(18)J. K. Kliwer, J. J. Kraushaar. R . A. Ristinen, H. Rudolph, and W. R. Smythe. Bull. Am. Pbys. SOC.,17, 504 (1972). (19)R. Akselsson and T. E. Johansson, Nucl. lnstrum. Methods, 91, 663664 (1971). (20)J. T. Routti and S. G. Prussin, Nucl. lnstrum. Methods, 72, 125-142 (1969). (21)J. Harrison and R. Eldred, "Advances in X-ray Analysis," C. L. Grant, C. S. Barrett, J. B. Newkirk, and C. 0. Ruud, Ed., Vol. 17,Plenum Press, New York, NY, 1974,pp 560-570. (22)H. C. Kaufmann and R . Akselsson, "Advances in X-ray Analysis." W. L. Pickles. C. S. Barrett, J. B. Newkirk, and C. 0. Ruud. Ed., Vol. 18,Plenum Press, New York, NY, 1975,in press. (23)R . Akselsson and T. E. Johansson, Z. fbysik, 266, 245-255 (1974). (24)W. Bambynek, E. Crasemann, R . W. Fink, H.-U. Freund, H. Mark, C. D. Swift, R. E. Price, and P. Venugopala Rao, Rev. Mod. Pbys., 44, 716813 (1972). (25)E. Merzbacher and H. W. Lewis, "Encyclopedia of Physics," S. Flugge. Ed., Springer-Verlag, Berlin, 1957,Vol. 34,pp 166-192. (26)J. S. Hansen, J. C. McGeorge, D. Nix, W. D.Schmidt-Ott, I. Unus. and R . W. Fink, Nucl. lnstrum. Methods, 106, 365-379 (1973). (27) Micro Matter Co., 197 34th St. East, Seattle, WA 98102. (28)E. Storm and H. I. Israel, Nucl. Data Tables, A7, 565-681 (1970). (29)D. C. Camp, J. A. Cooper, and J. R . Rhodes, X-Ray Spectrom., 3, 4750 (1974). (30)J. Cookson and M. Poole, New Sci., 45, 404-406 (1970).
(31)J. M. Jakievic, F. S. Godding, and D. A. Landis, E€€ Trans. Nucl. Sci., NS-19 (3),392-395 (1972). (32)J. W. Nelson, I. Williams, T. B. Johansson, R . E. Van Grieken, K. R . Chapman, and J. W. Winchester, E€€ Trans. Nucl. Sci., NS-21 (l), 618-621 (1974).
RECEIVEDfor review March 11, 1974. Accepted January 6, 1976. This study, part of a general study of the pollution sources of trace metals in the atmosphere, was supported in part by Grant R802132 from the U.S. Environmental Protection Agency and by Grants NSF-GU-2612 and NSFGP-25974 of the National Science Foundation. One of us (Thomas B. Johansson) is grateful for travel support from the Swedish Board for Technical Development (Grant STU 72-573/U485) and from the Swedish Atomic Research Council (Grant AFR 1213-1). During the course of this investigation, R. E. Van Grieken, on leave from the Institute for Nuclear Sciences, Rijksuniversiteit Gent, Belgium, received NATO and NFWO Fellowships.
Absolute Determination of Phosgene: Pulsed Flow Coulometry Hanwant Bir Singh,' Daniel Lillian,2 and Alan Appleby Department of Environmental Science, Agricultural Experiment Station, Cook College, Rutgers UniversityJersey, New Brunswick, N.J. 08903
Pulsed flow coulometry (PFC) was developed for the absolute analysis of reactive electron absorbing air pollutants, which undergo decomposition in a gas chromatographic column, and was used successfully for the determination of sub-ppb mixtures of phosgene in air. Compared to permeation tube standards, an error of less than 15% was easily achieved when ionization efficiencies were greater than this error 75 YO.At ionization efficiencies greater than 85 YO, could be reduced to 4 % . The electron capture detector response to phosgene was comparable to that to carbon tetrachloride, one of the strongest known electron absorbers. It was demonstrated that ppb mixtures of phosgene in air are quite stable in the presence of moisture and undergo rapid heterogeneous decay on surfaces. PFC should be useful for the analysis of any unstable electron absorber (e.& peroxyacetyl nitrate (PAN), chlorine dioxide, trichloroacetyl chloride) in air and offers a wide scope of application in air pollution and industrial hygiene.
