AcaJAXa is independent of concentration. This is a n important consideration since i t enables the number of carbon atoms in the n-alkanal derivative to be established without determining the concentration of the unknown or duplicating exactly the conditions by which the sample KBr pellet is prepared. It is often difficult t o establish the identity of compounds isolated from natural products because suitable reference compounds necessary for comparison with the unknown may not be available. This is particularly true in the case of the higher molecular weight carbonyls. A distinct advantage of the present technique is that the ratio value provides information useful in establishing the identity of n-alkanal derivative without the need for obtaining or preparing costly or otherwise unavailable standards. For example, a calibration curve similar to that shown in Figure 2 can be constructed by determining the Aca,/ANa values of a fen* representative n-alkanal 2,4-DSPHJb. These ratios are plotted against chain length and any intermediate, and higher or lower values are then established by extrapolation of the curve, since as 5hoM-n by
Figure 2, the curve is linear from CI to C16. To use the curve in the identification of a n n-alkanal 2,4-DSPH, the unknown derivative is merely mixed with 150 mg. of potassium bromide, the pellet is prepared, and the ratio of the CHz to X H absorbance is determined from the infrared spectrum as previously described. The identity of the n-alkanal is obtained by comparison of the observed ratio with these standard values. -41ternately, if the parent carbonyl is branched or if the unknown can not be established as t o its class (ketonic, olefinic, etc.), an excellent approximation of the number of methylene groups prcwnt in the molecule can be made by using the right hand ordinate of the curve shown in Figure 2. ;\n example of this is shown in Table 111 in which the CH2/K” absorbance ratio was used to determine the number of methylene groups in three representative 2,4DSI’H’s other than the n-alkanalsLe., 2-hexenal, 9-undecenal, and 4heptanone. These data suggest that the CH2,:K€I absorbance ratio is not only applicable in differentiating 2,4DSPH’s of the saturated aliphatic aldehydes but can also provide informa-
tion useful in establishing the chain length of other carbonyl compounds (10). LITERATURE CITED
(1) .411en, C. F., J . Am. Chem. SOC.52,
2955 (1930).
( 2 ) Bellamy, L. J., “The %frared Spectra
of Complex Molecules, 2nd ed., pp. 14, 15, Wiley, Kew York, 1959. (3) Ibid., p. 251-2. ( 4 ) Braude, E. A., Jones, E. R. H., J . Chem. Soc., 1945, 498. (5) Corbin, E. A., Schwartz, 0. P., Keeney, M., J . Chromatog. 3, 322
( I m). (6) Gordon, E., Wopat, F., Burnham, IT., Jones, L., AKAL. CHEIM.23, 1754 \ - - - - ,
(1951). (7) Jones, L. A, Holmes, J. C., Seligman, R. B., Zbzd., 28, 191 (1956). (8) Pippen, E. I,., Xonaki, h I , Jones, F. T., Stitt, F., Food Res. 23, 103 (1958). (S).Shri?er, R. L., Fuson, R. C., “Identification of Organic Compounds,” 3rd ed., Wiley, Kew York, 1948. (10) Stitt, F., Saligman, R. B., Resnik, F. E., Gong, E., Pippen, E. L., Forss, D. A., Spectrochzm. Acta 17, 51 (1961). (11) Timmons, C. J., J. Chem. Soc., 1957, 2675. RECEIVED for review December 21, 1962 Accepted July 5, 1963.
A Rapid and Simple Method of Deuterium Determination EDWARD M. ARNETT and PETER McC. DUGGLEBY Department of Chemistry, University o f Pittsburgh, Pittsburgh 7 3, Pa.
b An improved procedure for generating HD-H2 from mixed isotopic water samples for deuterium determination b y thermal conductivity uses standard gas chromatography apparatus. The method is at least as accurate and precise as other common techniques for deuterium assay and is much faster, more convenient, and easily learned. It does not require purification of the water and may b e used routinely over the entire range of D 2 0 - H 2 0solutions down to background levels. Safety precautions for handling hydrogen as a GLC carrier gas are described.
