showed that the amount of tungsten-dithiol complex extracted into the organic phase varied slightly with time up to seventy minutes, but remained constant after this time through the twenty-four hour period studied. After the addition of the hot reducing agents to the sample containing tungsten, the temperature of the solution is -90 “C. Thus, to avoid errors that would arise from adding chloroform to a hot acidic solution, time is allowed for the solution to cool to room temperature and during this time equilibrium is established also. Table I shows the results obtained by the radioisotope dilution method using “ideal” tungsten solutions. The precision (expressed as the standard deviation of a single determination) using this method is very good. This was achieved by using very stable counting equipment and by collecting at least 106 counts. In Table I1 the results of the tungsten analysis in SRM-480 are shown. The quantities X =tt $ / d i provide 95 confidencelimits for the true mg(W) as measured,
where 2 is a pooled estimate of the standard deviation (9). Molybdenum is separated from the tungsten as described in the procedure. Tracer experiments showed that ' 322
L10
-20
l30 - 40 L 50 - 60
70
Mole % Ba
Figure 1. Retention-liquid phase composition plots for inert solutes, n-butylbenzene and ethyl heptanoate, on columns prepared from Ba(OH)2/diglycerol mixtures Diglycerol content each column, 4.00 g; column temp, 60 "C
present in approximately equal molar amounts, served as sample. When overlapping of sample peaks became troublesome, the corresponding amount of each pure component was injected individually. NMR Spectra. NMR spectra were obtained for diglycerol and barium,/diglycerol mixtures containing 50 and 67 mole % barium using a Varian A-60 spectrometer. Solutions were prepared by adding the appropriate amount of standardized 0.5M barium hydroxide in water to an accurately weighed sample of diglycerol. These aqueous solutions were 0.2M in diglycerol and had a pH of 12. The uncomplexed diglycerol solution was adjusted to pH 12 with potassium hydroxide.
RESULTS Figure 1 is a plot showing the variation in net retention volume per equivalent of vicinal hydroxide for two inert solutes, n-butylbenzene and ethyl heptanoate, against the mole per cent of barium in a series of chromatographic columns prepared with constant amount of diglycerol and varying amounts of barium hydroxide. For both solutes distinct breaks in the curve were observed at compositions of barium and ca. 73 mole %. of 33 mole %, 50 mole %, 71 mole Figure 2 shows a similar plot of net retention volume per equivalent of vicinal hydroxide for the solutes, n-amyl alcohol and chloroform, two components capable of entering into hydrogen bond formation with the liquid phase mixture. Although the data for the solutions were scattered owing to considerable asymmetry of the alcohol peak as liquid phase barium composition was increased and the short retention of
z,
00 I
80 90
IO
20
I
30 40 50 60 70 80 90 Mole % Bo
Figure 2. Retention-liquid phase composition plots for hydrogen bonding solutes, n-amyl alcohol and chloroform, on columns prepared from Ba(OH)2/diglycerol mixtures Diglycerol content each column, 4.00 g; column temp, 60 "C
chloroform, a general decrease in retention with increase in mole per cent barium is apparent, at least through compositions of 50 mole %. DISCUSSION
At least two factors will affect the solubility of an inert solute in a mixture as one molecular species in the liquid phase is converted to another: (1) changes in relative amounts of the various species present will most certainly affect solubility; (2) a salting out effect should produce a sharp decrease in solubility at the point at which chelate formation is complete and an excess of barium hydroxide exists in solution, Maxima or minima in solute solubility occur at a point in the curve corresponding to the stoichiometric composition of complexes formed. The results shown in Figure 1 give evidence of complexes containing barium/diglycerol ratios of 1 :2, 1 :1, and 2 : l from the discontinuities observed at 33, 50, and ca., 71 mole %. The decrease in retention of the solute between 71 and 73 mole % barium has been attributed to the salting out effect produced by excess uncomplexed barium hydroxide with its accompanying decrease in solubility of the solute in the liquid phase. Above ca. 73 mole % barium a residue of undissolved barium hydroxide is observed in the liquid phase, and little change in retention of the solute occurs. The general decrease in retention observed in Figure 2 provides evidence that a decrease in hydrogen bonding occurs between the solute and the liquid phase mixture at compositions of barium at least as high as 50 mole per cent. Both VOL. 41, NO. 11, SEPTEMBER 1969