Phosgene finds wide use in industry as an intermediate for the production of a myriad of chemical compounds (11. Accordingly, there are numerous possible industrial hygiene and localized air pollution problems associated with its use. Potential industrial exposure to this compound is enhanced because of the widespread use of chlorinated hydrocarbons, many of which will undergo thermal and photochemical decomposition to form phosgene (2-9). Phosgene may also be important in air pollution work, since evidence exists that i t is synthesized in the lower troposphere during photochemical smog reactions involving halocarPresent address, Stanford Research Institute, Menlo Park,
CA.
94025. Present address, U.S. Army E n v i r o n m e n t a l Hygiene Agency, Edgewood Arsenal, MD 21010.
860
ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975
The State University of New
bons (10, 11).An accurate and convenient analytical method accordingly was required for ambient as well as TLV (100 ppb) levels of phosgene in air. The wet chemical procedures reported in the literature for the determination of phosgene vapor have been reviewed by Kolthoff e t al. ( 4 ) and Jeltes e t al. (9). In general, they suffer the typical difficulties associated with wet chemical procedures-namely, lack of specificity, interferences, losses in sampling lines, and the requirement of large samples precluding real time analysis. Furthermore, such methods are often elaborate and require considerable experience for acceptable accuracy. Priestley et al. (12) reported the electron capture (EC) gas chromatographic (GC) determination of phosgene in air. Separation was achieved using an aluminum column packed with 30% didecyl phthalate coated on 100/120 mesh GC 22 Super Support. Dahlberg and Kihlman ( 1 3 ) used a similar GC procedure with a stainless steel column packed with 20% DC 200 on Chromosorb W. The column had to be treated with acetyl chloride to preclude unacceptable phosgene losses. A sensitivity of 1 ppb was achieved by both groups. Jeltes e t al. (9) used an aluminum column packed with 30% diisodecyl phthalate coated on SO/lOO mesh Aero Pak for phosgene analysis and reported a sensitivity of 0.2 ppm with the EC detector. Because of its reactivity and the probelms associated with wet chemical methods, phosgene analysis is best accomplished by on-site GC procedures. The three columns used by the above investigators, based on our experience, require routine calibrations because of the extreme reactivity of phosgene which causes variable column losses, depending on the column history. Clearly, an absolute method not requiring calibration or suffering from changing column characteristics is desirable. Reported here is an extension of absolute coulometric analysis (14, 15) which empirically corrects for column sorption through the use of pulse
Inserting Equations 3 and 4 into Equation 2, one gets
CC14
RL; 8.0 MIN
ac -+--ac at 1 + A ai -
6 . 5 ppb
CCP3 F RETENTION T I M E ( R . T I 25MIN
A
20
1.0
MIN.
and
4 0
6.0
8.0
10.0
CCP 2 F - C C P F 2 R.T. = 4.0 MIN 28 0 ppb
Dual
( 5)
a t t = 0,
C = Ofor 1 2 0 IV(t) f o r t I 0 C = 0 for t < 0
1
o+,
( i)
(ii)
The solution to Equation 5 with BC's (i) and (ii) is given by Singh et al. (16): C = W
w
20
4.0
standard mixtures of
(t -
'('7l "))exp(-;)
(6)
+
6.0
MIN.
MIN
EC chromatograms of CC12F-CCIF2
Figure 1. CCll
3.0
1 + A
For the present model, the following boundary conditions (BC's) are valid
at 1 =
JL -
-~ kC
CC13F,
Weight (or moles) of C out of the column Weight (or moles) of C in the column nm
Carrier gas (N2) flow rata 60 ml/min. Column 10% DC 200 on 30160 mesh A Chromosorb W ( 5 4 X %-ill. SS). FSD(l0-in.) =
Since flow kinetics. For brevity, the method is called pulse flow coulometry (PFC).
