Gas-phase coulometry by thermal electron attachment

University of Reading, Reading, Berkshire,U.K.. E. R. Adlard. Shell Research Limited, Thornton Research Centre, P.O. Box 1, Chester CHI 3SH, U.K...
0 downloads 0 Views 378KB Size
Gas-Phase Coulometry by Thermal Electron Attachment J. E. Lovelock and R . J. Maggs University of Reading, Reading, Berkshire, U.K.

E. R . Adlard Shell Research Limited, Thornton Research Centre, P.O.Box I , Chester CHI 3SH, U.K. With intensely electron-absorbing substances, the electron capture detector tends toward destructive detection in which a large proportion of the substance entering the detector is ionized irreversibly. This tendency can be developed to the point where the detector can function as a gas-phase coulometer. The paper is concerned with the physical basis, experimental verification, and practical conditions for coulometric analysis with the electron capture detector.

it. The reactions within the detector have been stated previously (2) but, briefly, they are:

THEEXQUISITE SENSITIVITY of the electron capture detector made possible the discovery of the ubiquitous distribution of halogenated pesticides in the natural environment. Because of the requirement for chemical asepsis in its use, electron attachment has tended t o be regarded as a n imprecise method of analysis; certainly, if improperly used, it can give anomalous or even false response ( I ) . It was observed, however, that with intensely electron-absorbing substances and under favorable conditions of analysis, 90% or more of the molecules entering the detector were ionized. There follows a description of the basis, experimental verification, and practical considerations of the use of the electron capture detector in gas-phase coulometry. When operated coulometrically, thermal electron attachment is both at its most sensitive and also an accurate and absolute method of analysis.

Coulometry requires first that the equilibrium of reaction (a) is almost entirely to the right, or that AB- is rapidly scavenged from the system by reactions (b) o r (c). The concentration of positive ions within the detector is approximately 1000 times greater than the concentration of electrons and this would favor reaction (b). The encounter between oppositely charged species is an exciting event, as much as 15 electron volts of energy being released; there is little chance of a polyatomic molecule surviving this type of event intact and, consequently, the products of reaction (b) are very unlikely t o be as electron-absorbing as A B since intensely electron-absorbing substances are rare. It is possible however that the products of reaction (c) are electronabsorbing substances, in which case a greater than coulometric response might be observed; so far, however, this has not been the practical experience. In certain circumstances, the physical situation would seem t o favor the ionization of most of the molecules entering the detector and, hence, a coulometric response. The experiments made to test this hypothesis were based o n the simple physical model of a stirred reactor. Consider a detector of volume V into which carrier gas is flowing at a rate U. In the detector, a source of ionizing radiation is maintaining a concentration of electrons and into the carrier gas stream a n electron-absorbing substance is being introduced at a constant rate a . In the detector, the average concentration of the electrons is in considerable excess so that the rate of the reaction can be described by a pseudo first-order rate constant K. The gas within the detector is assumed t o be completely mixed. I n these circumstances, there are two processes for removing the electron-absorbing substance from the gas within the detector. First, the reaction with free electrons and second the ventilation of the detector by the flow of carrier gas so that the concentration of the substance at equilibrium C , is simply described as;

PHYSICAL BASIS

In the pulse sampling mode of operation ( 2 ) the electron concentration ( P ) within a n electron capture detector is related to the time interval between pulses according t o the relationship ;

where A is the rate of production of electrons by beta-ray ionization, V the detector volume, KD the rate of electron removal by all processes, and t the time available for electron accumulation between pulses. This relationship is established on a firm experimental basis for most of the commonly used electron-transparent carrier gases. Under typical conditions of operation, the average electron concentration is in the region of 10-13 mole liter-' or 6 X lo7 electrons ml-l. Now the minimum detectable concentration of intensely electron-absorbing substances, for example, SFe, CCh, halogenated pesticides, etc., is in the region of lo6 molecules mIk1, so that in the detection of these substances, the electron concentration within the detector is in excess of the solute concentration over a considerable range, especially since reaction with electrons implies that the solute concentration within the detector is lower than in the gas stream which enters

Z

AB-

+ AB$

+ @ -,neutral products

c, =

1962

(a) (b)

+&

/ A ' AB-A\

+ B'

(c)

a KV+ U ~

It follows directly that the proportion p of the substance ionized under these conditions is given by the dimensionless expression ; KV (3) p

(1) J. E. Lovelock, ANAL.CHEM. 35,474 (1963). (2) W. E. Wentworth, E. Chen, and J. E. Lovelock, J. Phys. Cliem., 70,445 (1966).

