transfer of a proton from a reactant ion peak (RIP) to the analyte. The RIP in air is the H+(H20),, ion; dopants can be added to form a different RIP, usually to increase selectivity. Ammonia is detected at parts per billion (ppb) concentrations, because ammonia has a high affinity for protons. Ethylene has a low affinity for protons, but the pi electrons would be expected to interact weakly with them. Thus the effect of large concentrations of ethylene on ammonia detection by IMS are not readily predictable. This paper reports experiments we carried out to assess how well ammonia could be detected in the presence of ethylene. Three considerations are important for analyzing ammonia in the presence of large concentrations of ethylene. First, ethylene is present in the IMS cell, because a silane polymer membrane in the spectrometer inlet allows ethylene diffusion into the cell. Although the membrane prevents complete mixing and IMS is not very sensitive to ethylene, the quantity passing through the membrane far exceeds the ppb concentrations usually encountered by IMS instruments. Also once in the IMS cell, poor mixing could lead to anomalous effects. For simplicity and accuracy, adding a step to separate the ethylene prior to analysis is undesirable. Second, ethylene's flammability poses a safety issue in a spectrometer as the electrical elements in the cell present a small potential for sparking. Third, radiation can initiate ethylene polymerization [7]. Many IMSs, including the one used here, have 63Ni ionization sources. The beta radiation produced by the nickel source would be unlikely to initiate global polymerization, unlike the deep penetrating (e.g., gamma) radiation reported to be used. Localized polymerization, how-
ever, could foul the source, and greatly reduce the spectrometer's performance. To minimize the potential for fouling and to improve selectivity for ammonia, we began with the nonanone-doping method developed by Eiceman et al. [8] for measuring ammonia in air.
2. Experimental section We used an E-IMS (Graseby Ionics, Watford, UK), a spectrometer that allows operating conditions to be varied easily. The spectrometer has a polydimethylsilane membrane inlet and chambers for introducing dopants into the drift or source gas. 5-Nonanone (Aldrich Chemical Company, Milwaukee, WI) was doped into the drift gas from an open, test-tube-shaped, 2-ml volumetric flask initially containing 50 ~tl liquid. Another dopant, acetone, was doped from a Dynacal permeation device (VICI Metronics, Santa Clara, CA; 42 ng rain J at 90°C) at ambient temperature. To eliminate the possibility of combustion in the cell, the drift (220 ml rain J) and source (80 ml rain-~) gas was nitrogen; removing oxygen broke the 'fire triangle' of oxidant, fuel and ignition source. The instrument was operated in a hood to remove and dilute exhaust streams safely. The cell temperature was 25°C, and the cell pressure was set 17 mm Hg below ambient. Mixtures of ammonia in air at 0.4 (5.404 1 rain-~) and 1.5 ppm (1.468 1 rain ~) were generated from a permeation device (1.510 gg rain - ~ at 30°C) in a 585 Precision Gas Standards Generator (both from Kin-Tek, LaMarque, TX). The mixtures were delivered through a PTFE tube terminating in a cup placed over the inlet nozzle. Ethylene with 1.5 ppm NH 3 was prepared by Alphagaz (Walnut Creek, CA) and was used without further analysis. It was dispensed to the EIMS through stainless steel tubing with needle valves that could divert a portion through 3A molecular sieve to remove the ammonia. This arrangement allowed coarse control of the ammonia concentration between 0 and 1.5 ppm. At the E-IMS inlet, a tee split the flow between the inlet and an exhaust tube open to the atmosphere. One side of the tee was connected to the inlet with a
21
J.H. Cross et al. Talanta 45 (1997) 19 23
plastic tube that slipped firmly into the bore of the inlet nozzle, which led to the membrane. A p u m p on the inlet pulled in the sample across the membrane at roughly 800 ml r a i n - ' Ethylene flow was increased until a flowmeter on the exhaust tube showed a slight positive flow. Waveforms from the E-IMS were averaged and displayed using Advanced Signal Processing hardware and software (Graseby Ionics). Spectra normally consisted of 32 averages of 640 points collected at 30 kHz after a 2-ms delay. Gating pulses 180 I~S long occurred at the rate of 30 Hz. A m m o n i a concentrations were measured from peak heights taken between two cursors set at nearly the same absolute baseline value. The height of the H + ( H 2 0 ) , , ion peak or protonated nonanone ion peak in air was measured similarly.