( f - L(l
71 +
"I)
dt =
Lm
W(t) dt
(8)
inserting Equations 6 and 8 into 7 , it follows that
THEORY The theory and application of gas phase coulometry has been discussed in detail (14, 15).For an injection of V mi of component C in air a t t "C, the absolute concentration of C in ppb's in the absence of any column losses is given by
where Ci is the inlet sample concentration, C, is the exit concentration expressed as the analogous mixing ratio of eluting C to V ml of mixture, and X Iand X2 are the respective serial EC detector responses in coulombs. I t has been shown ( 1 4 ) that use of Equation 1 leads to accurate analysis only at high ionization efficiencies (1 ( X * / X l ) )provided no losses take place in the GC column. Reactive compounds such as phosgene, however, exhibit a significant loss in the GC column requiring the novel application of PFC for absolute analysis. In this procedure, a pulse of the reactive constituent is studied by coulometric analysis of the exiting constituent to obtain the decay kinetics in the column. Consider a packed column of length L and cross-sectional area S through which a carrier gas flowing a t a velocity u sweeps a slug of a reactive gas mixture, say phosgene in air. For conditions of no axial dispersion, mass balance gives the instantaneous moles of phosgene traversing the column in time and space as,
where r is the rate of reaction of compound C in the GC column, C* is the absorbed moles of C in equilibrium with the vapor phase, t is the time, and 1 is the length along the column axis. Assuming first-order kinetics and a linear adsorption isotherm, one can write
(v) = kC
1-w
(3)
C* = AC (4) where k and A are the first-order rate and adsorption constants, respectively.
where Fo ( U S )is the carrier gas flow rate and V,(LS) is the column volume. Equation 9 can be rewritten as
In C, = In C i
k T.',
--
FO
(10)
where C, is given by Equation 1. Clearly, C, = Ci (Equation 1)in the absence of any column reactions ( k = 0). However, the interesting aspect of this development is that it gives the inlet concentration Ci as an intercept on a linear plot of In C, us. l/Fo.
EXPERIMENTAL A special purpose GC equipped with dual 15-mCi 63Ni EC detectors in series was used in this study. This apparatus has been described in detail (14). Phosgene was used to test the PFC method. Standard mixtures of phosgene were prepared with a TFE Teflon permeation tube (17 ) and conventional dynamic dilution in an allglass system. GC separation was achieved a t room temperature (23 "C) on a 10-in. X %-in. 0.d. aluminum column packed with 30% didecyl phthalate on lOOil20 mesh Chromosorb P. Prepurified nitrogen passed through a sequence of traps (charcoal, anhydrous calcium sulfate, and molecular sieve) was used as a carrier gas. Flow rates were measured by a bubble flowmeter and GC injections were made using all-glass syringes. Standard samples of phosgene were prepared a t several concentrations (10 to 167 ppb), and each sample was injected into the GC at several known carrier gas flow rates. Equation 1 was used to determine the column exit concentration a t each flow rate. A plot of In C, us. 1/Fo was made for each standard sample, and the intercept (CJ was compared with the standard concentration.
RESULTS Individual dual EC chromatograms of CC13F, CC14 and CC12F-CClFZ are shown in Figure 1 for standard injections of air-halocarbon mixtures prepared with their respective permeation tubes. A comparison of the standard concentrations with the coulometrically determined concentrations for these compounds a t their indicated ionization efficienANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975
861
c
'0°'0
\
lO0OC 760
g Y
754 7 52 7.50
1
1
I
1
!
1
1
1
1
co C P 2
n
'
I
1
1
I
I
COCP 2
I
'
1
COCP 2
n
A
I
IO DDb S C l
1.0 j
1 001
0.00
I 003
I
a02
y I 005
I
004
.
J
006
'IF0 MIN/ml
Flgure 4. A plot of In C, vs. 1/Fo for PFC analysis. Standard concentration (S.C.) p>0.75 for all data points. Curves a, b, c, d, and e = used column. Curves f and g = freshly conditioned column I .8
A- LJL
OET 2.
