Simple direct combination of gas chromatography and vapor phase

Determination of epoxy side groups in polymers. Infrared analysis of methyl methacrylate-glycidyl methacrylate copolymers. Swaraj. Paul and Bengt. Ran...
0 downloads 0 Views 640KB Size
Table V I . Double Bonds of Triglyceride Compositions for Cow Milk Fat, Bovine Fat, Rat Blood, and Rat Liver Cow milk Rat Carbon Double No.

bonds

Mol Wi

50 50 50 50 50 52 52

4 3 2 1 0 6 5 4 3 2 1

826 828 830 832 834 850 852 854 856 a58 860 862 872 874 876 878 880 882 884 886 888 890 900 902 904 906 908 91 0

52 52

52 52 52 54 54 54 54 54 54 54 54 54 54 56 56 56 56 56 56

0 9 8 7 6 5 4 3

2 1 0 5 4 3

2 1 0

fat

14 32 54

23 45 32

6 17 30 33 14

Bovine fat Rat blood

... 50 50

13 54 33

15 55 40

27 50 23

50 33 17

8 17 33 29 9 4

liver

20 41 32

7

13 39 30 8

9 12 19 31

25 4

12 19

27 17 15

10

zard method and GC-MS method. In the case of the Bezard fractionation method, C54 could not be measured because of its small quantity and high boiling point. But by the GC-MS method this can be easily measured-it provided data up to c56 in rat liver sample. These data generally agree quite well with those of the fractionation method.

The only great differences are the values for Cs-Clo-Clz ~ c30 T G (22.5 times), and of CZST G (20%), C S - C S - C ~of c14-c14-c16 of c44 T G (18.5 times). Since it was proved that the GC-MS method is useful for the study of triglyceride mixtures, we proceeded in the analysis of more complicated samples which require a higher column temperature, such as triglycerides in cows’ milk fat, bovine fat, rat liver, rat blood, etc. The results are shown in Table V. As for the positions of fatty acid combination of glycerine, Barber (10) and Lauer ( 2 1 ) reported that [M RCO&Hz]+ ion formed from positions 1 and 3 is useful. But this method is useful for only one type of TG and is not useful in the case where many TG’s of the same carbon number exists. We described the method to determine the degree of saturation of fatty acids in the former chapter. But the number of double bonds can be determined from the parent ions. Hites (12) found that the degree of saturation could be directly calculated from the M+ ions. The Hites method differs from the GC-MS method in the manner of measurement. In the Hites method, the TG mixture is directly introduced into a mass spectrometer and the peaks that do not overlap those of the M + ion are used for calculation. On the other hand, in the GC-MS method, since the TG mixture is fractionated by GC, the M + ion is not overlapped by the other mass fragments. Table VI shows the result of the GC-MS method. A comparison of Tables V and VI suggests a possibility of further extension of the application of this method. Table V shows, for example, that C54 TG of the rat blood Z Oof, c16-C18-c20, and 92% contains 2% of C ~ ~ - C ~ O - C6% of Cls-Cls-Cl~. Table VI shows, on the other hand, that the fatty acids of the C54 T G of the rat blood have two or more, up to seven, double bonds. Received for review November 6, 1972. Accepted February 12, 1973. Presented in part at the International Congress on Analytical Chemistry, 1972, Kyoto, Japan.

Simple Direct Combination of Gas Chromatography and Vapor Phase Infrared Spectrometry J. E. Crooks, D. L. Gerrard, and W. F. Maddams Epsom Division, Research & Development Department, B P Chemicals International Ltd., Great Burgh, Epsom, Surrey, England

A combined gas chromatography/infrared spectrometry system is described. The components of a sample are separated by gas chromatography and are passed separately into a heated multireflection gas cell. A spectrum is obtained for each component, using conventional scanning conditions, and while the spectrophotometer is running, the flow of carrier gas through the column is stopped. When the spectrum of a particular component has been obtained, the carrier gas flow is resumed and the next component passes into the cell. Spectra may be obtained for as’ many as five components in one sample without seriously impairing the column resolution. Useful

spectra are obtained from 100-pg quantities of most organic compounds which, for a total sample size of 10 pl, permits examination of components down to the 1% level. The system is readily assembled and the constituent gas chromatograph and infrared spectrometer may be used for other analytical work.

