i t is apparent that there is a significant difference in lability. For EDTA, most + 2 metal ions form complexes in which the nitrogen-metal bond has a short lifetime while for CyDTA and BDTA, in almost all cases, the lifetime is relatively long enough so that i t is outside of the XAlR “window.” For example, in the case of Zn, the E D T A complex has a short niitrogen-metal bond lifetime at room temperature, but in the BDTA comples the lifetime is long even a t 90’ C. Furthermore, even compleses of metal ions such as the alkaline earths in which the bonding is almost entirely electrostatic have long nitrogenmetal bond lifetimes with CyDTA and in some cases BDTA. This indicates that the longer lifetime of the metalnitrogen bond is causisd by steric effects rather than changes in the bond. In order to break the nitrogen-metal bond, it is probably necessary for some other ligand such as a water molecule t o replace the nitrogen. Thus, the water molecule must enter between the metal ion and the nitrogen a tom. This would be easier in the case of EDT-4 because the ethylenic backbone is relatively flesible; in the case of BDTA and especially CyDT-1, the additional groups tend to make the chelate structure more rigid, so that it IS more difficult to break the nitrogen and its two acetate arms away from the metal ion. I n the NXR method, a nitrogen-metal bond
with a short lifetime means that the two acetate groups on one nitrogen are averaged out. This requires that the following process take place rapidly: First, the nitrogen-metal bond is broken; then the nitrogen undergoes umbrella inversion, and the --N(Ac)z group rotates about the (Ac)zN-C bond so that the lone pair of electrons again face the metal ion, reforming the bond. Therefore, steric effects can slow down averaging of the acetate groups by slowing down the bond-breaking step and by slowing down the rotation of free nitrogen, so that it might be able to invert back to the original position and not eschange the acetate positions. However, the fact that the acetates are rapidly averaged at room temperature for the free ligands for the steric effects does not interfere with the rotation by an excessive amount. Therefore, a significant part of the steric effect is caused by the inhibition of nietalnitrogen bond fracture which is a result of the nitrogen being sterically held in a bonding position. Although every metal-nitrogen bond rupture will not lead to interchange, as long as the rate of interchange is of the same order of magnitude as the rate of metal-nitrogen bond fracture, the NMR spectrum will still yield a good qualitative indication of the relative labilities of the metalnitrogen bonds.
LITERATURE CITED
(1) Anet, F. A. L., J . Am. Chem. SOC. 84, 747 (1962). (2) Bailar, J. C., J . Znorg. Nucl. Chem. 8 . 165 (1958). (3) .Bothner-By, A. A., Naar-Colin, C., J . Am. Chem. SOC. 84. 743 (1962). (4) Day, R. J., Reilliy, C. N., ANAL. CHEM.36, 1073 (1964). (5) Dwyer, F. P., Garvan, F. L., J . Am. Chem. SOC.81, 2955 (1959). (6) Zbid., 83, 2610 (1961). (7) Gutowsky, H. S., J . Chem. Phys. 37, 2196 (1962). (8) Kula, R. J., Sawyer, D. T., University \ - - - - I
of California, Riverside, Calif., personal communication, 1965. (9) Kula, R. J., Sawyer, D. T., Chan, S. I.. Finlev. C. hI.. J . Am. Chem. SOC. 85. 2930”(’1963). ’ (10) ‘Lind, hI. D‘., Lee, B., Hoard, J. L., Zbid., 87, 1611 (1965). (11) AIori, &I., Shibata, X I Kyuno, E., Nakajima, H., Bull. Chem. SOC.Japan (12) 29, Saunders, 887 (1956). AI., Yamada, %I.,J . Am.
Chem. SOC.85, 1882 (1963). (13) Schwarzenbach, G., Helv. Chim. Acta 32, 839 (1949). (14) Sudmeier, J. L., Reilley, C. K., ANAL.CHEM.36, 1707 (1964). (15) Wright, D. L., Holloway, J. H., Reilley, C. N., Zbid., 37, 884 (1965).
RECEIVEDfor review February 1, 1965. Accepted May 7, 1965. Division of Analytical Chemistry, 149th Meeting, ACS, Detroit, llich., April 7, 1965. Research supported in part by Sational Institutes of Health, Grant RG-08349. One of the authors ( R . J. D.) gratefully acknowledges the help of a National Science Foundation Cooperative Graduate Fellowship.
