Anal. Chem. 1002, 64, 2096-2100
2008
Increased Response of the Reversal Electron Attachment Detector and Modeling of Ion Space-Charge Effects S. Boumsellek and A. Chutjian' J e t Propulsion Laboratory, California Institute of Technology, Pasadena, California 91109
D o a n and wnrltivlty tests of a modified version of the 80called reversal eioctron attachment detector (READ) are presented. The new verdon uws a sphwkalcathod. capabk of emitting higher electron currents. As In the original READ (which wed a planar emitter) electrons are focused into an ekctrortatic mirror whkh reverses their traJectorles. I n the reversal region electrons have essentlaily zero energy and attach to target moiecuierto form negative ions. The electron gun lens system has been modifiedWlng a field and traJectory cod. with space charge included. Electrontrajectories have beoncalculatedfor 1mA current focused into a reversalregbn of W-mm diameter. The detection ilmH of the apparatus is approximately 25 times lower than for the original READ. Nonlinearityin the measured dgnaiVI eiwtroncurrent k dercrlbed by a model In which a spherical bail of ions expands outward with veloclty determined by the space-charge force and the initial velocity of ion formation.
pressure species4 into vacuum and to detect them with high efficiency. Work is currently underway in this laboratory to address the former problem. In efforts to increase the response of the READ itself,a newer version has been designed and tested. An indirectly-heated spherical cathode is used to give a larger emission area relative to the planar cathode used earlier.' Because of the larger currents and the focusing nature of the emitter, the electron gun lens system had to be redesigned. In order to demonstrate whether an improvement in response was realized, the method of standard additions6 is used with mixtures of CC& or C ' c 8 6 in Nz. Both test molecules have the same essential property as explosives: the attachment cross sections peak a t or near zero electron energy, consistent with the Wigner threshold law.6 Finally, the question of linearity of signal with electron current is addressed herein, and the observed variations are modeled in terms of a spreading, spherical ball of ions formed at the reversal region.
EXPERIMENTAL SECTION
INTRODUCTION There exists an ongoing need in trace-species analysis to develop detectors which are sensitive, specific, and resistant to interferences from nontargeted chemical components. In previous work the concept, design, and performance of the so-called reversal electron attachment detector (READ) was presented.1J The READ utilizes the fact that there are many classes of molecules which have negative-ion resonances at low energies (below about 5 eV, say) and hence can form one or more negative ions upon electron attachment. Unlike other higher-pressure corona- and glow-discharge devices or chemical ionization techniques the READ can provide a 'match" between the peak of the electron energy distribution function and the peak energy of the target resonance state. Thus READ can access a broader range of targets: molecules which not only attach thermally but also attach at higher electron energies.l-3 The detection limits of the READ will depend upon, among other factors, the space-charge-limited electron current, at the resonance energy, that can be delivered to the reversal region. Operation of the READ has been demonstrated on several important classes of molecules: the explosives RDX, PETN, and TNT3 and several halogenated molecules.2 The trace detection of explosives is vital in securing airports, nuclear installations, harbors, embassies, etc. The challenge with explosives is to transfer efficiently the extremely low vapor
* To whom correspondence should be directed.