Although the combination of gas chromatography and high resolution fast scanning mass spectrometry has proved a very powerful weapon for the characterization of the components of rather complex mixtures of organic

ANALYTICAL CHEMISTRY, VOL. 45, NO. 11, SEPTEMBER 1973

1823

0 STOP VALVE

OUT

< I

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _'

'

I

r'

I1

I I

/ I

been marketed, but their overall performance characteristics, particularly the resolution, leave a good deal to be desired. The only published work using instruments of this type appears to be that of Bober and Burner (IO). The aim of the present work has been to make use of a conventional infrared spectrometer and to follow the dictum of Welti (6) to adapt'the gas chromatograph column to suit the requirements of the infrared spectrometer. This has been achieved by a much simplified version of the interrupted elution technique described by Scott et al. (11). The eluted fractions are examined in a heated multireflection gas cell which, for conventional packed GC columns, offers advantages over the minimum volume gas cell of the light pipe type. The overall objective was to design a system using, as far as possible, equipment to be found in a smaller spectrometry laboratory, which would give satisfactory spectra with 100-g samples of materials having moderately strong absorption bands. This objective has been achieved. EXPERIMENTAL Spectrometer and Gas Cell. A Grubb Parsons "Spectromaster" spectrophotometer fitted with an ordinate expansion accessory was used for all the measurements, in conjunction with a Grubb Parsons multireflection minimum volume gas cell with sodium chloride windows. This is of the type first described by White et al. (12) and has three mirrors, one fixed and two movable, by means of which the optical path length may be varied. The maximum path length obtainable is 100 cm, but a compromise has to be made between increasing path length and decreasing transmission. The volume of the cell is about 40 cm3 but approximately one half of this is dead space, not traversed by the radiation. As received, it had a heater mounted in the base plate which gave a ceiling temperature of 110 "C. This is too low for general use, with materials of relatively low volatility, and a 150-W cartridge heater in a shaped Dural block was fitted to the side of the cell. This increases the ceiling temperature to 220 "C, and no serious problems have been encountered in running the cell a t this level over an extended period of time. The Neoprene gaskets gradually deteriorated and were replaced by ones cut from Viton sheet. Gas Chromatograph and Ancillary Equipment. A Pye Series 104 Model 44 chromatograph has been used, thus permitting the use and ready interchange of standard columns. The interruption to flow (parking) of the column is achieved with a F'ye 6-port valve mounted in the column oven. Katharometer detection is used, permitting the whole of the sample stream to pass to the gas cell. The connection between the column and the cell is 70 cm of 3-mm 0.d. 24-gauge stainless steel tubing, heated by resistance wire and lagged with glass fiber tape. It is held at 250 "C.There is a T-piece connection to a line supplying a rapid flow of helium for flushing out the gas cell. About 20-cm length of this flush line is heated to prevent condensation of eluted material in the dead end. A block diagram of the overall system is shown in Figure 1.

Philpotts, "Proceedings of the Tenth Colloquium Spectroscopicum Internationale." E. R. Lippincott and M . Margoshes, Ed.. Spartan Books, Washington, D.C., 1966, p 577. (2) J. G. Grasselli and M. K. Snavely in "Progress in Infrared Spectroscopy." VOl. 3, H. A . Szymanski. Ed., Plenum Press, New York, N.Y., 1967, p 55. (3) M . D. D. Howlett and D. Welti, Analyst (London),91, 291 (1966). (4) I. A. Fowlis and D. Welti, Analyst (London), 92, 639 (1967). (5) H. Copier and J. H. Van der Maas, Specfrochim. Acta, 23A, 2699

RESULTS AND DISCUSSION Spectrometer Performance. The general performance of the spectrometer was examined a t several scanning speeds, to determine if the quality of the spectra deteriorated a t faster scan rates because of the time constants of the servo beam balancing system and the chart recorder. At the highest rate, three minutes to cover the range 2 to 5 pm and an additional two and a half minutes for the interval 5 to 15 p m , some loss of performance was discernible. With a scan rate of one half of the maximum value, that is eleven minutes to cover the range 2 to 15 pm, no loss of resolution or foreshortening of bands was evident. This is therefore the optimum speed.