S
years ago we published a preliminary description ( 1 ) of a n apparatus for generating H2-HD mixtures from isotopic water samples for subsequent determination of deuterium content by gay chromatographythermal conductivity procedures (f-4, 6, 8-1a). I n t h e meantime, we have increased t h e speed, convenimce, sensitivity, and accuracy of t h e method cori~idern1)lv anti give n t l t 6 I i i t ive &wription herc. The over-all principle of the technique is very simple. EVERAL
1420
ANALYTICAL CHEMISTRY
A small aliquot of water containing
HzO, HOD, and D20is delivered into a sample of calcium hydride in an evacuated tube. Since the selectivity for the reaction of calcium hydride with OH and OD bonds is practically unity ( 5 ) , the relative quantities of Hz and H D that are formed from the reaction correspond exactly to the relative quantities of hydrogen and deuterium in the water sample (no Dz is formed, of course). A portion of the gases from the generation step is passed into, and measured in, a standard GLC gas sampler at atmospheric pressure. From here it is released into the stream of a standard GLC apparatus using hydrogen as carrier gas (9,9, 10). After going through a short column of activated charcoal to remove any volatile impurities, the gas mixture enters the thermal conductivity cell and, since hydrogen is the carrier gas, only the H D content of the gas sample is detected by the cell. The size of the H D peak in the chromatogram recording is directly proportional to the deuterium content of the original water sample. C‘onvcniivwc a i d awiirucy ha[,? been improvctl by mutlifiriitions of the apparatus. The use of twin generating tubes, one of which coiitains hydrogen
froin a standard HzO-D20 solution, provides a convenient means of Calibrating the instrument at any moment to correct for short-term fluctuations in operating conditions. Almost all of t h e glass stopcocks in the original apparatus have been replaced with leakproof fixtures, eliminating a major source of nuisance and error. I n particular, t h e water sample is now introduced directly into the calcium hydride by means of a microhypodermic syringe which is thrust through a serum stopper in the top of the generating tube. Most of the stopcocks in the system for handling the gas have been replaced with pressure-tested toggle valves. Replacement of the original katharometer by a Burrell KD thermal conductivity cell has increased the sensitivity about tenfold. The use of a ball and disk integrator permits precise comparison of the peak areas obtained, giving a better criterion for deuterium content than the peak heights used previously. Thanks to these innovations, the perfectly 1ine:tr relation41ip between peak : ~ X Laiitl 1no1ti pcr mit tleuterium (above background) suggested in our original paper has been confirmed for
Generatino
Bu17:
t
I 7J
Leveling
:m
11
Silicone Rubber Diaphragm Mug GENERATING CARTRIDGE
Rubber Stopper
Figure 1.
Diagram of gas-generating system
the entire range from about 100% D20 down to the background. Particularly. in the important region for samples containing from 1 to 10 mole % DzO, the average relative standard deviation from a constant ratio for peak area (A) + mole % DzO is = t O . l % , The convenience and speed hare been improved to the point n-here an unskilled worker can learn t o make analyses within a day and can analyze 20 to 30 unknowns in an 8-hour day. EXPERIMENTAL
Apparatus and Chemicals. The apparatus used for generating and handling the gas sample is depicted in Figure 1. We h a l e used models made entirely of glass with glass stopcocks or alterntttively made of steel hypodermic connecting tubing and vacuum-tight toggle valves. The latter arrangement is better, although more difficult to build. From the gas sampler the gas strean was conducted through a 1-meter column of activated carbon a t a flow rate o f about 40 ml. per minute. Although a grade 11-1-1 carbon from Uarnebey Cheney Co., Columbus 19, Ohio, was used in these experiments, other grades f'; carbon should nork equally well. rhe column mas permitted to remain z t room temperature. I t was conditioned overnight a t about 300' C. to remove any- adiorbed moisture, hut follon irig thiy initial treatment it could be uscd for moiith. without further care. The choice of column length was g,overned by the need for a short relention time for