3. Results The spectrum of 1.5 p p m ammonia in ethylene in the nonanone-doped spectrometer (Fig. 1) resembled the pattern described by Eiceman et al. [8]. The only indication of ethylene's presence was a slight enlargement o f the usual shoulder upfield of the RIP. A m m o n i a appeared clearly ( S / N = 12.5) at a drift time much longer than that of the RIP. The long drift time probably resulted from the size of the ion cluster, since Eiceman et al, [8] found an a m m o n i u m ion cluster containing two nonanone molecules. Sensitivity to ammonia, however, was unacceptably low. The peak height
NonanoneRIP~ Ammonia /, ......
~
' ~
' ' ~l
l ' ~l
'
1~01,21,41~61'8
~0
DRIFTTIME(ms) Fig. l. Spectrum of ammonia in ethylene with 5-nonanone doping. To define the ammonia peak more clearly, 128 waveforms were averaged.
Table 1 Responses to ammonia in ethylene or air in the undoped spectrometer Ammonia conc. 0.0 (v/v ppm in ethylene) 0.1 0.5 1.5 0.4 (ppm in air) 1.5
Peak height 2 _+1 (mVP~ 59+_6 165 + 4 524 + 5 1319 _+5 2178 +_2
~'Average_+standard deviation of five sequentially collected spectra. for 1.5 p p m was 24 mV (Table 1), and the R I P showed no depletion. When the spectrometer was doped with acetone, two broad peaks due to the presence of ethylene were noted. One of these peaks had the same drift time as the ammonia peak, and no further experiments were conducted with acetone doping. But while the preceding result was being established, the spectrometer was exposed repeatedly to ethylene; since no consistent diminution of the RIP after exposure was observed, we concluded the source was not fouling from ethylene polymerization. Exposing the undoped, 'water chemistry' [H + (H20),, reactant ion peak] spectrometer to ethylene produced a broad band of partially resolved peaks (Fig. 2(A)). A m m o n i a appeared as a sharp peak on the small plateau at the furthest upfield part of the band (Fig. 2(B) and (C), drift time = 6.305 ms). The drift time was shorter than that of the nonanone-doped system, because the a m m o n i u m ion cluster containing water was presumably smaller. A m m o n i a was detected at 100 ppb (Fig. 2(B), S / N = 12). At 1.5 ppm, the peak remained narrow and well resolved (Fig. 2(C)). The relationship among the four concentrations measured (including 0 ppm) was nearly linear (Table 1). Exposing the undoped E-IMS to ammonia in air produced stronger responses (Table 1) at a drift time of 6.205 ms ~. The difference of 0.1 ms between drift times in air and nitrogen is a combination of the time resolution, 0.033 ms, in
J.H. Cross et al./Talanta 45 (1997) 19-23
22
During undoped experiments, the E-IMS was exposed to ethylene for numerous 5-10 min periods and one period exceeding 30 min. Over the entire period of experimentation, the RIP again showed no diminution after exposure.
A
B
i
=
i
i
I
C
~
0
'
'
'
I
2
'
'
'
I
'
4
'
'
I
6
'
" 1 ' ' '
8
I
10
'
'
'
I
12
'
' ' 1 ' '
14
*1
16
DRIFT TIME (ms) Fig. 2. Spectrum of ammonia in ethylene without a dopant. The vertical pairs of lines bracket the drift time of the ammonia peak. The other peaks are assumed to be clusters of ethylene and the reactant ion peak. (A) 0 ppm; (B) 0.1 ppm; (C) 1.5 ppm.