(1967). (6) D. Welti, "Infrared Vapour Spectra." Heyden & Son Ltd., London, 1970, p 45. (7) A. M . Bartz and H. D. Ruhl, Anal. Chem., 36, 1892 (1964). (8) R. A. Brown, J. M . Kelliher, J. J. Heigl. and C. W. Warren, Anal. Chem., 43,353 (1971). (9) P. A. Wilks and R. A. Brown, Anal. Chem., 36, 1896 (1964).

(10) H. Bober and K. Burner, Fresenius, 2. Anal. Chem., 238, 1 (1968). (11) R. P. W . Scott, I . A Fowlis, D. Welti, and T. Wilkins, "Gas Chromatography 1966. ' A. B LittleWood, Ed., institute of Petroleum, London, 1967, p 318. (12) J. U. White, N . L. Alport, W . M . Ward, and W . S. Gallaway, Anal. Chem. 31, 1267 (1959).

(1) A. R.

1824

ANALYTICAL CHEMISTRY, VOL. 45, NO. 11. SEPTEMBER 1973

Gas Cell Performance. The interdependence of the path length and the transmission was measured and the results zre given in Table I. These values were obtained at a cell temperature of 110 "C and although the performance is somewhat temperature dependent, the process of manual adjustment is more conveniently done a t 110 than a t 220 "C. The path length of 76 cm, which corresponds to five traversals between the mirrors plus, entrance and exit paths of 8 cm, clearly poses no spectrometer operational problems because attenuation of the reference beam to give 38% transmission still leaves ample energy. If additional sensitivity is required, satisfactory results may be obtained with 88-cm path length and 20% transmission. The transmission is somewhat wavelength dependent, as a result of emission from the hot cell, but this has not caused practical problems. Similarly, although the additional 22 cm of path length in air in the sample beam, the distance traversed by the radiation in that part of the optics not within the gas cell, leads to some absorption from water vapor and carbon dioxide, this is not large enough to prove troublesome. The time required to flush the cell is clearly of considerable importance in relation to the spectrometer scanning speed and the operating conditions of the GC column. The first approach was to use the carrier gas flow for this purpose. The efficiency was measured by filling the cell with a mixture of 1% of carbon dioxide and 99% of helium and flushing with pure helium a t the usual carrier gas flow rate of 60 cm3 per minute. The disappearance of the carbon dioxide was monitored by infrared spectrometry and complete flushing required two and a half minutes. When a rapid helium flow from a separate line is used, flushing is ccmplete in about ten seconds, a time which is negligible by comparison with that required for scanning the spectrum. Subsquently, the flushing of a wide variety of samples, many of them high boiling polar materials, has been followed by observing the disappearance of the C-H stretching absorption at 3.4 pm. In no case has the required flushing time exceeded twenty seconds, and in the majority of cases it is significantly less. Hence, there are no problems from memory effects. Gas Chromatograph Performance. The conclusion drawn from the optimization studies on the spectrometer performance is that there should be an interval of approximately twelve minutes between the collection of successive fractions in the gas cell. It is therefore necessary to park the GC column for this time interval, and to repeat the process several times if a multicomponent mixture is being examined. This must not lead to a significant loss of resolution, and the operating conditions must be chosen to deliver the unknown in a volume of carrier gas comparable to the volume of the cell. The possible loss of resolution from this parking procedure was studied with a number of test mixtures. For example, the four-component system methyl hexanoate, methyl octanoate, diethyl malonate, and methyl benzoate was run on a nine-foot column of 10% 2,2,6,6-tetra(nbutyl-9-propionate)cyclohexanoneon Anachrom AB a t 125 "C. Under these conditions, the respective retention times were 5.9, 17.7, 21.9, and 29.1 minutes. In a repeat run, the column was parked for ten minutes after the emergence of the methyl hexanoate peak and for an additional ten minutes after the appearance of the methyl octanoate peak. The peak widths of the diethyl malonate and methyl benzoate were unchanged and, hence, the resolution of the column was unimpaired. Subsequent experience with a wide range of mixtures has shown that a t least five parking operations are possi-

Table I. Path Length and Transmission Values for the Grubb Parsons Multireflection Gas Cell Path length, c m