speedy analyses, while still permitting a satisfactory separation of the H D peak from that of any impurity. The only function of the column was to separate volatile contaminants from the mixed H r H D sample, and no attempt was made to separate these isotopic species from each other. The main gaseous impurity was air from small leaks and this was easily separated by the 1-meter column. The appearance of an air peak served as a valuable check for leaks in the system. After traversing the column, the stream passed through a Burrell K D thermal conductivity cell. Since hydrogen was used as the carrier, it was possible to operate routinely a t a very high filament current of 390 ma., the temperature of the cell being thermostated a t 100' C. I n working near the background level where base line noise could cause serious errors, the entire apparatus was enclosed behind curtains in order to help maintain constant temperature. The output from the thermal conductivity cell went to a standard Alinneapolis-Honeywell 1-mv. recorder equipped with a disk chart integrator Model KI-4. The linearity of the integrator rras checked frequently and correction factors for the attenuator on the recorder were determined over the whole range of attenuation from 1 to 4096. Powdered calcium hydride was obtained from RIetsl Hydrides, Inc., Beverly, Mass., frebhly sealed in polyethylene envelopej coiitainirig nbout 10 grams ~ ' e r cuvt.lol)c. Thehe were handled in a dry box as described below. Hydrogen for use as carrier was a standard electrolytic grade obtained
froin Air Reduction Gorp. I t mas passed through a purifier tube containing dried molecular sieves to remove moisture. No other purification was used. Standard deuterium samples were prepared from 99.5% DzO sold by Isotope Specialties, P. 0. Box 688, 170 W. Providencia, Burbank, Calif. The purity of this material was checked by a density measurement on a large sample (?) and an independent analysis by J. Xenieth, 303 W. Washington, Urbana, 111. Samples were prepared by successive dilution of DzO with distilled water from the same sample to prevent possible errors from changing background concentrations (an unnecessary precaution, as it turned out). Since D20 is only about 10% heavier than R20( 7 ) , it is very difficult to determine small concentrations of D20 accurately by density measurements. We therefore used high precision volumetric ware, which would introduce a considerably smaller error in standardization of solutions than could be detected by most analytical techniques, including the one discussed here. Since the density difference between DzO and I&O corresponds almost exactly to the percentage difference in molecular weight, preparation of solutions by volume per cent gives the mole per cent composition directly. Safety Precautions. Since hydrogen must be used as the carrier gas, a number of precautions were taken against explosion. If these are followed there should not be much more danger in the use of hydrogen than in the ube of illuminating gas. I n our opinion, the best over-all safety step is to keep the entire apparatus in a walk-in hood so t h a t any possible leaks will not result in accumulation of hydrogen. Since the only purpose of the hood is to remove hydrogen, a simple canopy of plywood or other inexpensive material suspended over the equipment and backed by a window with an inexpensive exhaust fan a t the top is adequate. M'e have also employed as another protective device a solenoid valve to shut off the flow of hydrogen from the tank in case of sudden leakage while the system is not being watched. A Ross Solenoid valve (Green Model 1371A2017, normally open) with electric reset was attached by a suitable nipple directly to the reduction valve on the hydrogen tank. This could be activated by the output from a Thermocap capacitance relay, the impulse wire of which was wrapped around one of the legs of the mercury-filled U-tube flowmeter used to measure the hydrogen flow. A sudden increase in hydrogen flow due to a broken line would actuate the relay and cause the solenoid valve to shut off hydrogen flow completely until started again by the operator. The power supply to the filaments in the detector cell was also operated by a s~ itch through the rapacitance relay w that the filamcnt~ \\ciultl riot lie burriecl out if the 111tlrcigcii flow were stopped. Actually, the back pressure from a short column of the coarse VOL. 35, NO. 10, SEPTEMBER 1963
1421
~~
a
Table 1. Conditions for Analyses Deuterium, Cell size, Sample Detector, mole % ' ml. size, pl. ma. 100-5 1.5 20 300 5-0.5 1.5 20 350 0.5-0.2 350 6 35 0.2-0.1 10 38 350 0.1-0 10 38 390 9-mm. 0.d. tubing. 18-mm. o.d. tubing.