4. Discussion
The most important reactions in positive-mode IMS are the proton transfers governed by the relative proton affinities of the RIP and other chemicals in the reaction region. The essence of this study was to find a set of conditions that would favor proton transfer to ammonia in the presence of expected large concentrations of ethylene. Nonanone doping minimizes the effect of ethylene as indicated by the similarity to previous work [8]. Using an open container gives a nonanone concentration higher than normally used for doping in IMS. The IMS response, however, is insufficient to detect ppb concentrations of ammonia, probably because the nonanone competes too well with ammonia for protons. Undoped experiments, in contrast, produce spectra with broad, poorly resolved peaks that suggest formation of numerous weak ethylene-containing clusters undergoing exchange in the drift tube. Since ethylene has a proton affinity lower than water, formation of these ions from the H + (H20) . RIP indicates (by LeChatelier's principle) high concentrations of ethylene in the drift tube. Nonetheless, ammonia, which has a proton affinity higher than water, competes successfully for protons. In addition, the short drift time of the ammonia ion places it in a part of the spectrum free from interference by the ethylene related ions. Thus, analysis of ppb concentrations of ammonia is possible. The nearly linear response with concentration (Table 1) shows that the upper limit of sensitivity is greater than 1.5 ppm, because an IMS response is exponential, reaching a plateau at the upper limit of sensitivity. Extrapolation of the signal-tonoise ratio from 100 ppb suggests that the lower limit of detection is 25 ppb. The lower response to ammonia in ethylene versus ammonia in air suggests that competing for protons with high concentrations of ethylene in the cell does reduce the response. The reduction is much less than that caused by nonanone with its much higher proton affinity. The drift time of ammonia in air was the same as that for ammonia in the mixture of nitrogen
J.H. Cross et al. /Talanta 45 (1997) 19-23
and ethylene in the IMS cell. The mobility equation [9] K = 3q/16N(2rr/#kT)l"2(1 + ~)/f~D(T)
contains two variables that are coincidentally the same for ethylene and nitrogen. The reduced mass, /t, is m M / ( m + M ) where m is the mass of the ion and M is the mass of the neutral drift gas, 28.0 for both ethylene and nitrogen. N, the number density (molecules/unit volume), is almost identical for nitrogen and ethylene, because the molecular weights and density (1.250 g 1 J at STP) [10] are almost equal. These coincidences minimize the effects of operating the IMS with ethylene inside the cell. With another hydrocarbon, the drift time of the a m m o n i a ion would likely depend on the hydrocarbon concentration in the drift tube. If protons in the ethylene ions are bound loosely to the pi electrons, they could initiate cationic polymerization, which would foul the spectrometer. We noted no cumulative decrease in the R I P after repeated exposures, indicating that the source did not foul appreciably. In conclusion, analysis of ammonia in ethylene by IMS appears to be relatively simple. No preseparation or preconcentration is required, because crucial mobility parameters for ethylene and nitrogen are equal. The IMS requires no dopant and appears to function safely in spite of ethylene's chemical reactivity.
23
Acknowledgements The authors acknowledge the use of equipment and facilities at NASA/Johnson Space Center, Houston, Texas, to conduct this work as a dual technology utilization effort. We also thank Patricia A. Inners for generating the figures from the data.
References [1] G.A. Eiceman, Z. Karpas, Ion Mobility Spectrometry, CRC Press, Boca Raton, FL, 1994, Ch. 6. [2] A.T. Bacon, US Patent No. 5,234,838, 1993. [3] C.B. Shumate, H.H. Hill Jr., Anal. Chem. 6t (1989) 601. [4] M.A. Morrissey, H.M. Widmer, J. Chromatogr. 552 (1991) 551. [5] G.A. Eiceman, Z. Karpas, Ion Mobility Spectrometry, CRC Press, Boca Raton, FL~ 1994, Ch. 5. [6] R.H. St. Louis, H,H. Hill Jr., CRC Crit. Rev. Anal. Chem. 21 (1990) 321. [7] R.A.V. Raft, Ethylene polymers, in: Encyclopedia of Polymer Science and Technology, vol. 6, Wiley, New York, 1967, p. 283. [8] G.A. Eiceman, M.R. Salazar, M.R. Rodriguez, T.F. Limero, S.W. Beck, J.H. Cross, R. Young, J.T. James, Anal. Chem. 65 (1993) 1696. [9] E.A. Mason, Ion mobility: its role in plasma chromatography, in: T.W. Carr (Ed.), Plasma Chromatography, Plenum, New York, 1984, p. 50. [10] S. Budavari (Ed.), The Merck Index, llth ed., Merck, Rahway, N J, 1989, pp. 597 598 and 1044.
