cients ( k ) ,the peaks of which appear very shortly after the inert gas peak. Thus one is able to analyze quickly the low boiling hydrocarbons (CI t o C,) above room temperature with open tube columns. B y using various pulverized solids highly variable stationary phases can be produced. The lifetime of thin layer open tube columns is at least comparable with that of capillary columns, but the charging capacity of thin layer open tube columns is higher according t o the greater contents of the stationary liquid. Perhaps possibilities may present themselves for “preparative capillary gas chromatography’’ here, which might obtain importance in identification of unknown peaks in connection with gas chromatographic analysis. LIST OF SYMBOLS
average film thickness
retention time of inert gas retention time of methane = retention time of n-heptane 21 = average linear gas velocity DQO = gaseous diffusion coefficient at 1 atm. Di = liquid diffusion coefficient F, = volumetric flow rate of the carrier gas measured at outlet pressure and temperature of column L = height equivalent t o HETP = n a theoretical plate L HEETP = - = efiective height equivaN lent t o a theoretical plate K = permeability constant (Equation 2) T2 &hear. = - (Equation 4) 8 L = column length N = number of effective plates (Equation 8) R = peak resolution V = volume = gas volume of column Vt = liquid volume of column VN = net retention volume 7 = carrier gas viscosity to
tR,Ci tR.C,
= =
v,
partition coefficient V k = ratio of liquid phase V, capacity t o gas phase number of theoretical plates pressure column inlet pressure column outlet pressure Pi - Po average column radius time time of analysis of R = 1.5 (Equation 10)
-‘
ACKNOWLEDGMENT
We thank E. E. Wegner for discussions of details during this work. LITERATURE CITED
(1) Adlard, E. R., “Gas Chromatography 1956,” D. H. Desty, ed., p. 98, Butter-
worths, London, 1957. (2) Ambrose, D., et al., “Gas Chromatog-
raphy 1960,” R. P. W. Scott, ed., p. 423, Butterworths, London, 1960. (3) Desty, D. H., Goldup, A., “Gas Chromatography 1960,” R. P. W. Scott, ed., p. 162, Butterworths, London, 1960. (4) Desty, D. H., Goldup, A., Swanton, W. T., “Gas Chromatogra hy 1961,” D. Weiss, N. Brenner, J. E. Callen, eds., p. 105, Academc Press, New York, 1962. (5) Golay, M;, J. E., “Gas Chromatography 1958, D. H. Desty, ed., p. 36, Butterworths, London, 1958. (6) HalBse, I., “Gas Chromatography 1961,” N. Brenner, J. E. Callen, M. D. Weiss, eds., p. 137, Academic Press, New York, 1962. ( 7 ) HaUse, I., HorvAth, C., Nature (London) 197,71 (1963). (8) Halhse, I., Schreyer, G., 2. Anal. Chem. 181, 367 (1961). (9) HaUsz, I., Wegner, E. E., BrennstoffChemie 42, 261 (1961). (10) HalBsz, I., Wegner, E. E., Nature 189, 570 (1961). (11) Keulemay, A. I. M., “Gas Chromatography, 2nd ed., p. 142, Reinhold, New York, 1960. (12) Purnell. J. H.. Nature (London) 184, ’ 2009 (1959). ’ (13) Purnell, J. H., Quin?, C. P., “Gas Chromatography 1960, R. P. W. Scott, ed., p. 184, Buttervorths, London, 1wn (14) Schreyer, G., Dissertation, Universitat Frankfurt am Main, Germany, 1961. (15) Wegner, E. E., Dissertation, Universitat Frankfurt am blain, Germany, 1961.
d
_1--.
RECEIVEDfor review January 28, 1963. Accepted January 28, 1963. Presented a t the International Symposium on Advances in Gas Chromatography, University of Houston, Houston, Texas, January 21-24, 1963.
Some Applications of Gas Chromatography in Inorganic Chemistry C. S. G. PHILLIPS and P. L. TIMMS lnorganic Chemisfry laboratory, Oxford University, England
b Despite the fantastic growth of gas chromatography over the last 10 years, its application to essentially inorganic problems has been relatively slight. It is the purpose of this paper to suggest some of the ways in which gas chromatography may b e used in an inorganic laboratory, and to demonstrate this by examples taken from recent work in the authors’ laboratory. The discussion is divided into the study of volatile inorganic substances using conventional column packings, and the use of active inorganic column materials.