Analysis of Deuterium Oxide-Water Mixtures and Measurement of the Deuterium Isotope Effect on the Acid-Base Equilibrium of N,N-Dimethylbenzylamine Using Nuclear Magnetic Resonance DONALD E. LEYDEN1 and CHARLES N. REILLEY Departmenf of Chemisfry, University o f Norfh Carolina, Chapel Hill, N. C.
b Advantage i s taken of the difference in the NMR spectrum of H+A and of D+A in which A is an amine to determine the D l O content in DZO-HZO mixtures. The ratio of the area of the spectrum of D+A to that of H+A i s a quantitative measure of the D 2 0 . Samples ranging from 2 to 80Y0 DzO were analyzed usinig a calibration curve, and relative errors of 1 to were obtained. The extension of the technique to the investigation of the isotope effect on the acid-base equilibria of amines and related compounds is also discussed, and the isotope effect on the dissociation of H+A and D+A in H20 and DzO, respectively, has been determined for N,N-dimethylbenzylamine.
2Y0
T
of the deuterium content in DzO-HzO mixtures has been the subject of many investigations. Kirshenbaum (8) has summarized and discussed many of the known methods. More recently, techniques such as infrared spectrophotometry ( I @ , vapor pressure measurement ( I I ) , radiometric methods ( I ) , and gas phase detection of Hzand H D generated from D20-H20 mixtures (2) have been employed. Goldblatt and Jones (5) have recently evaluated three methods of deuterium determination including one employing nuclear magnetic resonance previously described by Goldman (6). This latter method is based on standard additions of small amounts of HzO to a D20-Hz0 mixture which conHE DETERMINATION
tain a high percentage of DzO. Extrapolation of a plot of the integrated area of the H D O signal to zero area gives the original protium content. Standard addition methods such as this are ideally suited for samples containing low protium content. For samples containing DzO in the range of approximately 10 to 90%, the addition of a suitable internal standard and subsequent comparison of the area of the proton signal in the water and in the standard should represent one reasonable and straightforward approach. Factors which need to be considered for selection of a,n internal standard and for proper measPresent address, Department of Chemistry, University of Georgia, Athens, Ga. VOL. 37, NO. 1 1 , OCTOBER 1965
1333
urement of the relative areas have been discussed by Paulsen and Cooke (12). The above X h I R methods yield an estimate of the D2O content by difference and, hence, are not ideally suited for samples containing lorn DzO concentrations. For such samples, a method whereby positive signals indicate the concentration of both the DpO and the HzO would be more desirable. A scheme for achieving this goal constitutes the purpose of this paper. BASIS OF THE METHOD
The method proposed herein is based on the addition to a D20-Hz0 sample of a reagent ( R ) which interacts to yield products D R and HR, each exhibiting a different characteristic S h I R spectrum, I n this way, the ielative amounts of the two isotopes can be directly compared. The selection of the chosen reagent was governed by certain general considerations, and the final choice was made without an exhaustive study. The reagent qhould react rapidly and in a reproducible manner with the two isotopes and should have the usual desired properties of easy availability, stability, and solubility. The design of the reagent should be such that HR and DR exhibit simple, nonoverlapping spectra of sufficient amplitude such that excessive amounts of reagent can be avoided. Because the two isotopes would not be expected to yield significant chemicalshift differences in the protons in the reagent molecule (S), advantage must be taken of the difference in the spinspin interaction in the two cases. The proton ( I = zero quadrapole moment, and magnetic moment of 2.79 magnetons), would spin couple to the reagent protons in a manner different from that of deuterium ( I = 1, quadrapole moment of 2.57 X and magnetic moment of 0.86 magneton). Thls difference in spin-spin interaction results in separate characteristic spectra for HR and DR and, hence, permits quantitative measurement of the two forms. The magnitude of spin-spin interaction will be greatest when the reagent proton(s) and the H (or D) nuclei are separated b y a minimal number of bonds. The maximum interaction would be obtained in D-H and H-H where H is considered to be the reagent-Le., from a suitable hydride. Although H-H is strongly spin coupled, the two protons have identical chemicalshift values, and a single K M R signal is observed. I n D-H the proton resonance is a triplet. and its central peak is superimposed upon the H-H signal. This overlapping of the two spectra plus the inconvenience of handling the gaseous products eliminated the consideration of hydride as the reagent. The next case considered was one in which the isotope nuclei are separated 1334