Talanta 45 (1997) 19 23
Determination of ammonia in ethylene using ion mobility spectrometry John H. Cross a.,, Thomas F. Limero ~, James L. Lane b, Fusheng Wang c KRUG Life Sciences, Suite 120, 1290 Hercules Drive, Houston, TX 77058, USA b Shell Deer Park Chemical Plant, Highway 225, Deer Park, T J( 77536, USA Shell Oil Products Company, Westhollow Technology Center, 3333 Highway 6, Houston, TX 77082, USA
Received 9 October 1996; received in revised form 18 March 1997; accepted 18 March 1997
Abstract
A simple procedure to analyze ammonia in ethylene by ion mobility spectrometry is described. The spectrometer is operated with a silane polymer membrane, 63Ni ion source, H + (H20),, reactant ion, and nitrogen drift and source gas. Ethylene containing parts per billion (ppb) (v/v) concentrations of ammonia is pulled across the membrane and diffuses into the spectrometer. Preconcentration or preseparation is unnecessary, because the ethylene in the spectrometer has no noticeable effect on the analytical results. Ethylene does not polymerize in the radioactive source. Ethylene's flammability is negated by the nitrogen inside the spectrometer. Response to ammonia concentrations between 200 ppb and 1.5 ppm is near linear, and a detection limit of 25 ppb is calculated. © 1997 Elsevier Science B.V. Keywords: Ammonia; Analysis; Detection; Determination; Ethylene; Ion mobility spectrometry (IMS)
I. Introduction
Ion mobility spectrometry (IMS) has the sensitivity, selectivity, robustness and operational simplicity necessary for performing analyses in the field. Field monitoring of drugs, chemical warfare agents, and explosives among others are reviewed in a recently published m o n o g r a p h [1], which also provides background information on IMS. The qualities that make IMS attractive for these field analyses are also important considerations for
* Corresponding author. Tel,: + 1 281 4838800; fax: + 1 281 4833058; e-mail: [email protected].
on-line chemical process analysis, and a patent has been issued for an IMS-based process monitor for N H 3 [2]. The patent is licensed to Molecular Analytics, formerly Environmental Technology Group, which has made a process ammonia analyzer commercially available. The major difference between field analyses and process analyses is the sample medium. In field analyses, the samples consist of air drawn from the atmosphere. In process analyses, the samples contain significant fractions of organic chemicals drawn from process equipment such as reactors, pipelines, or stacks. The most specific reports in the literature about using IMS for media other than air regard liquid
0039-9140/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0039-91 40(97 )00095-7
20
J.H. Cross et al./Talanta 45 (1997) 19 23
chromatography or supercritical fluid chromatography (SFC). For IMS with liquids, the usual, simple radioactive ionization source was abandoned in favor of a more complex corona discharge ionization source [3]. When IMS is used as a SFC detector, the mobility spectra can show drift time effects when the mobile phase is CO2 or CO2 with modifiers [4]. Discussions of media effects on IMS can be found in Eiceman and Karpas [5] and St. Louis and Hill [6]. Ion mobility spectrometers operating in the positive mode detect compounds by transfer of a proton from a reactant ion peak (RIP) to the analyte. The RIP in air is the H+(H20),, ion; dopants can be added to form a different RIP, usually to increase selectivity. Ammonia is detected at parts per billion (ppb) concentrations, because ammonia has a high affinity for protons. Ethylene has a low affinity for protons, but the pi electrons would be expected to interact weakly with them. Thus the effect of large concentrations of ethylene on ammonia detection by IMS are not readily predictable. This paper reports experiments we carried out to assess how well ammonia could be detected in the presence of ethylene. Three considerations are important for analyzing