Transmission, %

100 88 76 64

5 20 38 55

ble before there is a significant loss of resolution. If a particularly complex mixture is being examined, it may be necessary to make two separate runs, to deal respectively with the earlier and later groups of peaks. There is sufficient flexibility in the GC operating conditions to meet most requirements and, in particular, the one that the peaks shall be reasonably sharp, to keep the volume of effluent to about 40 cm3, can be met by the use of short columns. a t higher temperatures or by temperature programming. The 6-mm 0.d. columns have been used, as narrower columns do not have the load capacity required to meet the sensitivity criterion laid down for the technique. In general, the conditions used may not be optimum for the highest efficiency but, nevertheless, no difficulty has been encountered in adapting the GC conditions to meet the relatively inflexible operating parameters of the spectrometer and gas cell. It is also necessary to assess when the sample concentration in the gas cell reaches a maximum; this will occur after the peak height reaches its maximum on the GC trace. The sample concentration in the gas cell is conveniently monitored, in most instances, by setting the spectrometer to 3.4 pm and observing the intensity of the C-H stretching band. When this reaches a maximum, some thirty to forty seconds after the maximum in the GC trace, the cell is isolated from the column, which is then parked. It is preferable to switch off the katharometer a t this juncture and, if temperature programming is being used, the programmer is simultaneously switched from start to hold. The performance of the overall system was first assessed with the four-component ester mixture used for the studies on column parking. An initial qualitative run, which gave very satisfactory spectra, was followed by a quantitative study. One microliter of a mixture of methyl hexanoate, methyl octanoate, diethyl malonate, and methyl benzoate in equal proportions was injected and the spectra of the emergent 0.25-pl components were recorded. The spectrum obtained for the methyl benzoate, shown as Figure 2, is typical. It is clear that, using ordinate expansion, a spectrum adequate for characterization purposes may be obtained from 20 pg of methyl benzoate. High quality spectra of more than adequate intensity are obtained from these 0.25-p1 components and there is no difficulty in working a t the 1OO-pg level without resort to the spectrometer ordinate expansion accessory. Thus, with a 10-pl sample, 1%components may be examined. The system has subsequently been used for the successful characterization of a wide range of unknowns, and confirmation has been forthcoming that 100 pg of most compounds will give an adequate spectrum. No difficulty has been experienced in dealing with materials covering a very wide range of boiling points. At the lower end of the range, a mixture of unsaturated C4 hydrocarbons, which could not be completely characterized by GC/mass spectrometry because of lack of specificity in the cracking patterns of the components, was examined. The four major components were shown to be 1,2-butadiene, 1,3-butadiene, butyne-1, and vinyl acetylene. By contrast, the tech-

ANALYTICAL CHEMISTRY, VOL. 45, NO. 11, SEPTEMBER 1973

* 1825

1.0 Od

0.6

0.4

Y

II *

0.2

0

6

i

4

4

ib

ii

r2

(3

IS

1%

WAVE LENGTH I N p m

Vapor phase infrared spectrum of a 250-pg sample of methyl benzoate separated from a four-component mixture

Figure.2.

nique has proved useful for dealing with the small amounts of high boiling material which sometimes exude from plasticized polymer films. These can be washed from the surface with a suitable low boiling solvent, most of which is subsequently evaporated, and injected onto the column. A typical unknown gave three satisfactory spectra, those of dibutyl phthalate and two high boiling secondary alcohols. No problems have been encountered with polar polyfunctional molecules; a range of compounds containing twelve carbon atoms and up to four oxygen and/or nitrogen atoms has been examined. Contrary to what has sometimes been stated the interpretation of vapor phase spectra, particularly of higher molecular weight compounds where rotational fine structure is absent, presents no difficulties. Indeed, there are advantages in some instances because there is no interference from solvent bands. Welti (6) has provided spectra of three hundred and six organic molecules, covering a wide range of functional groups, in the vapor phase at elevated temperatures, and he has also written a useful account of frequency correlations among vapor spectra. His assertion that vapor phase spectra are as useful for the characterization of organic compounds as solution spectra has been confirmed by the present work. The success achieved with the simple system described above is primarily the result of three decisions, to use a conventional general purpose spectrometer operating under normal conditions, to adapt the operating conditions of the gas chromatography equipment to meet those of the spectrometer, and to examine the vapor phase fractions in a heated multireflection gas cell. All three are compatible with the primary requirement that the system shall use a minimum of equipment not available in a modestly equipped analytical laboratory. This arrangement has the additional advantage that the two building blocks, the spectrometer and the GC equipment, can be and are used independently for other work, as required. The ability to adapt the GC operating conditions to meet those of the spectrometer has been greatly facilitated by the success of the simple stop/start elution technique which, although clearly of wide applicability, was little used prior to the present work. 1826