V
volatile inorganic substances are highly reactive; they are often toxic or can explode in contact with air. It is therefore convenient ERY X ~ N Y
to be able to handle them in an inert atmosphere and in small quantities, as is normal in gas chromatography. They are often prepared by reactions which lead to a very complex mixture of products, so that an efficient technique is required to separate them. Moreover, as to be shown, the gas chromatographic results can often be used to identify a n inorganic substance, even nhen this substance ha5 never been isolated before. We shall emphasize this point by our discussion of t n o rather simple series of compounds, the volatile hydrides of silicon and germanium, and the substituted borazol(1s. The first group are analogues of the paraffin hydrocarbons and the second of the aiomatic hydrocarbons. We shall also use these tn-o series to suggest that the realm of volatile inorganic compounds is larger
and perhaps more fascinating than is commonly supposed. The volatile hydrides of silicon and germanium may be prepared in a number of ways, of R-hich the simplest is the hydrolysis of an alloy of magnesium with silicon and ’or germanium. With the mixed Ng-.Si-Ge alloy the proportion of silicon-rich-e.g., Si ,GeHlo-or germanium-rich hydrides may be varied readily by using an appropriate silicon-germanium ratio in combination with magnesium. The simple silicon or germanium hydrides (4)have so far turned out to be the analogues of the straight and branched chain saturated aliphatic hydrocarbons, with the same number of isomers for a given number of chain atoms-Le., other than hydrogen. I n the case of the mixed hydrides, however, the level of VOL. 35, NO. 4, APRIL 1963
505
Table
I.
Number of Isomers of Saturated Hydrides
No. of isomers (C, Si, or Ge only)
No. of atoms in chain
No. of isomers (mixed Si and .Ge, includmg simple hydrides)
si3ns
2
1 1
3
complexity will rise very markedly as is shown by Table I. Furthermore, having separated a mixture into its constituent compounds, complexity may be readily reintroduced by submitting one, or even better two of these compounds together, to a pyrolysis or an electric discharge (7). The former tends to produce a variety of products with mainly smaller chains, while the latter gives rise to both smaller and larger chains. Figure 1 shows a chromatogram ( 1 ) obtained (on squalane at 20' C.) after passing a discharge through a small quantity of &Ha, itself isolated from a complex silane mixture resulting from the hydrolysis of magnesium silicide. Mixed C-Si-Ge hydrides might possibly be prepared by a modification of Simmons' methylene insertion reaction (14). Many of the individual compounds in our chromatograms have been idrntified by first isolating sufficiently large quantities by preparative gas chromatography, and then making some suitable physical measurement. Thus monosilane, disilane, and trisilane, which had been prepared by previous workers (16, 16) were identified by their vapor pressures. Mass spectrometry and nuclear magnetic resonance have also been of considerable assistance, although the interpretation of the latter is here often very tricky because of the small chemical shifts and the large coupling constants (4). Infrared spectrometry has been particularly helpful in the borazole field. In a number of cases, molecula1 weights have been determined using either the Martin or the Gon--Mac (Standard Oil) pas density balance.
Figure 1 . Silanes: Chromatograms of (a) trisilane and after passing an electric discharge through trisilane
With hydrogen as carrier gas, molecular weights may be measured to 4=3 t o 4% with samples of the order of 10 pg. by comparing the response in a katharometer with that in a density balance (signal a function of weight). This method depends upon the observation that with a katharometer silicongermanium hydrides of similar retention times give an approximately molar response. This is not found, for example, with hydrocarbons. More precise values (&I%) may be obtained using 50-pg. samples by combining a pressure-volume measurement with that
Table II. Molecular Weight Determination
Method Katharometer and density balance Pressure-volume and density balance
Compound SilGeHD iso-SisGeHlo SipGeHi n-SirHla
~~~
506
0
ANALYTICAL CHEMISTRY
Sample size, pg.
Found
20 10 70 50
139.0 172.0 138.2 120.9
Mol. wt. Required 136.7 166.8 136.7 122.2
(b) products
obtained
of a density balance (17'). Some typical results are given in Table 11. For a mixed silicon-germanium hydride, the silicon-germanium ratio is obtained conveniently by passing the hydride over heated AuCl8, which converts the silicon quantitatively to Sic14 and the germanium to GeCL. These latter are then separated at 20' C. on a 3-ft. X 3-mm. gas chromatographic column packed with 13% Silicone 702 on Celite, and estimated with a gas density balance. I n some casese.g., if the gold chloride is not heated sufficiently-incomplete reaction may occur and a small SiHC13 peak also appears, but this is readily allowed for in the calculation. Some results are shown in Table III. Because of the small sample size, great care must be taken to ensure there is no contamination by traces of a previous sample (which, for example, may readily have been retained by tap grease) ; otherwise gross errors can occur. The method could no doubt be improved further to give the
L0910v; for l l o o' C.