grade of charcoal used in this work is so small that a broken gas line would not result in a leak of more than 100 or so milliliters of hydrogen per minute. This would probably be dissipated so rapidly by diffusion that an explosive mixture would not be approached in a room of reasonable size and activity. Preparation of Calcium Hydride Cartridges. The key point in the analytical procedure is t h e generation of hydrogen isotopes from t h e mixed water sample. There is considerable evidence t h a t this is the major source of uncontrollable experimental error. For best results, freshly powdered calcium hydride that has never been exposed to moisture should be used and protected against humidity during the packing of generating cartridges. A large number (perhaps 100) of 4-inch lengths of 8-mm. o.d. glass tubing are heated in the middle and pulled out to yield a pair of 2-inch lengths, each sealed a t one end, from each of the original 4-inch pieces. I n the sealed end of the tube a small wad of hydrophobic cotton is placed to prevent the blowing out of hydride during the generation step. The cotton is prepared by treating ordinary cotton with a chlorosilane to remove adsorbed moiature which would gradually react with hydride after packing. For best results the use of hydrophobic cotton is mandatory and is far superior to glass wool or ordinary cotton. The cartridges are now dried extensively in a vacuum oven and removed to a dry box to cool. There they are subsequently filled nearly to the top with powdered calcium hydride, capped with silicone rubber diaphragm plugs (Burrell Corp. silicone seals, puncture type, Catalog No. 261-9-01), and stored in a desiccator inside the dry box until required. Analytical Procedure. After the cell and recorder are stabilized to give a good base line at t h e expected attenuation, a cartridge is removed from t h e desiccator and pushed part way through a rubber stopper as shown. The end of the seal is now crushed with a pair of pliers and t h e stopper seated tightly in t h e top of t h e generator tube. JVith the vent valve closed and t h e generator valve open, the leveling bulb lowered to its base position, and stopcock B rotated to 90" from the position shown so that the sampler tube may be evacuated, the vacuum valve is opened to permit evacuation of the system. When the mercury has risen to maxi-
1422
ANALYTICAL CHEMISTRY
Generator Small'
Smalla Largeh Largeb Large*
mum height, the vacuum valve is closed. If the mercury does not drop a t all within 5 minutes, the system may be considered vacuum tight and the generator valve closed. If, however, it falls slowly, this indicates the presence of a leak which must be eliminated before proceeding further. The most likely source of kaka is stopcock B or borne other part of the pa. sampler assembly since the usual arrangement of glass stopcocks i b prone to leak under vacuum and must be kept well greased. Other likely sources of leaks are: the seat of the silicone plug, ball joint C, and the joint of the rubber stopper with the top of the generating tube. If the apparatus is made of hypodermic tubing with pressure-tested toggle valves for the generator, vent, and vacuum valves, there should be no difficulty. In general, it is qafest to use a fresh silicone diaphragm plug with each cartridge, although we have used them four or five times without this becoming a source of leaks. To generate a sample, 15 to 30 pl. of mixed isotopic water are drawn into a clean, dry Hamilton 50-111. syringe with a ll/*-inch needle and a gastight barrel. After any air in the syringe has been displaced, the needle is thrust through the plug and well domn into the calcium hydride. The sample of water is then forced out into the hydride. This injection may be fairly rapid, although if it is too fast the diaphragm plug may blow out. Aside from this, no particular care is needed in this step. S t this point the mercury will have been displaced by the gas uhich will have been formed. The generator valve is now opened and the gas sampler filled with the generated gas while pressure is brought to slightly above atmospheric by raising the leveling bulb. The pressure is now reduced to atmospheric by gently opening the vent valve and closing i t when the level in the bulb and in the generating tube become equal. After the vent valve and the generator valve are closed, the sample is delivered to the gas stream by rotating stopcock B to trap an aliquot of gas in the sampler, and then stopcock d is opened. Previously undetected air leaks will appear on the final recorder tracing. Considerable flexibility is possible through variation of the size of the sampler tube. Best results will be obtained if the sample of water to be used is as small as possible, the limiting factor being the requirement of a peak of reasonable size in the final chromato-
gram. The size of water sample and gas sampler should be adjusted so that four or five aliquots of gas can be taken from each generation and the peak height will be at least half the width of the chart paper. I n analyzing a sample containing a high deuterium titer, it will be possible to UT a small (1- to 5-ml.) sampler and still use a small (15-PI.) sample of mater. For samples containing very little deuterium one may need to use larger amounts of water (perhaps 40 pl.) and a 20-nil. sampler. 911 that is necessary is t o use enough water to gile a good-sized peak in the recording a t the lowest attenuation possible. I n the normal range of deuterium analysis of labeled carbon conipouiidq, deuterium level, arc apt to be betneen 1 and 10%. Often the sample available from combustion of labeled compounds i, smaller than 15 p1. I n such a case the sample should be accurately diluted with distilled n ater (usually containing about 0.016 mole % background DzO). A ten-to-one dilution v, ill still maintain a high enough deuterium level for reasonably accurate analysis, even though the original sample might have been only 1 pl.