ANALYTICAL CHEMISTRY
from the reagent’ protons by two bonds, the products being represented symboland D-Xically as H--X-(H). Two cases exist, the first being where H and H have identical chemicalshift values and the second where they do not. I n the first case, the N l I R patterns are similar to those of H-H and D-H except that the spin-spin interaction is smaller and the possibility exists that’ the D-X-(H), will exhibit no spin-spin coupling because of rapid relaxation of the deuterium nucleus by electrical field gradients in the molecule ( I S ) . I n the latter event, separate spectra would not be observed for the two products. Where X is an oxygen atom, chemical exchange of the protons removes the spin-spin interaction. When X is a nitrogen atom, proton exchange can be removed by use of a highly acidic medium, but spin-spin interaction between the nitrogen nucleus and the protons so complicates the spectrum that reagents of this t’ype were not considered feasible. Consideration may be given to reagent in which X is CRR’. Where R = R’ = H , the product is gaseous and the spectra overlap. Where R # R’ # H, the two protons may still be in similar magnetic environments, and the spectra nil1 again overlap. An interesting case arises if either R or R’ contains an asymmetric center (R*). The two protons in HC(R*R)H, although attached to the same carbon atom, may be in different magnetic environments and, therefore, exhibit an AB pattern. At the same time, because of the asymmetry present in DC(R*R)H, the deuterium-proton coupling may not be effective, and, in the absence of long-range spin-spin coupling, a single resonance signal should be observed for H. Horvever, there are further considerat’ions in the appIication of such a reagent, and these are explained by the use of Figure 1. For generality, it is assumed that the reagent is capable of accepting two protons; this would lead to the product shown, and the relative concentrations of these would depend upon the D20-toHzO ratio in the sample. Products of the type HC(R*R)H would exhibit an .-iBpattern whereas products of the type DC(R*R)H could exist in t\yo forms, each of which exhibits a single signal with a separation of AvaB between the two (4). This can be understood by consideration of the structures shown at the bottom of Figure 1 in which R* is represented as C(syz). Kone of these structures are equivalent because of the asymmetry of C(zyz). Therefore, if one proton and one deuteron are present, the chemical-shift value of the observed signa1 will depend upon t’he configuration of the C(RHD) carbon. Because reaction of the sample with a reagent designed to give products such as those
represented in Figure 1 would, undoubtedly, require considerable time and effort along with the lack of advantage over later considerations, the selection of a reagent of this type was eliminated. Consideration may now be given to the final case considered in which the reagent protons are separated from the isotope nuclei by three bonds, the products being represented by H-X-[C(RR’)Hl, and D-X-[C(RR’)H],. Cases involving further separation of the reagent protons and the isotope nuclei were not considered because the larger separation would reduce the spin-spin interaction and sufficient separation of the different signals would not be obtained. I n the case where X is C(RR’), the spectra of both possible products would be identical for the reasons given for D-X-(H), and H-X-(H),. In the case where X is C(R’’R”), the spectrum of H-C (R”R”)-C(RR’)--H would exhibit spin-spin coupling between the two protons whereas the spectrum of D-C(R”R”)-C(RR’)-H would not. However, a reagent which contains such a degree of asymmetry and yet not exhibit interfering spin-spin interaction between the R groups and the reagent proton would be difficult to