1457
amyl alcohol and chloroform have active hydrogens capable of entering into hydrogen bond formation. Considering these results the following chelated species may be suggested. CHZO,~ HOCH2 \
~
CHO
I CHz I
I
’
‘Ea
0
I
CHz
I
CHOH CHzOH
I
‘OLH
I CHz I 0 I CHz I CHOH
I
’
CHO
I
‘OCH
I
0
I CHz I I
I
I
CHz
I
CHO
1
0 ‘
1 0 I
CHz
CHz
OCH >ai “
CHO
CHZOH CHZOL kOCH2 1 :2 Chelate 2 . 2 Chelate
CHz
ANALYTICAL CHEMISTRY
Mixture
1
1
67 mole % Bariurn/Diglycerol Mixture
I
CHz
I
CHO
\
CHzO 2 : 1 Chelate
Two possible structures might be proposed for the complex barium, but the dimerized species indicontaining 50 mole cated above is more likely formed than a 1 :1 monomeric species as determined from the decrease in solubility (retentionj of the two inert solutes at 50 mole % barium. A lower solubility for the polarizable n-butylbenzene and ethyl heptanoate would be predicted in the sterically hindered, dimeric 2 : 2 species than would be expected in the more polar 1 : 2 complex. The accessibility of the solute to the metal and the donor oxygens of the diglycerol moiety is greatly reduced by the closed structure of the 2:2 species. On the other hand, a large decrease in solute solubility would not be predicted for a 1 :1 monomeric species because its open structure would allow the solute access to the free hydroxyl groups and the metal, and relatively large solubility for polarizable solutes would be maintained by Debye or induction cohesion forces. The additional coordination sites on the 2 :1 complex may be filled with the methanol solvent and desolvated during evacuation stages in the preparation of the packing material. Confirming evidence for a dimeric 1 :1 metal chelate species is given in Figure 3 which shows a portion of the NMR spectra of diglycerol, a 50 mole % barium/diglycerol mixture, and a 67 mole % barium/diglycerol mixture. The peaks at about 5 ppm are the hydroxyl groups of the diglycerol as well as water and its spinning side bands. The set of peaks at ca. 3.8 ppm is the result of various carbon-bonded protons. Little difference exists in the 3.8 ppm region for all three spectra because chelation has almost no effect on these protons. The peak at 6 = 5.0 for the aqueous diglycerol solution, however, is considerably broader than that observed for either the 50 mole % or the 67 mole mixtures owing presumably to exchange broadening of free hydroxyl protons. Because the NMR spectrum for the 50 mole % mixture is essentially
1458
I
1
I 6.0
I
I
4.0
5.0
S
I
3.0
(ppm)
Figure 3. NMR spectra of ( A ) diglycerol, ( B ) 50 mole barium/diglycerol mixture; ( C ) 67 mole barium/diglycerol mixture Water solvent; each solution O.2M in diglycerol; pH, 12
identical to that of the 67 mole % mixture, there is no evidence of free primary or secondary hydroxyl groups as would be present in a monomeric 1 :1 chelate species. Thus, NMR studies would appear to support the gas chromatographic evidence for the dimeric species. This new gas chromatographic technique provides yet another method for the determination of the structures of nonvolatile metal chelates for systems difficult to study by other means. ACKNOWLEDGMENT
We thank Michael L. Crosser for determining impurity levels in the diglycerol liquid phase. Received for review April 7, 1969. Accepted July 7, 1969. Presented in part at the Sixth International Symposium on Gas Chromatography and Associated Techniques in Rome, Italy, September 1966, and at the 156th National ACS Meeting in Atlantic City, N. J., September 1968. This work was supported by the National Science Foundation under Grant GP-
5151X.