The value of the multireflection gas cell, by comparison with that of the more widely publicized light pipe, has not been generally appreciated. In some measure, this is the result of the tendency to assess both types of accessory in terms of the path length/volume ratio. This parameter is only meaningful if the volume of the cell is related to the volume of sample being examined, so that the total amount of sample in the radiation beam is considered. This can be appreciated in terms of a specific example. Typically, with a 4-mm i.d. packed column, a fraction will contain 5 pg/cm3 in a total volume of 40 cm3, Hence, 200 pg of unknown is available but, because half of the sample in the gas cell is not traversed by the radiation, the effective amount of material is 100 pg. In the case of a light pipe used in conjunction with the same type of packed column, the volume is small by comparison with that of the carrier gas plus sample and it is clearly expedient to fill it at a point on the time scale corresponding to the GC peak maximum. Here the sample concentration will be approximately twice that over the peak as a whole, that is about 10 pg/cm3. The volume of the light pipe used in the Wilks Model 41 system is 0.6 cm3 and the weight of the sample in the radiation beam is 6 pg. This difference is, of course, a direct consequence of both the smaller path length and volume of the light pipe. To give the same absorbance as the multireflection gas cell with a 76-cm path length, a light pipe with a 38-cm path length is required. In terms of the Wilks light pipe, for which the optical path length is very little greater than the geometrical path length, this would increase the volume to about 4 cm3 and fractions one order of magnitude smaller than a t present could be examined. However, a light pipe with a geometrical path length of 38 cm is clearly impracticable for use with conventional spectrometers. In the case of a support coated open tube column or a capillary column, where the gas flow is slower and the efficiency is greater than that of a 4-mm packed column, the light pipe is to be preferred because the maximum sample size is very much smaller and the peak volume is of the same order as the volume of the light pipe. However, the overall sensitivity of the light pipe, whether used in conjunction with packed columns or support coated open tube columns, is appreciably lower than that of a multireflection gas cell used in conjunction with a 4-mm i.d. packed column. Unless the efficiency of light pipes can be improved to the point where the optical path length is much greater th,an the geometrical path length, the superiority of the multireflection gas cell is undeniable. The performance characteristics of this cell are well suited to those of a typical GC packed column and the combination is ideal for the requirement that useful spectra are to be obtained from 100-pg samples. Indeed, the figure can be set at 20 pg in many instances. This, of course, is by no means the ultimate sensitivity which may be achieved by vapor phase measurements or by the more conventional condensation techniques. As Freeman (13) has pointed out in his useful review, the literature is replete with reports on ancillary techniques and modified instrumentation whose aim is to obtain interpretable infrared spectra on small samples. In considering these various approaches, it should not be forgotten that there must always be a compromise between simplicity, sensitivity, speed, and capital outlay. (13) S. K. Freeman in "Ancillary Techniques of Gas Chromatography,'' L. S. Ettre and W. H. McFadden, Ed.. Wiley-lnterscience, New York, N. Y., 1969, p 227.

ANALYTICAL CHEMISTRY, VOL. 45, NO. 1 1 , SEPTEMBER 1973

At one end of the scale, interferometric spectrometers provide high sensitivity and scanning speed but are very expensive. Specialist spectrometry laboratories using conventional dispersion instruments with a range of accessories come in an intermediate position. This approach is typified by the work of Freeman (13) who, using an ultramicro cavity cell and a beam condenser, has obtained a good spectrum from 1 pg of eugenol in solution in 1pl of carbon tetrachloride. In this type of work, the unknown from the GC separation is condensed into a glass capillary and washed out into the ultramicro cell with the requisite small volume of solvent, and sensitivity is gained at the expense of sample handling time. It is also possible to trade spectrometer performance for scanning speed, for “on-the-fly” work, and this point is well covered by Freeman (13).

The objective of the present work, as noted at the outset, has been to provide a simple system based, as far as possible, on the equipment to be found in a smaller spectrometry laboratory, for use by staff who do not specialize in infrared spectrometry. In these circumstances, ease of operation, particularly in the handling of small samples, is of primary importance, and this has led to the vapor phase approach. The sacrifice made in terms of sensitivity is not a serious one in practice; the ability to obtain useful spectra from 1oO-Ig samples and 20-pg samples in many instances has proved adequate for the great majority of the problems encountered. Received for review January 10, 1973. Accepted March 27, 1973. Permission to publish this paper has been given by The British Petroleum Company Limited.