1.5 P 1.4
Figure 3. Dual EC chromatograms of replicate ppb injections of
:;
phosgene
1.0
u
1.0
0.3-1
2.0
1.0
MIN.
MIN.
1.31
u
3.0
2.0
3.0
MIN.
Carrier gas flow rate = 37 ml/min. FSD (10-in,) = 0.4 ml
0
IO
20
A. Sample size =
x ) 40 50 60 70 80 CARRIER GAS FLOW RATE Fg d / M I N
90
100
Figure 5. Effect of flow rate on the ionization efficiency (p) of phosgene.
Table I. Comparison of the Coulometrically Determined Concentrations of CC13F, CClI, and CClZF-CClFz with Permeation Tube Standards Compound
Standard Coulometric Ionization concenconcentraefficiency,$ kation, ppb tion, ppb
CC1,F
0.70
cc1,
0.81
2.4 6.5
CCl,F-CClF,
0.28
28.0
j
0
40
%
2.1
14.3
6.1 5.1
6.2 81.8
DISCUSSION Coulometric analysis can be used for the absolute determination of electron absorbing compounds which do not suffer column losses. This is illustrated in Figure 1 and Table I, the latter of which shows less than a 15%error between standard ppb injections of CC13F and CC14 and their ANALYTICAL CHEMISTRY, VOL. 47, NO. 6, MAY 1975
8.0 I
Error,
cies is given in Table I. Figure 2 is a plot of phosgene permeation tube weight us. time. Presented in Figure 3 are chromatograms of replicate injections of a standard phosgene mixture. Figure 4 shows semi-log plots of C, us. 1/Fo for several standard concentrations of phosgene. A used and a freshly conditioned column were employed in these experiments. A comparison of the standard concentrations with those determined by PFC is presented in Table 11. Figures 5 and 6 show the variation of ionization efficiency with flow rate and input sample concentration respectively. Figure 7 is a plot of phosgene decay in a conditioned glass vessel in the presence and absence of water vapor. Also shown in Figure 7 is the dry phosgene decay profile in a ground-glass vessel having a large specific surface area.
862
Inlet solute weight = 0.08 ng (ppb X sample size ml = 20)
09
I
-
p p b x SAMPLE SIZE
12.0 I
-
16.0
ml I 20.0
- +
24.0 r
28.0
(01
0.5 0
20.0
40.0
€0.0
80.0
100.0
120.0
I DDb I X SWPLE SIZE m i
Figure 6. Effect of solute concentration on the ionization efficiency of phosgene
coulometrically determined concentrations (Equation 1). For compounds exhibiting low ionization efficiencies, however, significant errors can be expected as is demonstrated by the 82% error for CC12F-CC1F2 analysis a t 28% ionization efficiency. The reactivity of phosgene suggested the use of a T F E Teflon permeation tube, as a primary standard, for the dynamic preparation of ppb phosgene-air mixtures. After a few days of conditioning, the tube exhibited a constant rate of weight loss (Figure 2) and provided an extremely reliable phosgene output (Figure 3). Figure 3 further shows that phosgene is an excellent candidate for coulometric analysis
3oooL c 1 0 0 0 ~
5,O
190
MINUTES I?
2O :
Zt"
310
MO
.