obtain. Where X is an oxygen atom, the products are alcohols represented as H-0-C(RR’)H and DO-C(RR’)H. Generally, spin-spin interaction between the hydroxyl proton and the C-H proton is observed only in alcohols of high purity a t low temperatures, or in special solvents such as dimethylsulfoxide. Traces of acids or bases accelerate exchange of the hydroxyl proton and eliminate spin-spin interaction. For this reason, we eliminated the use of alcohols as the reagent. Where X is a nitrogen atom, the products may be protonated amines (H+A) or DfA) having proton exchange rates such that chemical equilibrium is established in a reasonable time but sufficiently slow that spin-spin interaction may occur between the N-H and N-alkyl protons. The rate of this exchange depends upon the hydrogen ion concentration in the solution and the pK, of the amine (7, IO). Because this type of compound was considered to be well suited for the purpose of providing direct S l I R signals indicating the relative amount of DzO and HzO, it was chosen as the class of reagents to be used and is discussed below. AMINES AS THE REAGENT
Seglecting other considerations for the present, the selection of trimethylamine as a possible reagent can be made on the basis of its solubility, stability, pK, value, and the sensitivity resulting from the nine equivalent methyl protons. An example of the effect of
R* \ /
R
H
H
R R* \ /
Figure 1 . Effect of deuterium substitution on a carbon atom adjacent to an asymmetric site
changes in the N-H proton exchange rate upon the magnetic resonance spectrum of the methyl proton in trimethylamine (TMA) in the absence and presence of deuterium ions is shown in Figure 2. The NMR spectrum of a 0.5M solution of ThIA a t p H 4.50 appears as shown in Figure 2a. The nine equivalent protons of the methyl groups exhibit a strong singlet,, Figure 2b shows a similar solution of T M A at p H 2.42. This spectrum shows the broadening of the singlet as a result of the decrease in the proton exchange rate. If the p H of the solution i:i further decreased, the methyl protons are split into a sharp doublet by the N-H proton as is shown by the spectrum in Figure 2c. The K-H proton resonance is not readily observed because of the intensive splitting by the methyl protons as well as spin-spin interaction with the "4 nucleus (10). If, along with protiom, deuterium ions are present in the solution, a similar result is expected to occur for the protonated species. However, as a result of the deuterium quadrapole moment, spin-spin interaction between the deuterium nucleus and the reagent protons is not expected to occur. A spectrum of TMA in an acidic solution containing
J"
J " P * 0.1
5.4
Figure 2. NMR spectra of aqueous trimethylamine a t various p H values and in 10% D 2 0
deuterium ions could be expected to exhibit a doublet as shown in Figure 2c as well as a single peak located between the two doublet signals. The ratio of the area of the single peak of D+T;LIA to that of the two doublet signals of H+TMA should be proportional to the ratio of deuterium ions to protium ions present and, therefore, provide a quantitative measurement of the deuterium content of a D20-H20 mixture. However, a spectrum of the methyl protons of T M A in 1.OF hydrochloric acid containing 10% DzO in Figure 2d shows that the center portion of the spectrum is a triplet. That the triplet nature of the signal is caused by spin-spin interaction of the deuterium nucleus (instead of the W4 nucleus) with the methyl protons is evidenced by the fact that no triplet character is
observed when deuterium ions are not present and that the amplitude of the triplet signal is proportional to the deuterium concentration in the solution. The fact that a triplet resonance signal is observed indicates that the deuterium nucleus is located in a symmetrical electrical environment although this was not expected initially. This symmetrical electrical environment of the deuterium nucleus is probably in part a result of the hydration of the cation acid. I n order to obtain quantitative measurement using the methyl proton spectrum, a singlet resonance signal for D+A is more desirable. Because of the importance of electrical symmetry in the quadrapole relaxation mechanism, substitution of a different group for one of the methyl groups would be expected VOL. 37, NO. 1 1 , OCTOBER 1965
1335
/
2 0 '10 D2 0
Figure 3.