(1) Bernius, M. T.; Chutjian, A. J. Appl. Phys. 1989, 66, 2783. (2) Bernius, M. T.; Chutjian, A. A d . Chem. 1990, 62, 1345. (3) Boumaellek, S.; Alajajian, S. H.; Chutjian, A. J. Am. SOC.Mass Spectrom. 1992,3,243. Chutjian,A.; Boumsellek,S.;Alajajian, S. H. In Proceedings of the First International Symposium on Explosiue Detection Technology; Khan,s. M., Ed.: Federal Aviation Administration: Atlantic City, NJ, 1991; p 571. 0003-2700/92/0364-2096$03.00/0
System Description. The READ system has been detailed in earlier work,'s2 and only a brief description is given here. Referring to Figure 1, READ consists of an indirectly-heated, spherical cathode (element V1) from which electrons are extracted, focused with a shim element (Vl'), and accelerated by a five-element lens system (Vl-V5). They are then focused into an electrostatic mirror by elements V5-V7. The mirror decelerates the electron beam to zero longitudinalvelocity and nearzero radial velocity at the reversal plane R. At R, the reversed electron beam crosses or 'touches" the molecular target beam. Negative ions formed at R by the attachment or dissociative attachment (DA)process are then extracted by elements V5-V8. The electron beam is square-wave-modulated by fast switches s1-S~ with a nearly 50% duty cycle. These switches are power MOSFET-based to ensure a fast (50-11s) risetime of the lens voltages.' Electron attachment to the target takes place at R during half of the pulse cycle (electronsON, ions OFF). During the second half (electrons OFF, ions ON) the negative ions are extracted,focused,deflectedby a 90° electrostaticanalyzer (ESA) to ensure the sign of charge, and further focused onto the entrance plane of a quadrupole mass spectrometer (QMS).The ion signal produced at the channel electronmultiplier is amplified, and the intensity of each mass peak is read on a pulse counter after a 10-8 integrating time. Electron Optics and Ray Tracing. Careful consideration had to be given to both electron and ion space-charge effects in the READ.' Intense space charge is encountered many times in the design: (a) at the cathode, where electron currents are high, and velocities low; (b) at an intermediate crossover in the gun lens system; (c) at the reversal region R where currents are again (4) Dionne, B. C.; Rounbehler, D. P.;Achter, E. K.; Hobbs, J. R.; Fine, D. H. J. Energ. Mater. 1986,4,447. (5) Willard, H. H.; Merritt, L. L., Jr.; Dean, J. A,; Settle, F. A., Jr. Instrumental Methods of Analysis, 6th ed.; Van Nostrand New York, 1981; Chapter 29. (6) Wigner, E. P. Phys. Reu. 1948, 73, 1002. (7) Bernius, M. T.;Chutjian, A. Rev. Sci. Instrum. 1989,60,779;1990, 61, 925.
0 1992 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1902
q’ pu
ESA
R
L
t
I I I
I
VOLTAGE
DISPLAY
I
T
2097
41 I
$
v,
0
I
QMS
F
I
AMP
Flgu. 1. Schematic diagram of the READ devlce,showingthe spherlcal cathode region (F, V1, Vl‘), electron gun lens and reversal system
and Ion extraction lens system (V5-V8). Also shown are the reversal region R, target Inlet T, 90’ electrostatlc analyzer (ESA), quabupolema88 selector (QMS),and channel electronmultlpller (CEM). Lens elements V5-V7 are cycled by the fast swltches S1-S3, respecthrely. (Vl-V7),
high and the electrons are slowed to zero and near-zero longitudinal and radial velocities; and (d) again at R where the slowmoving (negative)ions are formed in the presence of the (negative) electron space charge. All designs and ray tracing were carried out using the Herrmannsfeldt field and trajectory code implemented on a personal computer.8 The spherical cathode region was designed using cylindricalsymmetry. The electron gun, as redesigned, consisted of a spherical cathode (Vl), a “shim”electrode (Vl’), an anode (V2) with a 5.2-mm-diameter aperture, and an electrode (V3) held at near-ground potential. A three-element lens (V3-V5) transports the electron beam to an electrostaticmirror (V5-V7) where it is reflected back toward the cathode. The Neumann and Dirichlet boundaries,as well as electrontrajectories are shown in Figure 2. Trajectories of the negative ions formed at R have been calculated previously’ and remain unchanged in the present modification. The starting condition for the program included Child’s law on a spherical cathode with a radius of curvature of 8.15 X m and an emission area of 5.6 X m2. The design requirements of an electron current I, of 1mA at a 650-V acceleration voltage Vz (potential on the anode element V2, with the cathode taken as zero of potential) resulted in a gun perveance I J V Z ~=/6.0 ~ X 1VA/V3/2. In order to have good spatial resolution within the limits of the mesh-point capacity of the program, the problem of boundaries describing the READ gun portion (Vl-V7) was split into two parts: part 1included the electrodes V1, Vl’, V2, V3, and V4; and part 2 the electrodes V2-V7. Three electrodes, V3-V5, were made common to both parts to ensure that fringing fields from part 1were included in part 2. The mesh size in part 1 (0.43 “/mesh unit) was one-third that of part 2 to be able to resolve small details in the gun structure. After running part 1the output trajectorieswere injected (asinitial conditions) into part 2 at a position on the lens axis where the potential was (8) Herrmannsfeldt, W.B. EGUN-An Electron Optics and Gun Design Program. Stanford Linear Accelerator Report No. SLAC-Report-331; Stanford University: Stanford, CA, October 1988. See also: Becker, R. NucZ. Instrum. Methods Phys. Res. 1989, B42,162.