Multicolumn Radio-Gas Chromatographic Analysis of Recoil Tritium Reaction Products Darrell C. Fee and Samuel S. Markowitz Department of Chemistry and Lawrence Berkeley Laboratory, University of California, Berkeley, Calif. 94720

A general radio-gas chromatographic analysis system has been developed for hydrogen and C1 to C7 alkanes and alkenes. Peaks were monitored at a constant flow rate in the same detector (a 0 proportional counter) and the injection volume was large. A system of four columns used in series gave adequate resolution of more than 20 products from the whole sample. This system was a combination of sbp-flow, recycle, center-cut, stepwise temperature programming, and stepwise pressure programming techniques.

Gas chromatography has been widely applied in the separation and analysis of multicomponent systems. If the components are radioactive, the effluent from a chromatographic column may be mixed with a counting gas and the radioactivity measured as the mixture flows through an internal proportional counter ( I ) . This immediate radio assay is called radio-gas chromatography. The radiogas chromatographic analysis of tritium-labeled hydrocarbons is of particular interest to us. We are studying the reactions of recoil tritium atoms. There were several a priori considerations for the design of a general radio-gas chromatographic analysis system for the products of recoil tritium reactions. (1) The expected (tritium-labeled) products differed widely in boiling points and physicochemical properties. The expected products ranged from HT and CH3T to the tritiated parent hydrocarbon (we intended to eventually study the recoil tritium reactions of cyclohexene and methlycyclohexene) and included nearly every straight chain alkane-t and alkene-t species in between (for a review of recoil tritium reactions, see Ref. 2 and 3 ) . In addition, we wanted to separate the methylcyclohexene-t isomers. (This would (1) R. Wolfgang and F. S. Rowland, Anal. Chem., 30,903 (1958).

determine whether or not direct T-for-H substitution was accompanied by a shift of the double bond (2-4).) A normal “boiling point” column would not separate 3-methylcyclohexene-t from 4-methylcyclohexene-t. The three methylcyclohexene isomers had been individually resolved on a saturated silver nitrate/ethylene glycol column (5). The methylcyclohexene-t isomers and the smaller tritiated alkenes from recoil tritium reaction would be individually resolved on a saturated silver nitrate/ethylene glycol column. However, all alkane-t species would emerge as one peak from such a column (6). This suggested an aliquoting procedure. The tritiated alkenes and the methylcyclohexene-t isomers could be assayed using one aliquot. The tritiated alkanes could be assayed using another aliquot. Upon further consideration, we decided that no aliquoting procedure would be possible. Aliquoting might lead to unequal fractionation of low vapor pressure parent compounds. Consequently, we decided to inject the entire sample at once. The typical gaseous sample was contained in a glass capsule, 6 cm long with an internal diameter of 1.5 cm. [The dimensions of the capsule are fixed a t such large values to minimize the loss of recoil tritons to the capsule wall following the 3He(n,p)T reaction (7).]The glass capsule would be mechanically crushed directly in the stream of the chromatograph. This led to (2) a large sample injection volume. The sample is initially distributed throughout the 2O-cm3 volume of the mechanical crusher. This sample volume is swept onto the gas chromatographic column in about 100 sec, assuming typical radio-gas chromatographic flow D. S. Urch, MTP (Med. Tech. Publ. Co.) Int. Rev. Sci.: lnorg. Chem., Ser. Eight, 1972,149. (3) R. Wolfgang, Progr. React. Kinet., 3,97 (1965). (4) H. Bieler, F. Battig, and P. Jordan, 2. Phys. Chem. (Frankfort am Main), 72, 1 (1970). (5) E. Gil-Av, J. Herling, and J. Shabtai, J. Chromatogr., 1, 508 (1958). (6) M. E. Bednas and D. S. Russell, Can. J. Chem., 36,1272 (1958). (7) J. W. Root and F. S. Rowland. Radiochim. Acta, 10,104 (1968).

ANALYTICAL CHEMISTRY, VOL. 45, NO. 11, SEPTEMBER 1973

1827