I n Table I are shonn the conditions n e used for performing analyses on samples TI ith different deuterium levels. DISCUSSION AND RESULTS
Contamination of Sample. A major difficulty with some of the most frequently used methods of deuterium determinations ia their sensitivity to small amounts of contaminants. This often requires repeated distillation of small quantities of water. The present method appears to eliniiiiatc this difficulty completely. Volatile contaminants are separated by the carbon column, n hile nonvolatile oneb remain in the hydride cartridge. Table I1 demonstrates that contamination of the water does not influence the analytical precision. Here, samples of our qtandard solution containing 4.97 mole % added D20 mere contaminated with 5y0by weight of the compounds shown. The average normalized peak area and Cvalue for this solution are, respectively, 2515 and 506 with a standard deliation s representing h1.1770 of these ~ a l u e for 10 D.F. The C values obtained from contaminated d u t i o n s agree within experimental error n i t h the reference solution, although the experimental error for contaminated solutions appears t o be slightly greater. Contaminants containing exchangeable hydrogens could not be used in such a test. X few preliminary trials wggest that deuterium levels in urine or plasma may be determined by the direct injection of these fluids into calcium hydride with no
prior distillation nor purification. In such cases, however, care would have to be taken to discriminate between free D20 in solution and t h a t bound t o substrate. RESULTS
Xfaiiy n ater solutions were prepared and tested in the apparatus. Table I11 summarizes the results of this work. The size of the peak area for a particular solution tends t o vary because of fluctuations in carrier gas, flow rate, dtltector current, ainiiient conditions, etc., and it is necessary t o employ a normalization factor t o compensate for this. One particular solution (4.74 mole yo added D20) was chosen as a standard and all the peaks obtained for it 11 ere normalized t o the average peak arca obtained c n a n arbitrarily chosen day.
EXAMPLE.On April 27, 1962 (the standard day), the average area of the HD peak for 4.74 mole % added D20 n a s 2407. On April 2 4 1962, the same solution gave a peak area of 2521 Therefore, all peaks obtained for samples analyzed on April 24 nere multiplied by 0.9548 (2407/2521) to give the normalized peak areas. The standard 4.74 mole yo DzO d u t i o n n o > analj zed frequently throughout each su biequent working day to give accurate normalized areas for the other t c - t solutions. Attempts t o correlate fluctuations with atmospheric pressure clianges and thereby correct for them were unsuccessful. The linear relation betaeen average normalized peak arca (-2%)and mole per cent added D20 ( P Ii5 demonstrated by the constancy of ihe C value representing thc quotient of these two value, shonn in Tabli, 111. For solutions n i t h P between 1.244 and 9.48YG (the mo3t common E nalytical range) the average value of C iq 502 area unit> per mole yo added DzCl with an average relative standard de ciation of only i ~ O . 1 7of~ this value. From 11.37 to 99.5 mole % added DzO the same akerage value of C ik found, but the a\ erage relative standard deviation is reduced to zt0.0627, As we go to DrO levels below lYo the halues of C gradually increase a h tlic influence of background DzO from t h e distilled water is felt. This niay normally be expected to run in the neighborhood of 0.016 mole y’, in thi, part of the country (7). We may still expect a linear relationship to hold between th: total deuterium content of the water and the normalized peak area in the background region. If we again let P represent the added mole per cent D 2 0 and B represent the apparent background contribution-of DPO, I%e may expect that C, = A,,/ (P B ) , R here Ct is the C value based
+
on total DkO. From Table 111 it is seen t h a t above 1 mole yo added D20 the difference between C t and C becomes negligiblc. Table I11 shows clearly t h e linearity between LTnand ( P B ) which holds down to the background concentration of DzO, although a t this lower limit the error in C, becomes large. Over t h e range from 0.00 t o 0.995 mole yo added DzO t h e relative standard deviation of C t from an average value of 545 is *12.0yo and clearly most of this error is attributable to t h e very lowest deuterium titers. If the carrier gas had the same deuterium content as t h e background water, we would expect a linear plot of P us. AT,, to pass through the origin.
On the other hand, if the carrier gas contained no deuterium at all, i t should intersect t h e P coordinatefat a point below t h e origin corresponding exactly to the background - level of our distilled water, since A , = C, ( P B ) may be
+
Table 11.