0
5
10
15
AXIAL DISTANCE (cm) Flgurs2. Electron trajectorlesuslng the Spherical cathode, calculated up to the reversal region R during the electron ON cycle.
identical in both parts. Typical values of the voltages used to compute the trajectories in Figure 2 are V I = 0 V, VI, = 150 V, Vz = 650 V, V3 = 2 V, V , = 1340 V, Vs = 112 V, Ve = -30 V, and V , = -30 V. The total electron current calculated in the program was 1mA, and the beam diameter at R was 3.5 mm. Calculations were carried out on a 386-based PC, and a typical running time for both parts was 5 min.
RESULTS AND DISCUSSION Sensitivity Tests. Using the method of standard additions6 the sensitivity of the new version of the READ was measured. Test mixtures of CCl, or C - c a 6 in Nz were prepared. At zero electron energy C1- is the only fragment ion produced in DA to CCl,, while the c-ca6- molecular ion is the only product in attachment to c-C$e.Q With their large attachment cross sections peaking at zero energy, these molecules served as good simulantsof the nitrogen-containing explosives currently in use. Naturally, the ultimate detection limit with the explosives themselves will depend on their attachment cross sections. These have not been measured in any explosives molecule. Sample mixtures were prepared in an all-stainless steel vacuum system. All lines were kept warm a t 370 K to prevent “sticking” of the sample to the walls during preparation and transfer to the collision region. Targets were used in their reagent grade and subjected to ten freeze-thaw cycles to remove dissolved gases. They were mixed with Nz (99.99% purity) in a concentration ratio C (particle density) = 1(pure target) to 1X lo-”. The quadrupole mass spectrometer was tuned to either mle = 35 or 186, and the negative-ion signals were measured as a function of the fractional concentration C. In addition, to understand the signal linearity properties of the READ, a series of measurements of count rate v8 electron current was made a t three values of C = 1 X 1V,1 X 10-2,and 1. The total pressure in the main vacuum chamber was fixed at 2.7 X 10-5 P a (2.0 X le7Torr) throughout. Between all measurements where C was changed, the lens system was baked to 415 K using a quartz-iodine lamp placed near the collision region. This served to remove residual target
2098
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
1061'
"'1
'
'"I
' "'1
' "'I
'"'I
' / ' ' I ' " " ' "'I ' ' " 1
'
'''I
' "'I
'
"j
CCP, + N,
2
1
I
lo-',
10'0
109
10"
104
10.2
100
CCe, FRACTIONAL CONCENTRATION Figure 3. READ sensitivity evaluation using the method of standard dilutions for various concentratlons C of CC, In NP. The solM line represents a least-squares fit to the data. 105
3
1
O
0
4
0.4 0.8 1.2 1.6
2.0 2.4
2.8
Y
ELECTRON CURRENT I, ( ~ o - ~ A )
Negativalon Ci- signal vs 1. for the three concentratlons and (c) C = 1 (pure CCI,). SolM (a) C = 1 X lo4, (b) C = 1 X lines are calculated fits to data using eq 6. Flgurr 5.
1021012
1010
108
106
CmC6F6FRACTIONAL
Flguro 4. Same as
104
102
100
CONCENTRATION
Figure 3, but for c-C,,F,, as the target.
molecules from the immediate inlet lines and surfaces, and prevented "memory" effects in the measurements. A 10-min baking time was sufficient to remove residual but a 2-h bake was required to remove the "stickier" CCb. Blank runs were performed, and the background was measured to be only electronic-noise background in the mass ranges 3040, or 180-190 m u , before introduction of the next mixture. At each concentration, the signalwas monitored over a period of 0.5 h to ensure stability of the mixture at the 3% level. Results of sensitivity curves obtained for CC4 and c-Ca6 are shown in Figures 3 and 4, respectively. The errors represent the quadrature sum of the statistical counting error and the error in reading the two pressure gauges used to make up each fraction C and are expressed a t the 1 . 7 ~(90%) confidence level. In the case of CCl, at a common concentration of 10 pptr, the signal rate is approximately 25 times greater in the spherical READ system than in the earlier planar-emitter version.2 Count rates with are lower than with CC4, consistent with the smaller attachment cross section and rate constant for the former.9 Also, from Figure 3 one could extrapolate to a count rate of 13 kHz at a mixture C = 1 X 10-l2. As a caveat, however, such an extrapolation (9) See, for example: Orient, 0. J.; Chutjian, A.; Crompton, R.W.; Cheung, B.Phys. Rev. A 1989,39,4494 (CCl,). Chutjian, A.; Alajajian, S. H.J. Phys. B 1985, 18, 4159 ( C - C ~ S ) .