ammonia in the presence of large concentrations of ethylene. First, ethylene is present in the IMS cell, because a silane polymer membrane in the spectrometer inlet allows ethylene diffusion into the cell. Although the membrane prevents complete mixing and IMS is not very sensitive to ethylene, the quantity passing through the membrane far exceeds the ppb concentrations usually encountered by IMS instruments. Also once in the IMS cell, poor mixing could lead to anomalous effects. For simplicity and accuracy, adding a step to separate the ethylene prior to analysis is undesirable. Second, ethylene's flammability poses a safety issue in a spectrometer as the electrical elements in the cell present a small potential for sparking. Third, radiation can initiate ethylene polymerization [7]. Many IMSs, including the one used here, have 63Ni ionization sources. The beta radiation produced by the nickel source would be unlikely to initiate global polymerization, unlike the deep penetrating (e.g., gamma) radiation reported to be used. Localized polymerization, how-
ever, could foul the source, and greatly reduce the spectrometer's performance. To minimize the potential for fouling and to improve selectivity for ammonia, we began with the nonanone-doping method developed by Eiceman et al. [8] for measuring ammonia in air.
2. Experimental section We used an E-IMS (Graseby Ionics, Watford, UK), a spectrometer that allows operating conditions to be varied easily. The spectrometer has a polydimethylsilane membrane inlet and chambers for introducing dopants into the drift or source gas. 5-Nonanone (Aldrich Chemical Company, Milwaukee, WI) was doped into the drift gas from an open, test-tube-shaped, 2-ml volumetric flask initially containing 50 ~tl liquid. Another dopant, acetone, was doped from a Dynacal permeation device (VICI Metronics, Santa Clara, CA; 42 ng rain J at 90°C) at ambient temperature. To eliminate the possibility of combustion in the cell, the drift (220 ml rain J) and source (80 ml rain-~) gas was nitrogen; removing oxygen broke the 'fire triangle' of oxidant, fuel and ignition source. The instrument was operated in a hood to remove and dilute exhaust streams safely. The cell temperature was 25°C, and the cell pressure was set 17 mm Hg below ambient. Mixtures of ammonia in air at 0.4 (5.404 1 rain-~) and 1.5 ppm (1.468 1 rain ~) were generated from a permeation device (1.510 gg rain - ~ at 30°C) in a 585 Precision Gas Standards Generator (both from Kin-Tek, LaMarque, TX). The mixtures were delivered through a PTFE tube terminating in a cup placed over the inlet nozzle. Ethylene with 1.5 ppm NH 3 was prepared by Alphagaz (Walnut Creek, CA) and was used without further analysis. It was dispensed to the EIMS through stainless steel tubing with needle valves that could divert a portion through 3A molecular sieve to remove the ammonia. This arrangement allowed coarse control of the ammonia concentration between 0 and 1.5 ppm. At the E-IMS inlet, a tee split the flow between the inlet and an exhaust tube open to the atmosphere. One side of the tee was connected to the inlet with a
21
J.H. Cross et al. Talanta 45 (1997) 19 23
plastic tube that slipped firmly into the bore of the inlet nozzle, which led to the membrane. A p u m p on the inlet pulled in the sample across the membrane at roughly 800 ml r a i n - ' Ethylene flow was increased until a flowmeter on the exhaust tube showed a slight positive flow. Waveforms from the E-IMS were averaged and displayed using Advanced Signal Processing hardware and software (Graseby Ionics). Spectra normally consisted of 32 averages of 640 points collected at 30 kHz after a 2-ms delay. Gating pulses 180 I~S long occurred at the rate of 30 Hz. A m m o n i a concentrations were measured from peak heights taken between two cursors set at nearly the same absolute baseline value. The height of the H + ( H 2 0 ) , , ion peak or protonated nonanone ion peak in air was measured similarly.