+
A = 2 Ct
written as P - B. Actually, it was found t h a t the least squares intercept is -0.0082 mole % added DzO or about half of the background concentration t o be expected for this locality. This seems reasonable in view of the fact t h a t we are using electrolytic hydrogen as carrier and the hydrogen from which the calcium hydride mas prepared was also produced by electrolysis. The separation factor in electrolysis
Effect of Adding Contaminants to Aqueous Solution Containing 4.97 Mole Added D%O Degrees of Rel. std. d na Cb freedom Std. dev. dev., y6 Contaminant 15 . 0 11.17 506 10 2515 0 1.12.7 1 2 51 507 10 jC A NaCl 2520 A 8 7 1 1 74 502 10 2495 5% SaaSOc 570 THF 2515 506 10 1 8 8 11.73 5% piperidine 2535 510 10 110.5 12.06
70
-4verage normalized area. cent added UZO. See Results.
* Peak area per mole per Table 111.
Experimental Results for Synthetic Samples of DzO
Degrees Pa
00.5 00.46 82.92
66.33 58,53 49.75 39 .80 28.43 19.90 11.37 9.48
9.04 8.12 7.11 5.68 4.97 4.74 3.55 2.58
2.49 1.244 0.995 0.746 0.497 0.249 0.124 0.099
0.075 0.062 0,050
0.044 0.030 0,025 0.011 0.0
d nb 50, ,596
45,721 41,808 33,267 29,229 24,793 19,943 14,272 10,010 5,679
4,750 4,543 4,033 3,610 2,882 2,513 2,407 1,774 1,298 1,226 622 493 383 267 135
70 5 57 3
47 4 36 7 32 0 27 0 21 0 19 0
of
Cc 508 505 504 502
499 498 501 503 503 499 501
freedom 5 5 5
? 0
5 5 5 19 19
49 7
10 19
507
18
502
508
17
306
10
500
80 20 17
508 503 492 500 405
513 537 542 569
578
ti31 5Y2 640
613
8 9
700 760 813
6 0
a3
10
10
10 10 7 20
11
11 10 10 11 12 10
10
10 11
Std. dev. 12.5 zt6.5 1 5 .0 15.0 13.4 19.5 13.3 13.2 1.3.4 14.6 119.5 115.4 113.3 zt14.3 16.8 110.1 14.8 18.1 18.8
Pa
+ Bd
c
... ...
... ... ... ... ...
...
... I
.
.
...
... ...
...
...
...
... ...
111.11
zk12.0 115.3 112.7 1 5 7 .0 130.6 112.3 113.6 127.4 127.2 123.3 135.4 144.8 546.6 1110.0 ...
te
1 :003 0 . 754 0.505 0.25
0.132 0.107 0 ,083 0 ,070
I1 ,058 0 . 052
0.038 0.033 0.019 0.008
... ... ... ... ... ... ... ...
... ...
... 492 508 529
526 534 535 !71 ,124
55%
519 553 576 468 750
Mole per cent added D20. * Bverage normalized peak area. AJP. Mole per cent apparent background DzO. e & / ( P $- B ) .
VOL. 35, NO. 10, SEPTEMBER 1963
1423
is variously estimated from about 3 to 10 in favor of hydrogen (7). If i t were 4, a completely fresh sample of water containing 0.016 mole yo DzO as background would be expected to give off hydrogen containing only one fourth this amount of deuterium and would thus show a n intercept of -0.012 mole % DzO on our scale. The observed intercept suggests t h a t t h e combined effects from the deuterium in t h e carrier gas and calcium hydride correspond to water that has been electrolyzed for some time and so is a good deal richer than background in deuterium. Although such a reasonable result in the extreme background region is a n encouraging demonstration of t h e flesibility and precision of our method, i t cannot be used for absolute determinations in this region, because of uncertainty about the deuterium levels in the carrier gas and calcium hydride. It also appears to be slightly too insensitive at present for quantitative comparisons a t natural abundance levels. T o test this possibility, samples of water from Potomac River steam
condensate (0.0149% deuterium) and Yellowstone Park snow water (0.0128% deuterium) were compared. We were unable to detect the difference between them by direct measurement, although if we had had a large enough sample of each to perform a dilution experiment comparable t o t h a t done with our distilled water, we might have been able to detect t h e 15% difference between them. As thermal conductivity cells of higher sensitivity are developed, the method may be adapted to useful measurements in the background region. ACKNOWLEDGMENT
We express our warm appreciation for t h e many contributions of Ronald Rabinowitz who performed a large number of the determinations and calculations reported here, and to Lloyd Guild of the Burrell Corp. for much assistance. The standard samples of natural abundance deuterium water were generously provided by Vernon Dibeler of the National Bureau of Standards.