NMR spectra of aqueous N,N-dimethylbenzylamine containing 20, 50, and 70% D 2 0
t o collapse the triplet because of introduction of asymmetric electric field gradients in the molecule. However, the new amine must also have a pK, such that the exchange reaction may be suitably decreased a t a reasonable hydrogen ion concentration. Of the several amines considered] NIN-dimethylbenzylamine (1IB.i) was selected. The spectrum of the methyl proton resonance signal of a 0.5F solution of this amine in a 2F hydrochloric acid solution containing various amounts of DzO is shown in Figure 3. The signal exhibited by D+llBA is a singlet. Hence, the electrical asymmetry introduced in the molecule by substitution of a benzyl group for one of the methyl group of trimethylamine has been sufficient to collapse the triplet observed with the latter amine. The chemicalshift value of methyl resonance signals of D+hIBA and H+MBA are identical within experimental error as would be expected (3). A similar effect is observed for the -CHzprotons of the benzyl group. Figure 3 also shows that the area of the portion of the spectrum associated with D+MBA increases with increasing DzO a t the expense of that associated with H+XBA. EXPERIMENTAL
All spectra were obtained using a Varian A-60 nuclear magnetic resonance spectrometer. N,N-dimethylbenzylamine was obtained from Distillation 1336
ANALYTICAL CHEMISTRY
Products Industries as Eastman White Label and was used without further purification. Deuterium oxide (99.8001,) was obtained from Columbia Organic Chemicals Co. A stock solution of the amine was obtained by preparing a solution 1.2F in the reagent in 12F hydrochloric acid. The stock solution could be kept for several days, particularly if stored in the dark. Samples of Dz0-H20mixtures were prepared by mixing weighed amounts of each constituent. Calibration curves were obtained by placing 500 rl. of the D20HzO mixtures in a NMR sample tube and adding 100 ~ 1 of. the reagent solution. The same procedure was followed for the test samples. Because paramagnetic ions broaden the methyl proton spectrum, care must be taken to prevent their presence. For example, no metal-tipped syringes may be used as paramagnetic species may be picked u p by the solvent. Although no isotopic exchange of C-H protons was expected or observed] spectra were taken immediately after sample preparation. The area of the signals arising from the methyl protons of the amine was measured by both electronic integration and peak-height measurements. The two methods gave similar results for the analysis of DzO-H20 mixtures. Because of the ease and theoretical foundation (17) of electronic integration, this method of area measurements was used for the analyses performed. A calibration curve was obtained by plotting the fraction of the total area associated with D+iMBA us. per cent DzO. The total area of the spectrum at 100% DzO
is not a result of only D+JiBA because the reagent was added as a light-water solution; therefore, there is always some HzO present in the final mixture. The application of the technique described in this paper to the study of isotope effects required more accurate measurement of the area of the two types of methyl resonance signals than was required for the preparation of a calibration curve for analysis. Because the doublet signal for H+MBA did not return to the base line between the two peaks in pure HzO and because of overlap of the two signals in D20HzO mixtures, a simple integration technique could not be used. A more accurate measurement of areas was made by comparing the area of the signal obtained in a mixture of DzOHzO with the area of a signal obtained in pure HzO which could be superimposed on the H+MBA portion of the spectrum of the amine in a DzO-HzO mixture. After a good superposition was obtained, the areas were measured with a planimeter. Corrections were made for HzO added in the form of the reagent solution. ANALYSIS OF DzO-H20 MIXTURES
Figure 4 shows a calibration curve obtained as described above. The extensive nonlinearity at both high and low p.er cent DzO values is in part a result of inaccuracies in the measurement of the differences of the very large and very small relative areas involved in these regions, There is also a small degree
J
r
z a
c3
2r---l
r
I -
v)
a I
-
N
nwn o
2
+: f
I A
-I
-
-2
-
uu
0
z
-I-
0
o
-I
0
(3
a a LL
*
4
-2
+a PERCENT Figure 4. mixtures
LOG
DZO
Calibration curve for analysis of DzO-HzO
+ D20 e
Figure 5.
Plot of
log
.w I
2
[D+A]2/[H+A]Z vs.
3
log
[ H I 0 I/[DzO I
of nonlinearity over the intermediate range of D20 percentages, and this will be discussed below. Table I shows the results of some sample analyses. These results indicate that the technique is applicable to a wide range of D 2 0 concentration with accuracy comparable to many of the techniques previoudy described. The technique has the advantage of being simple and rapid to perform. X o pretreatment of the sample is required, and only a fraction of a milliliter is needed. The instrumentation is available t o most laboratories. The method discussed in this paper requires the D20-H20 mixtures to be analyzed to be free of certain types of interfering impurities. Small concentrations of organic materials will not interfere with the ,analysis unless the A X R spectrum of the impurity coincides with that of the reagent. However, the presence of paramagnetic metal ions will cause the spectrum to be broadened. The nonlinearity of the calibration curve shown in Figure 4 is undesirable from an analytical point of view. However, this phenomenon draws at'tention to the need to consider isotope effects upon the various equilibria which exist in an acidic solution of a tertiary amine in a mixture of D 2 0 and HzO. For example, the equilibrium 2 HfA
0
-I
2 D+A
+ HzO
(1)
in which 11 represents an amine can be proposed as one 'equilibrium which exists. The equilibrium constant for this expression may bse written as
or in a logarithmic form
Table I.