3
2
Number of Atoms inc& 1O
2
3
4
5
6
7 Siatutns
Figure 2. Silanes: Log. retention times (specific retention volumes on Silicone 702 at 1 10' C.) plotted against number of silicon atoms.
I
2
111.
Determination of Ratio
Sammze, Ple Compound iso-SiZGezH iso-GerSiIl,o n-Si4GeHlz
pg.
10 5 10
Si/Ge
Si/Ge ratio Found Required 0.97 : 1 0 . 3 2 :1 3.8s: 1
1:l
0.33: 1 4:1
5
6
atoms
provisionally assigned as normal isomers
Table
4
Figure 3. Silanes, Germanes, and Silicogermanes: Log. retention times (relative to n-tetrasilane on Silicone 702 at 19' C.) plotted against number of silicon and/or germanium
a, b, c, e, and h a r e known normal silane iromenj n and v ore therefore
hydrogen content, as all the hydrogen appears to be converted to HCl. With our very short columns the HC1 has, however, been incompletely separated from the chlorine liberated in the decomposition of the AuC13. I t is perhaps worth emphasizing the value of gas density balances in the routine analyses of inorganic volatiles, as they need no calibration once the molecular weight corresponding to a peak has been established. When working with small quantities of highly reactive and novel compounds, this is clearly a very distinct advantage. Retention Data. When the logarithms of the retention times of paraffin hydrocarbons are plotted against the number of carbon atoms which they contain, a good straight line may be drawn through the points
3
for the n-isomers (8). The same is found for the simple silanes and germanes containing up to 5 silicon or germanium atoms. Further, the branched isomers lie in each case below the straight chain isomers in a manner very closely similar to that found for the hydrocarbons. This is shown in Figure 2 which plots retention times (logarithms) for supposed silanes on Silicone 702 a t 110' C. against the number of silicon atoms (S), the peaks lettered a to h having been identified as the appropriate silanes by various physical methods. Extrapolation of the line passing through the n-isomers (a, b, c, e , and h ) allons the provisional identification of peak n as n-hesssilane and peak v as n-heptasilane, and this identification is supported by the folloming evidence: (Ai)Peak n , when pyrolyzed a t 400" C., gives all the lower n-silanes, but no branched isomers. On the other hand, peak 1 gives iso-penta- and iso-tetra- in addition to the n-silanrs. (B) The sizes of peaks n and v relative t o supposed branched chain hesa- and heptasilanes are about those predicted froin the known ratios of normal to branched isomers Ivith tetra- and pentasilanes. (C) These two peaks as well as those corresponding to monosilane, disilane, trisilane, n-tetrasilane, and n-penta-
Straight chain isomers only.
silane are either removed or very much reduced by addition of 5A molecular sieve to the mixture of silicon hydrides, n hereas the other (branched-chain) peaks are not. (D) The n-isomer peaks and peaks n and v overload easily to produce sharp tails, whereas the branched-chain isomers a t similar concentrations remain symmetrical. In Figure 2 all the peaks emerging between n-pentasilane and the provisionally identified n-hesasilane have been plotted as though they mere hesasilanes, and similarly peaks o to u as though they were heptasiianes. We have no independent evidence for this and the most we can say is that the compounds to n-hich they correspond have the normal chemical reactivity associated with silanes. It is probably fortuitous that if n-e move m into the supposed heptasilanes we obtain just the right number of isomers [ 5 ] for Si6HI4and also [Q]for Si7HI6. Figure 3 shows a similar plot to Figure 2 for the mixed silicon-germanium hydrides, except that to reduce the complexity we have here omitted all values for branched-chain isomers. Yow it will be seen that all the n-isomers fall on a rather simple lattice of points, such that to a good degree of approximation the addition of -SiHz-- adds a constant to the log t~ ( K s , = 0.72)-i.e., VOL. 35, NO. 4, APRIL 1963
0
507
1.2
Figure 4. Borazoles: (Full circles and thick lines), log. retention times (on squalane at 100' C.)for various substituted borazoles plotted against number of carbon atoms external to ring Aromatics: (Open circles and thin lines), log. retention times (on squalane at 100' C.) for various substituted benzenes plotted against number of carbon atoms external to ring
I.o
0.8
2
0.6
0
0
s 2 U