LITERATURE CITED
(1) Arnett, E. M., Strem, M., Hepfinger, S . , Lipowitz, J., McGuire, D., Science 131, 1680 (1960). (2) Farkas, A., Farkas, L., Proc. Roy. SOC.London A144, 467 (1934); A146, 623 (1934). (3) Fujita, K., Kwan, T., Bunseki Kagaku 12, No. 1 (1963). (4) Furuyama, S., Kwan, T., J . Phys. Chem. 65, 190 (1961). ( 5 ) Hughes, E. D., Ingold, C. IC., Wilson, C. L., J . Chem. SOC.1934,493. (6) Hunt, P. P., Smith, H. A., J . Phys. Chem. 65, 87 (1961). (7) Kirshenbaum, I., "Physical Properties and Analysis of Heavy Waters," McGraw-Hill, New York, 1951. (8) Moore, W. R., Ward, H. R., J . Phys. Chem. 64,832 (1960). (9) Ohkoshi, S., Tenma, S., Fujita, Y., Kwan. T., Bull. Chem. SOC. J u m n 31, 722 (1958). (10) Riedel. 0.. Uhlmann. E.. 2. Anal.
Ibid., 64, 832 (1 RECEIVED for review January 15, 1963. Accepted June 12, 1963. Work supported by a grant (SSF-G-13513) from the National Science Foundation, for which we are most appreciative.
8 -Merca ptoq uino Iine as a n An a Iytica I Reagent Dissociation and Metal Chelate Formation Constants ALFIO CORSINI, QUINTUS FERNANDO, and HENRY FREISER Department of Chemistry, University of Arizona, Tucson, Ariz.
b The acid dissociation constants of 8-mercaptoquinoline and the first stepwise formation constants of some of its metal chelates have been determined spectrophotometrically or potentiometrically in 50% v./v. aqueous dioxane. Although both functional groups in 8-mercaptoquinoline are significantly more acidic than those of 8-hydroxyquinoline, the chelates of 8-mercaptoquinoline are of comparable stability to, and form a t lower pH values than, the 8-hydroxyquinoline chelates. A reversal in the usual stability order was observed for the Zn(ll) and Ni(l1) chelates of 8-mercaptoquinoline.
I
N CONTRAGT TO
8-hydroxyquino1inel
a white crystalline solid, 8-mercaptoquinoline is a hygroscopic blue liquid which forms a crystalline red dihydrate. This reagent was introduced first in 1944 by Taylor ( I S ) who stated that it vould riot be useful analytically because of ready oxidation to the t)i\-8-quinolyldisulfide. However, with reasonable precautions the oxidation can be largely 1424
ANALYTICAL CHEMISTRY
eliminated. Not until recently has 8mercaptoquinoline received much attention in analytical applications (12). 8-Mercaptoquinoline would appear to have potentialities as useful and as versatile as 8-hydroxyquinoline. For the fullest exploitation of a n analytical reagent, i t is important to have available equilibrium data for its reactions with metal ions. I n addition, there exists some indication t h a t sulfurcontaining ligands exhibit a reversal in the usual chelate stability order of divalent transition metal ions. This reversal in stability order has interesting implications in the problem of reagent selectivity. Accordingly, a study of the formation constants of metal chelates of 8-mercaptoquinoline was undertaken. EXPERIMENTAL
Preparation of 8-Mercaptoquinoline. The reagent [m.p. 58' to 59' C.; lit: 55" to 59" C. (6, 1 2 ) ; 58.5" C. ( I T ) ] was prepared arid purified by a niodified Edinger method ( I S ) . The reaction sequence was:
SOjH
S,,Cli HCl
C9H6h'S-Sn Salt
The purified material was kept under nitrogen but even under these conditions oxidation to the disulfide occurred slowly (small amounts of the lightcolored oxidation product were noticeaide after a few days). 8-Mercaptoquinoline hydrochloride, which is anhydrous and significantly more