AB-

=

F

U

The rate constant for the reaction of several compounds, with electrons is known, including that for SFe. The electron

ANALYTICAL CHEMISTRY, VOL. 43, NO. 14, DECEMBER 1971

Radioactive source and strength, mCi Nickel-63 15 Tritium 200 Americium-241 0.015 Rate of removal of (I

Table I. Detectors : Dimensions and Characteristics Chamber dimensions. cm Cathode Anode Diameter Length Diameter Length Volume, ml 1.o 2.0 1.25 1.45 0.1 0.9 0.2 0.5 1 0.03 1 .o 0.95 0.95 1.3 0.1 electrons from clean dry nitrogen at NTP.

concentration can be readily determined from the current flowing in the detector and can be varied by changing the interval between pulses. The volume of the detector and the carrier flow rate are also simple t o measure o r to change. The model can therefore be tested by direct experiment. Expression 3 above can be written as a linear relationship between l/p and (U), thus: ljp

=

1

+ UjKV

K D ,SK-' ~ 7 x 102 2 . 6 x 103 3 x 103

7.20

10.2

I O

08

&

400!~s

(4)

This is more convenient t o apply t o the test of experimental results than is Expression 3. Before proceeding t o the experimental section, it is useful t o describe the basis of a second model system which was frequently used for test purposes and could also be applied to practical use. Consider two identical electron capture detectors connected in series and let their signals be X and Y in coulombs, respectively. Let p be the fraction of a compound A B which is ionized in each detector and let Q be the quantity in molecules of a plug of A B which enters the first detector:

Saturation current, nA 8.55

0 6

P 04

\r

\

02

220us

IOOUS

0 0

I

I

I

5.

U, ml s-'

Signal from 1st detector

X

=

PQ

(5)

Signal from 2nd detector

From Equations 5 and 6 p = l - -

Y

X

(7)

It follows that X2

Q = ~-

x-

Y

This model is not only the basis of the coulometric application of the electron capture detector but also (from Equation 3) it is a simple direct route to reaction kinetic information. It is important t o note that in all of the foregoing discussion and in the experiments which follow, conditions were considered o r chosen, such that not more than 5 of the electron concentration within the detector was removed during detection. Coulometric analysis with larger signals is probably possible but was not sought in these first experiments. The relationship for the electron capture detector developed by Wentworth ( 2 ) does not apply under these coulometric conditions where the detector response is linear with concentration of the test substance. EXPERIMENTAL Detectors and Electronics. Three different electron capture detectors were used; all were cylindrical and made of stainless steel. The radioactive sources were a t the inner surface

Figure 1. Proportion of SF, ionized in the nickel-63 detector at different flow rates with pulse period as a parameter

of the detector bodies which were the cathodes. The anodes were metal rods mounted coaxially within the detector volumes. The principal dimensions and characteristics when tested in nitrogen gas a t N T P are listed in Table I. The polarizing voltage was from a D4147 Lab-Gear pulse generator with a single stage of additional amplification to raise the pulse amplitude t o 35 volts. The signals from the detectors were amplified using two Philbrick P2A operational amplifiers and applied to both channels of a Tektronix model 564 storage oscilloscope. Chromatography. All experiments were made with nitrogen as the carrier gas. The nitrogen was dried by being passed over activated 5A molecular sieve but was not further purified. The detectors and columns were mounted in a Pye Model 104 gas chromatograph. RESULTS AND DISCUSSION Figure 1 shows how p, the proportion of SF6 ionized, varies with flow rate. The experiment was with two, 2-ml volume nickel-63 detectors in series. Measurements were made at pulse intervals of 400, 220, 100, and at 30 microseconds. The observed points (circles) are compared with values calculated (the continuous curves) from Expression 3. In Figure 2 the results of this experiment are plotted as a graph of 1/p against U. Figure 3 shows a comparison of the proportions Of SFs ionized in the three detectors listed in Table I ; the ionization efficiency a t infinite pulse

ANALYTICAL CHEMISTRY, VOL. 43, NO. 14, DECEMBER 1971

1963

1.0

l o 0.8

4

0.6

0 AMERICIUM- 241

P

0 NICKEL-63

0.4

A

TRITIUM

I

- 3

?

0.2

0.0

I

2

U, m l s - I Figure 3. Proportion of SF6 ionized at infinite pulse period in the detectors listed in Table I

2

Table 11. Compounds Giving Coulometric Response under Favorable Experimental Conditions Telomer 2, SFs(CF2)2Cl Telomer 4, SFs(CFZ)&l Telomer 6, SF~(CFZ)~CI CFzBr2 Perfluorobenzene CFCls Carbon tetrachloride, CC14 Perfluorotoluene Imperial Chemical Industries trade name. 0