must be viewed with caution,since in both cases the sensitivity curves deviate from a slope of unity. This very likely occurs because most of the ions at the high fractional concentrations are lost through ion-ion repulsion. In the next section this question is addressed in terms of the principal nonlinear effect in the system-that of ion space charge. Space-ChargeEffwts in READ. In many applications of a trace-analysis instrument, for example in air or water pollution monitoring, one would like to relate a measured signal intensity to a sample concentration in a linear fashion. In other applications, such as in explosives detection, one is more interested in the presence of a sample, regardless of concentration (a so-called "go, no-go" situation). To explore the dependence of the READ signal on target density and electron current, a series of measurements with CC4 was carried out at three values of C (1X 10-4,l X 10-2, and 1)and over the range of currents I, = 1.0 X 10-6 to 3.0 X 10-4 A. This approach was preferred to the method of Figures 3 and 4, since the electron current is the only variable and the optical properties are left unchanged for each value of c. Shown in Figure 5 are results for the three concentrations. In general, two regions of slope can be distinguished: (a) the C1- signal rate is near-linear at lowl., followed by (b) a marked flattening of the signal at higher I,. Results at the higher currenta correspond, as will be shown below, to ions being lost by a rapid expansion due to space charge, and leaving the field of view of the ion extraction-lens system V5-V8. The physical picture adopted in understanding the shape of the sensitivity curves is that of a spherical ball of ions
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
created at R by the DA process, during the electron ON cycle. There is a time delay At introduced between the electron OFF cycle and the ion ON cycle. This delay is set experimentally at At = 2.1 ps and is a “settling time”for the electrons prior to pulsing out the ions. During this time, the negativeion cloud expands under the action of the space-charge force of the negative ions (electrons are OFF), the initial velocity of ion formation, and small fringing electric fields from V8, deflectors in V5, etc. The ion density within this cloud increases with increasing C and I,. The resulting increase in ion-ion repulsion will lead to a larger expansion velocity for the sphere, and hence a larger final radius rf after the delay At. The ion extraction system V5-V8 is only capable of focusing an ion ball of less than a certain radius rf. This is calculated from the Herrmannsfeldt code to be rf = 3.35 mm for the present system. Referring to Figure 6, one has a sphere of initial radius ro as set by the radius of the initial electron beam. During the period At in which the electrons are turned off, some portion of the sphere (radius rl) expands to aradius rfby the combined velocities of space-chargerepulsion and the initial velocity of ion formation in the DA process. The ions within that sphere are focused by the ion lens system, mass-analyzed, and detected. (The outer FO edge can expand beyond rf,but those ions will be lost.) At any C and Ie the extracted ion signal S will be given simply by
where K is an instrument transmission and detection efficiency, p is the ion charge density contained within the sphere of radius ro, and 4/3ar13is the volume of detected ions. The product p(4/3ar13) is the number of ions contained within the sphere of radius 1-1. To determine r1 one calculates the radial electric field E, at any radius r due to a charge density p contained within a radius ro. This is given from Gauss’s law as (MKS units)
lo9. From the relation edV = eE,dr
=
3
udu = (e/3c0m)(:)pdr
(3)
where m is the C1- mass and e is the electron charge. Upon integrating, one obtains the radial velocity (letting K = (2e/ 3co?72)ro3pand A uo2 + K/ro) u = dr/dt = (A -K/r)’j2
(4)
In time At the sphere surface at r1 will have moved to the distance rf,as determined by the integral of eq 4 At =
1 mm H
Flgure 8. Geometry for the spherical ball of negative ions formed at R. A sphere of CI- ions of radius ro is formed by the initial electron beam. During the electron OFF cycle, and before the ion ON cycle, a portion 4 of this sphere expands by space-charge repulsion, and the initial kinetic energy of CI- formation, to a maximumradius rf accepted by the ion extraction optics.