The values of [ D + A ] / [ H + X ]for various ratios of HzO to DzO are obtained directly from the nuclear magnetic resonance spectra. The values of [H20]/[DzO] can be obtained from the analytical concentration and the use of the important equilibrium Hz0 DZO e 2 HDO (4)
Taken
2 5 2 , 2 24,2 25,l 95, 2 29 4 41 4 45, 4 42,4 59, 4 50 8 24 8 12, 8 12,8 77, 8 16 12 50 12 50. 11 83. 12 7 5 21.69 21.37;22.00;21.30, 21.50,21.98 72,lO 70.25,70.25,71.31, 70.34,7 2 , O O 81.59 85.25,82.00,84.05, 82.10
The equilibrium constant for this equilibrium may be written as
The value of K' has been found by Urey (16) to be very nearly the statistical value of 4.0. Knowledge of the ratios [ D + A ] 2 / [ H f A ] 2 and [HzO]/ [DzO]allows calculation of K . According to Equation 3, a plot of log [D+X]Z/ [H+AI2 us. log [H20]/[Dz0] will be a linear plot with a slope of -1 as illustrated in Figure 5 . The value of log K is equal t o the value of log [ H Z O ] / [ D ~ O ] when log [ D + d ] 2 / [ H + A ] 2equals zero. A value of 1.04 for K is obtained from Figure 5 at the point indicated by the arrow. This value for K is a t first deceiving because it seems to indicate that there is little or no isotope effect on the acid-base equilibrium of N,N-dimethylbenzylamine. However, there are still further isotopic distribution equilibria to consider. For example, H+X HzO e H3+0 A K H = 1.2 X l o u 9 (6) represents the dissociation of the cation acid in HzO whereas D+.k DzO D3+0 A K D = 3.9 X lo-'' (7) represents a similar process in D 2 0 .
+
+
+
Per cent D 2 0 Found
AV.
2 25
2 21
+
+
NMR Determination of D 2 0 in DzO-HzO Mixtures
4 49 8 29 12 86 .. 21.63 ~~
70.83 83.35
The values of KH and K Dwere determined directly using a potentiometric technique similar to the one described by Salomaa, Schaleger, and Long (14). The expression
represents the isotope effect for converting hydrogen in the solvent to hydrogen in the solvated proton aggregate. The equilibrium constant for the isotopic distribution equilibrium is given by
where L is generally reported to be about 11.0 (9). The value of L has also been measured directly by use of an NhlR technique (9). An interesting result of the measurement of the equilibrium constant K by the 9 h I R technique described in this paper is that combination of Equations 2,6,7,and 9 yields
VOL. 37, NO. 1 1 , OCTOBER 1965
1337
The significar e of Equation 10 is that the direct rnwurement of K allows a comparison of these several equilibrium constants in terms of a simple relationship. As mentioned above, K H and K o are obtained independently by a potentiometric technique, and the resulting value of K H / K o is 3.09. The value of fiobtained is 3.38. Further investigation of equilibrium and kinetic deuterium isotope effects using NMR should provide interesting information concerning the nature of these effects. The investigation of isotope effects on the acid-base chemistry of amino acids would be of particular interest in view of the physiological effects of the presence of large amounts of DzO in biological systems. Such a n investigation should also include research into the nature of the electrical
symmetry and extension of the method described in this paper to the analysis of other isotopic mixtures. LITERATURE CITED
(10) Loewenstein, A., Meiboom, S., J . Chem. Phys. 27,1067 (1957). (11) Mazurek, M, Perlin, A. S., Can. J . Chem. 42, 710 (1964). (12) Paulsen, P. J., Cooke, W. D., ANAL. CHEM.36.1713 (1964). (13) Pople,‘ J. A., Schneider, W. G.,
( 1 ) Ameil, S., Peisach, AT., ANAL. CHEM. 34, 1305 (1962). ( 2 ) Arnett, E. M., Dugglesby, P. McC., Ibid., 35, 1420 (1063). ( 3 ) Bergquist, hI. S., Erickson, L. E. G., Acta Chem. Scand. 16, 2308 (1962). ( 4 ) Davis, D. R., Roberts, J. D., J . Am. Chem. SOC. 84, 2252 (1962). ( 5 ) Goldblatt, AT., Jones, W. hl., ANAL. CHEM.36.431 i1964). ( 6 ) Goldmin, hl., Arch. Sei. (Geneva) 10, 247 (1957).