3. Results The spectrum of 1.5 p p m ammonia in ethylene in the nonanone-doped spectrometer (Fig. 1) resembled the pattern described by Eiceman et al. [8]. The only indication of ethylene's presence was a slight enlargement o f the usual shoulder upfield of the RIP. A m m o n i a appeared clearly ( S / N = 12.5) at a drift time much longer than that of the RIP. The long drift time probably resulted from the size of the ion cluster, since Eiceman et al, [8] found an a m m o n i u m ion cluster containing two nonanone molecules. Sensitivity to ammonia, however, was unacceptably low. The peak height
NonanoneRIP~ Ammonia /, ......
~
' ~
' ' ~l
l ' ~l
'
1~01,21,41~61'8
~0
DRIFTTIME(ms) Fig. l. Spectrum of ammonia in ethylene with 5-nonanone doping. To define the ammonia peak more clearly, 128 waveforms were averaged.
Table 1 Responses to ammonia in ethylene or air in the undoped spectrometer Ammonia conc. 0.0 (v/v ppm in ethylene) 0.1 0.5 1.5 0.4 (ppm in air) 1.5
Peak height 2 _+1 (mVP~ 59+_6 165 + 4 524 + 5 1319 _+5 2178 +_2
~'Average_+standard deviation of five sequentially collected spectra. for 1.5 p p m was 24 mV (Table 1), and the R I P showed no depletion. When the spectrometer was doped with acetone, two broad peaks due to the presence of ethylene were noted. One of these peaks had the same drift time as the ammonia peak, and no further experiments were conducted with acetone doping. But while the preceding result was being established, the spectrometer was exposed repeatedly to ethylene; since no consistent diminution of the RIP after exposure was observed, we concluded the source was not fouling from ethylene polymerization. Exposing the undoped, 'water chemistry' [H + (H20),, reactant ion peak] spectrometer to ethylene produced a broad band of partially resolved peaks (Fig. 2(A)). A m m o n i a appeared as a sharp peak on the small plateau at the furthest upfield part of the band (Fig. 2(B) and (C), drift time = 6.305 ms). The drift time was shorter than that of the nonanone-doped system, because the a m m o n i u m ion cluster containing water was presumably smaller. A m m o n i a was detected at 100 ppb (Fig. 2(B), S / N = 12). At 1.5 ppm, the peak remained narrow and well resolved (Fig. 2(C)). The relationship among the four concentrations measured (including 0 ppm) was nearly linear (Table 1). Exposing the undoped E-IMS to ammonia in air produced stronger responses (Table 1) at a drift time of 6.205 ms ~. The difference of 0.1 ms between drift times in air and nitrogen is a combination of the time resolution, 0.033 ms, in
J.H. Cross et al./Talanta 45 (1997) 19-23
22
During undoped experiments, the E-IMS was exposed to ethylene for numerous 5-10 min periods and one period exceeding 30 min. Over the entire period of experimentation, the RIP again showed no diminution after exposure.
A
B
i
=
i
i
I
C
~
0
'
'
'
I
2
'
'
'
I
'
4
'
'
I
6
'
" 1 ' ' '
8
I
10
'
'
'
I
12
'
' ' 1 ' '
14
*1
16
DRIFT TIME (ms) Fig. 2. Spectrum of ammonia in ethylene without a dopant. The vertical pairs of lines bracket the drift time of the ammonia peak. The other peaks are assumed to be clusters of ethylene and the reactant ion peak. (A) 0 ppm; (B) 0.1 ppm; (C) 1.5 ppm.