adjusted parameter and may be calculated from the standard expression for the attenuation of an electron beam of length 1 and radius ro through a thin target beam of density n
S (Hz)= ip($rr;)
where l / ~ o= 36a X mudu, one obtains
2099
s“
dr ( A - K/r)1/2
(5)
This can be evaluated in closed form to give [letting u = (1
- (K/Ar))1/2]
At=A[v+ln(E)]uf A312 1 --2
(6) u,
For a given interval At, the value of r1 was determined from eq 6 at each p, C, and le.The charge density p is the only
Ienul P=-
(7)
UOa
where a = aro2 and u is the attachment cross section, taken as 5 X 1O-l’ m2. Values of the parameters used are listed in Table I together with the final calculated value of p. Since only the ratio l/a is determined from eq 7, their individual values are not unique but were kept reasonably close to trajectory-calculated results. The value for uo, corresponding to a 35Cl-energyof 0.14eV, is comparableto the formation energy (0.100 eV) measured in ref 10 for the DA process. (The slightly larger value here can arise from accelerations through fringing fields at R.) The value of n is an estimate from the pressure in the main chamber during operation, and that for u is an estimate based on the known attachment cross section for CC4.9 Results of the calculation are given as the solid lines in Figure 5. The agreement, especially in the pure CC4 case (greatest nonlinearity) is surprisingly good for this simple model. The calculation reveals both a linear and a nonlinear region. The linear region extends to higher Ie as the concentration of target is reduced to C = and 10-4(lower chargedensity p ) , in agreementwith the experimentalresults. It is interesting to note that the charge density p does not scale with the target density between C = lo4 and C = 1. This results from the fact that in Figures 3 and 4 the electronand ion-beam tuning conditions are changed between measurements at each concentration. To show this, a separate series of measurements was made at concentrations C = 1W2, lO-l, and 1. The electron current was kept low (I, = 0.051 X A) to minimize ion space-charge effects, and the optics were not retuned between concentration changes. In this case, the C1- signal and concentration scaled linearly (at the 2u or 98% confidence level) as 1.00:10.8:118 for the concentrations 10-2:10-1:1.0. And hence the detection response at small I, and fixed tuning conditions is linear with p. (10)Walter, C.W.; Smith, K.A.; Dunning, F.B.J. Chem. Phys. 1989, 90,1652.
2100
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
Table I. Values of Parameters Used in t h e Space-Charge Calculations 1 10-2
104
1 (varied) 1 (varied) 1 (varied)
2 X lOI7 2 X 1015 2 X 1013
5 5 5
1.0 4.6
9.0
8.8 8.8 8.8
6.3 x 10-5 7.5 x 10" 1.0 x 10-6
1.25 1.39 1.51
1.8 x 1 v 7.0 x 10-1 1.0x 10-7
*
a See text for details. See ref 9. c Calculated to be approximately one-third of a mesh unit, or 4.3 x 10-4 m. 0.14 eV for C1-, measured to be 0.100 eV in ref 10. e Electron beam area is calculated to be rro2 = 9.62 X 10-6m2.
Assumptions in the model are (1)neglect of space-charge neutralization due to positive ions (Nz+,N+,Nz2+,CCl,+, etc.), (2) use of a spherical ball, rather than a cylinder of charge generated by a near-cylindrical beam of electrons slowing into the mirror region (see the calculated geometry in Figure 2 of ref l),(3) a uniform electron density distribution within the radius of formation ro, (4) a uniform gas target density distribution at R, and (5) neglect of fringingfield gradients at R which would produce a distribution of starting velocities UO. These effects can, in principle, be included in a more detailed calculation but would very likely involve further parameters and detract from the simplicity of the present model. This one-parameter calculation gives good agreement with experiment and captures the main feature of the spacecharge effects. CONCLUSIONS
The READ technique was modified to provide a spherical electron emitter and hence greater electron currents at the reversal region. A lowering of a factor of 25 in detection limit was obtained relative to the earlier planar-emitter design. A
question of the linearity of signal with target concentration and electron current was addressed using a model in which a spherical ball of ions expands under space-charge force and the initial velocity of ion formation. Good agreement was found between the model calculation and the experiment, indicating that ion space charge is responsible for the observed nonlinearities at higher charge densities. With the increased READ response work is currently in progress to interface the reversal region to accommodatetrace species at atmospheric pressure using sample enrichment methods. ACKNOWLEDGMENT
This work was carried out at the Jet Propulsion Laboratory, California Institute of Technology, and was supported by the Department of Transportation, Federal Aviation Administration Technical Center, through an agreement with the National Aeronautics and Space Administration. RECEIVEDfor review April 29, 1992. Accepted June 17, 1992.