4100
4
\
\J
c
o+&-r
7 0
$0 MINUTES
8b
Id0
Id0
10.1 I40
Figure 7 . Decay of phosgene in contact with glass surfaces at two humidities because of its high ionization efficiency (>85%) a t a convenient carrier flow rate. The assumptions of a linear absorption isotherm and first-order kinetics for phosgene-column reactivity are proved valid by Figure 4, where semi-log plots of C, us 1/Fo are shown to be linear (Equation 10). As developed in the theory, the intercepts of these lines are the inlet phosgene concentrations (C,). From Table 11,PFC concentrations are seen to be in very good agreement with the standards (maximum error = 10.2%). This is attributable t o the use of ionization efficiencies exceeding 75% and the inherent applicability of the assumed kinetic model. Since the lines are parallel for a given column (Figure 4a-4e and 4f-4g), once the slope of one of them is determined, analysis is possible by a single injection a t one flow rate. The inlet concentration C, would then be determined as the intercept of the parallel extrapolated line. I t is recommended, however, that the slope of an analysis line be routinely checked. The rate constant k may vary gradually with column history as is suggested, in the extreme case, by the different slopes of the lines obtained for the used and the conditioned columns. In conventional GC procedures, a change in slope would correspond to a need for a new calibration, whereas, in PFC, the analogous requirement would be the injection of an unknown sample a t a few flow rates. This is substantially simpler than a rigorous calibration, particularly for reactive gases like phosgene. Since there is evidence (14, 1 5 ) that carrier flow rate and solute mass are the major factors governing the ionization efficiency and, hence, the accuracy of the analysis, the effects of these variables on phosgene ionization efficiency were studied. A linear plot (14, 15) of the reciprocal ionization efficiency us ilow rate (Figure 5 ) shows that an increase in ionization efficiency can readily be obtained by reducing the carrier gas flow rate. A t a given flow rate, however, Figure 6a clearly demonstrates that the ionization efficiency is nearly independent of the input solute mass up to about 0.1 ng. Beyond this, the ionization efficiency is seen to fall off, presumably because of the decreasing ratio of electrons to solute molecules (Figure 6b). Consequently, in the regime where solute mass has an effect on the ionization efficiency (>O. 1 ng for phosgene), the latter can be increased by a reduction of solute mass as well as of carrier gas flow rate.
Table 11. Comparison of Standard Phosgene Concentrations with PFC-Calculated Concentrations Standard ConcentTations, ppb
Calculated concennations, ppb
Emor, c/a
Efficiency range, p
10.0 35.0 40.0 80.0 167.0 80.0 167.0
9.6 33.0 38.0 76.0 150.0 76.0 150.0
4.0 5.6 5.0 5.0 10.2 5.0 10.2
>0.85 >0.80 >0.80 >0.80 >0.75 >0.80 >0.75
Comments
Used Used Used Used
column column column column U s e d column Fresh column Fresh column
Having established the validity of PFC for the absolute determination of phosgene in air, the method was used for a preliminary study of phosgene vapor stability in glass vessels. A 560-ml cylindrical glass vessel (45-cm X 4-cm i.d.) was flushed with 167-ppb mixture of phosgene in dry air for 60 minutes and was then capped and allowed to stand a t 23 "C for subsequent analysis. Figure 7 a shows the first-order decay of phosgene in this system. After three hours, water-saturated air was injected into this vessel by syringe to bring the water vapor concentration to 785 ppm. I t is clear from Figure 7a that water had, a t best, a minimal effect on phosgene decay. Figure 7 b shows a much faster decay of a phosgene-air mixture in a vessel of much higher surface area (ground glass), indicating that the role of water, if any, is manifested in heterogeneous surface reactions. Further, in tropospheric photochemical smog simulation experiments where phosgene was synthesized from C&14 in 200-liter Teflon bags ( I O ) , it was found that phosgene concentrations remained stable for periods exceeding 15 hours in the presence of 10,000 ppm of water vapor. We attribute the absence of surface reaction here to the passivation of the walls by photochemical reaction products CC13COC1. The ionization efficiency of phosgene is comparable to that of carbon tetrachloride, suggesting that femtogram (10-15g) analysis would be possible in the absence of significant sorption. The kinetics for the heterogeneous decomposition of phosgene in the GC column and glass sur-
CONCLUSION Pulse flow coulometry (PFC) facilitates the absolute determination of phosgene down to sup-ppb levels. The strategem of compensating for column losses by extrapolating to zero residence time in the column should offer wide application to similarly reactive electron-absorbing compounds such as peroxyacetyl nitrate (PAN), C102, CCl&OCl. The ionization efficiency of phosgene is comparable to that of carbon tetrachloride, suggesting that femtogram (lO-15g) analysis would be possible in the absence of significant sorption. The kinetics for the heterogeneous decomposition of phosgene in the GC column and glass surfaces is first-order at low concentrations (