4. Discussion
The most important reactions in positive-mode IMS are the proton transfers governed by the relative proton affinities of the RIP and other chemicals in the reaction region. The essence of this study was to find a set of conditions that would favor proton transfer to ammonia in the presence of expected large concentrations of ethylene. Nonanone doping minimizes the effect of ethylene as indicated by the similarity to previous work [8]. Using an open container gives a nonanone concentration higher than normally used for doping in IMS. The IMS response, however, is insufficient to detect ppb concentrations of ammonia, probably because the nonanone competes too well with ammonia for protons. Undoped experiments, in contrast, produce spectra with broad, poorly resolved peaks that suggest formation of numerous weak ethylene-containing clusters undergoing exchange in the drift tube. Since ethylene has a proton affinity lower than water, formation of these ions from the H + (H20) . RIP indicates (by LeChatelier's principle) high concentrations of ethylene in the drift tube. Nonetheless, ammonia, which has a proton affinity higher than water, competes successfully for protons. In addition, the short drift time of the ammonia ion places it in a part of the spectrum free from interference by the ethylene related ions. Thus, analysis of ppb concentrations of ammonia is possible. The nearly linear response with concentration (Table 1) shows that the upper limit of sensitivity is greater than 1.5 ppm, because an IMS response is exponential, reaching a plateau at the upper limit of sensitivity. Extrapolation of the signal-tonoise ratio from 100 ppb suggests that the lower limit of detection is 25 ppb. The lower response to ammonia in ethylene versus ammonia in air suggests that competing for protons with high concentrations of ethylene in the cell does reduce the response. The reduction is much less than that caused by nonanone with its much higher proton affinity. The drift time of ammonia in air was the same as that for ammonia in the mixture of nitrogen
J.H. Cross et al. /Talanta 45 (1997) 19-23
and ethylene in the IMS cell. The mobility equation [9] K = 3q/16N(2rr/#kT)l"2(1 + ~)/f~D(T)
contains two variables that are coincidentally the same for ethylene and nitrogen. The reduced mass, /t, is m M / ( m + M ) where m is the mass of the ion and M is the mass of the neutral drift gas, 28.0 for both ethylene and nitrogen. N, the number density (molecules/unit volume), is almost identical for nitrogen and ethylene, because the molecular weights and density (1.250 g 1 J at STP) [10] are almost equal. These coincidences minimize the effects of operating the IMS with ethylene inside the cell. With another hydrocarbon, the drift time of the a m m o n i a ion would likely depend on the hydrocarbon concentration in the drift tube. If protons in the ethylene ions are bound loosely to the pi electrons, they could initiate cationic polymerization, which would foul the spectrometer. We noted no cumulative decrease in the R I P after repeated exposures, indicating that the source did not foul appreciably. In conclusion, analysis of ammonia in ethylene by IMS appears to be relatively simple. No preseparation or preconcentration is required, because crucial mobility parameters for ethylene and nitrogen are equal. The IMS requires no dopant and appears to function safely in spite of ethylene's chemical reactivity.
23
Acknowledgements The authors acknowledge the use of equipment and facilities at NASA/Johnson Space Center, Houston, Texas, to conduct this work as a dual technology utilization effort. We also thank Patricia A. Inners for generating the figures from the data.
References [1] G.A. Eiceman, Z. Karpas, Ion Mobility Spectrometry, CRC Press, Boca Raton, FL, 1994, Ch. 6. [2] A.T. Bacon, US Patent No. 5,234,838, 1993. [3] C.B. Shumate, H.H. Hill Jr., Anal. Chem. 6t (1989) 601. [4] M.A. Morrissey, H.M. Widmer, J. Chromatogr. 552 (1991) 551. [5] G.A. Eiceman, Z. Karpas, Ion Mobility Spectrometry, CRC Press, Boca Raton, FL~ 1994, Ch. 5. [6] R.H. St. Louis, H,H. Hill Jr., CRC Crit. Rev. Anal. Chem. 21 (1990) 321. [7] R.A.V. Raft, Ethylene polymers, in: Encyclopedia of Polymer Science and Technology, vol. 6, Wiley, New York, 1967, p. 283. [8] G.A. Eiceman, M.R. Salazar, M.R. Rodriguez, T.F. Limero, S.W. Beck, J.H. Cross, R. Young, J.T. James, Anal. Chem. 65 (1993) 1696. [9] E.A. Mason, Ion mobility: its role in plasma chromatography, in: T.W. Carr (Ed.), Plasma Chromatography, Plenum, New York, 1984, p. 50. [10] S. Budavari (Ed.), The Merck Index, llth ed., Merck, Rahway, N J, 1989, pp. 597 598 and 1044.