(1956). (16) Urey, H. C., J . Chem. SOC. 1947, p. 562. (17) T’arian Associates, Palo Alto, Calif., Tech. Information Bull. 3 , 1 (1960).
(1957). (8) Kirshenbaum, I., “Physical Properties of Heavy Water,” NcGraw-Hill, New York, 1951. ( 9 ) Kresge, A. J., Allred, A. L., J . -4m. Chem. SOC.85, 1541 (1963).
RECEIVEDfor review May 11, 1965. Accepted June 7, 1965. Investigation supported by Public Health Service Research Grant RG-08349 from the National Institutes of Health. Division of Analytical Chemistry, 149th Meeting, ACS, Detroit, April 1965.
( 7 ) Grunwald,
E., Loewenstein, A., Lleiboom, S.J., J . Phys. Chem 27, 630
Bernstein, H. J., “High F l u t i o n Nuclear Magnetic Resonance, p. 214, hlcGraw-Hill, New York, 1959. (14) Salomaa, P. Schaleger, L. L., Long, F. A., J . Am. Chem. SOC.86, 1 (1964). (15) Trenner, N. R., Arison, B. B., Walker, R. W., ANAL. CHEM.28,.830
END OF SYMPOSIUM
Precision Activation Analysis with 14-Million Electron Volt Neutrons WILLIAM E. MOTT and JOHN
M. ORANGE
Gulf Research & Development Company, Pittsburgh, Pa.
As part of a program to develop, for routine industrial use, a series of analytical methods based on activation with 14-m.e.v. neutrons, an extensive study was made of the factors affecting precision. Both the comparator and the indirect flux monitoring methods of analysis were studied. In each case the precision was limited by conditions inside, rather than outside, the accelerator-viz,, by variations in deuteron flux over the beam area and by inhomogeneities in the density of tritium in the target. Before the work described in this paper was initiated, precision was generally limited to about a 1 to 2% standard deviation even when counting statistics predicted much lower values. At the present time observed fractional standard deviations (root-meansquare errors) consistently agree with the expected values (from counting statistics), which in some analyses are frequently as low as 0.3%.
D
in the literature indicate that the precision of analytical results obtained by the fast neutron activation method is usually limited b y factors other than counting statistics. Despite this there have been ATA GIVEN
1338
ANALYTICAL CHEMISTRY
7 5230
few definitive studies of this rather fundamental and very important problem, probably the most thorough to date being by Xnders and Briden (1). They examined many of the gross factors affecting precision in oxygen analyses including the monitoring of the neutron flux in the sample being analyzed. Gilmore and Hull (2) and Iddings ( 3 ) also studied flux monitoring techniques and their role in limiting precision in oxygen analyses. About the same time, hIott and Rhodes (4) found that for expected relative standard deviations below 1%, differences between observed standard deviations (rootmean-square error) and those expected from counting statistics could usually be related to variations of the flux in the unknown sample not reflected in the count of the flux monitor, whether a BF3 counter, a proton recoil counter, an associated particle counter, or a simultaneously irradiated comparator. These flux variations were attributed to changes in deuteron flux over the beam area and by inhomogeneities in the density of tritium in the target. They concluded that whereas relative standard deviations of 1 to 2% should be relatively easy to achieve, any signifi-
cant advancement in precision would require improved sample irradiationflux monitoring systems in which simultaneous irradiation of unknown sample and standard would be a necessary, but not a sufficient, condition for success. When considering the precision problem, two facts become clear. First, improved precision is a necessity if fast neutron activation analysis is to compete favorably on a broad scale with other methods of elemental analysis. Second, poor precision is not inherent in the fast activation approach, only in some of the techniques employed. It becomes of interest, therefore, to study beam and target effects and ways to compensate for them. This paper reports on such a study and describes irradiation systems now in use at Gulf Research that with homogeneous samples give observed standard deviations down to 0.3y0 when the appropriate number of counts are collected from sample and monitor. FAST NEUTRON ACTIVATION FACILITY
Neutron Generator. A 130-kv. deuteron accelerator, which was designed and constructed a t t h e Gulf Research Center in 1957, is t h e core of the fast neutron activation facility,