I

2

I

U,ml s-‘

Figure 2. Data used in Figure 1 plotted to show the linear relationship between l / p and U period was compared at different flow rates. In this experiment the voltage between the electrodes of the detector under test was altered from 100 volts dc (sufficient to remove all electrons before reaction could take place) to zero volts. The amount of SF6 passing through the detector for each of these conditions was measured by a second detector operating under fixed conditions and connected to the outlet of the test detector. That the 100-volt dc polarization was sufficient t o prevent electron capture was checked simply by removing the source from the first detector; the second detector signal was the same with no ionization source in the first detector as it was with the 100-volt dc polarization. This procedure, which uses an infinite pulse interval, has other possible applications in gas chromatography and in gas-flow studies generally. The application of a rectangular dc pulse to a n electron capture detector through which a low concentration of absorbing substance is passing provides a n almost instantaneous method of loading the gas stream with a rectangular plug of test substance. The experiment illustrated in Figure 3 also confirms the relationship in Equation 3. The proportion of molecules ionized at a n infinite pulse period is not much more than that ionized at a period of 400 microseconds (Figure 1) so that in practice the pulse interval of 400 microseconds approximates t o infinity. The use of the electron capture detector as a gas-phase coulometer was finally checked by comparing the calculated quantity of SF6 with known quantities at concentrations in air down t o parts in 1012 by volume. A room 85 cubic 1964

meters in volume was provided with positive pressure ventilation at 47 liters per second; a fan was placed within the room to ensure the mixing of the air. SF6, 35 PI, was introduced into the room and samples were taken over a period of three hours. Figure 4 compares the concentrations of SF6 calculated from experimental measurements o n the basis that the detector was functioning coulometrically (circles), with the concentrations of SFe calculated from the known ventilation rate of the room (straight line). Apart from a scatter of points near the origin of the experiment, possibly due t o incomplete mixing of the added SFEat that time, the observed quantities follow faithfully the decay of SFc concentration by ventilation within the room. SFs was chosen as the test substance in this and other experiments because of its extreme chemical and physical inertness. With this substance the probability of losses by irreversible adsorption or surface reaction is minimal. Some measurements were also made by conventional gas chromatography. Table I1 lists some compounds with absorption coefficients found sufficient to provide coulometric responses. Reference t o past papers o n electron capture (3-5) and to our own past (3) J. E. Lovelock, in “Physical Processes in Radiation Biology,” pages 183-193. L. Augenstein, Ed., Academic Press, New York, N.Y., 1964. (4) C. A. Clemons and A. P. Altshuller, ANAL. CHEM.,38, 133 (1966). ( 5 ) J. 0. Watts and A. K. Klein, J. Ass. Ofic. Anal. Chem., 45, 102 (1962).

ANALYTICAL CHEMISTRY, VOL. 43, NO. 14, DECEMBER 1 9 7 1

Figure 4. Decay of SF6 concentration in a room used as an exponential dilution vessel Solid line calculated from exponential decay equation; circles calculated from experimental results assuming coulometric response

3

2

I

t, hr

results frequently reveal conditions under which 50% or more of the substance analyzed would be ionized in the detector; this historical information is useful in checking the applicability of coulometry in the gas phase. It is almost certainly applicable t o halogenated pesticides, for example; it also explains the difference in response factors for the same compound listed by different workers. With conditions favoring complete ionization, such as low flow rates and long pulse intervals, all response factors for intensely absorbing substances tend t o the same value, that of a molar coulometric response. Thus, if the relative response factors for a weakly absorbing standard substance are compared with a n intensely absorbing test substance under different conditions of analysis, it is perfectly possible t o have different assigned values for the response factor of the strongly absorbing substance owing t o this approach t o a 100% ionization. Where information is available o n the rate constant for the primary reaction between electrons and the molecular species of interest, such as in the papers of Wentworth and his colleagues (6), a direct estimate of the proportion of molecules ionized can be calculated. In general, rate constants in excess of 3 X 10-9 ml molecule- sec- are necessary for coulometric analysis. Figures 1 and 2, which show the calculated and observed forms of the relationships in Equations 3 and 4 for the detector, reveal the principal characteristics of detector be(6) W. E. Wentworth and J. C . Steelhammer, Adcan. Chem. Ser., 82, 75-99 (1968).

havior which previously have tended t o be obscure. At high flow rates and low ionization efficiencies, the detector tends toward a simple concentration-sensing device. At high efficiencies of ionization and low flow rates, the detector tends toward one which is coulometric and responds t o the rate of mass input rather than t o the concentration of the test substance. In Expression 3, K , the rate of destruction of the test substance, and V , detector volume, are seen as a product. This implies that t o detect very low concentrations of a n intensely absorbing substance, a large volume detector is t o be preferred. Thus, if trace quantities of SFs in air are sought, a detector 100 ml in volume and a column flow rate of 3 I. a minute would provide the same ionization efficiency as a detector 1 ml in volume with a flow rate of 30 ml per minute. The sample size for the large detector could be 100 times greater than that with a small one and a corresponding improvement in sensitivity is possible. The theory also suggests that if a number of detectors are connected in series and their output signals taken in parallel, ionization efficiencies close t o 100% could be realized. A single detector of open tubular construction with a coaxial anode could function in the same manner as a number of detectors in series, particularly if the ratio of its length to diameter was great. Such a detector would not only serve practically but also would bring to gas chromatography the concept “height of a theoretical detector!” RECEIVED for review July 9, 1971. Accepted August 16, 1971.

ANALYTICAL CHEMISTRY, VOL. 43, NO. 14, DECEMBER 1971

e

1965