the full lines in Figure 3 are parallelwhile the addition of - G e H r adds another constant ( K c , = 1.07)-i.e., the dotted lines in Figure 3 are parallel. With four chain-atoms (Si and/or Ge) the retention times of the iso-compounds all correspond to approximately 0.7 (retention time of n-compound). We may also express this in terms of a constant to be added to a log t R , when an atom is added to the side of a chain. Thus log t R (iso-Si4Hlo) - log t~ (n-Si3Ha) = 0.57 = K'81, and similarly R'Q, = 0.86. These values of R and R' may then be used for the prediction of further retention times, as in Table IV. Pyrolysis and gas chromatographic analysis of the pyrolysis products is also proving a valuable aid to identification (9). Our results so far show that Ge-Ge bonds are more readily broken than Si-Ge bonds, d i l e Si-Si bonds are not broken under our pyrolysis conditions m-hich comprise heating for less than one second a t 270' to 300' C.e.g., Si6H14is not cracked. Many of the secondary reactions can be interpreted in terms of attack by a GeH2 radical. Thus m-e observe the following reactions :
0.4
2 4 3 0 v)
$
x-
0.2
z
0
G 0.0 3i I-
z
0 2
-0.2
w IW
m u
1: -0.4 I4 -1 W
a
W
5 -0.6
-0.8 -1.0
NUMBER OF CARBON ATOMS E X T E R N A L TO THE RING
-
may be prepared and isolated by reaction 3. On pyrolysis it gives n-tetrasilane and no iso-tetrasilane. On the other hand, the peak
L, L,
obcorresponding to tained from the hydrolysis of the MgSi-Ge alloy, presumably also contains
some (see last example in Table IV), for when this is isolated and pyrolyzed both n-tetrasilane and iso-tetrasilane are formed in the ratio of about three parts of the former to one of the latter. Finally we may make use of a change of column liquid. Thus on substituting
u 0
0
0
0.4
-
/-
/ , , I ,
2 ‘ 3
I4 0 0
AQ *5
/
cu ~
;c 0 c
W
-0.2
z Iz r?
g
-0.4
W
c w
a W
w
F: -0.6 U
2 W
a
Y
c3
3
V
I
I
I
I
I
I
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 LOG (RELATIVE RETENTION TIME 1 ON 13.2% SQUALANE AT 100’ C. Figure 5. Borazoles: Comparison of log. retention times on Carbowax and on squalane
1.0
-0.8
tritolyl phosphate for Silicone 702, we find that all the germanium-rich compounds are retarded relative to the silicon-rich, so that peaks nearly coincident on a preparative silicone column are often readily resolved with a column of tritolyl phosphate. Moreover, as the normal and branched chain isomers tend to respond in the same way to a change of column liquid, this also serves as a useful check on isomer identification. A plot similar to Figure 3 may also be made for the tritolyl phosphate results. As a second illustration of the application of gas chromatography to volatile inorganic compounds, we refer briefly to some work which is being carried out in our laboratory by Phillips, Powell, and Semlyen ( I I ) , a preliminary report of which has been submitted to the Chemical Society, with whose permission we reproduce Figures 4 and 5. Figure 4 is a plot of log t R against number of carbon atoms for a variety of alkyl-substituted borazoles (thick lines and full circles in Figure 4)-e.g., Me
Table IV.
Compound Si 0 Ge 0 (H omitted)
Compound from which retention time estimated
O-u-0 0 L ~~
J - 0
Predicted log t R
Experimental log t R
2.14
2.14
( n - ~ i , ~= , 2.oa)
2.12
m
2.76
2.74
0-C-a
2.41
2.43
c-0-c-a
2.88 2.92
o-o-0-0
2.87
2.86 2.86
Prediction of one borazole retention time is again possible on the basis of the smooth lines connecting homologous series-e.g ., Mer
Me
-
Predicted Retention Times
.. . . Me2 iso-Pr . . Me iso-Pra . . . iso-Prr
Moreover there is a remarkable similarity between the borasole retention time sequences and those found for the analogous aromatic compounds (thin lines and open circles in Figure 4). This is emphasized in the figure where the thick aig-zag line for the boraaoles VOL. 35, NO. 4, APRIL 1963
a
509
is the equivalent of the fine zig-zag line for aromatic hydrocarbons. Borazoles may possess one or more free N-H groups, and the effect of these is shown very clearly when one changes from a column liquid such as squalane to one such as Carbowax (treated with SiMe8C1 to remove OH end groups), which is capable of using its oxygen atoms to form H-bonds with the S - H groups of the borazoles. Figure 5 shows the differentiation between borazoles with 0, 1, and 2 N-H groups. ACTIVE
INORGANIC
COLUMN
MATERIALS
Over the last few years, we have been able to demonstrate (2, 5 , 12) how inorganic liquids such as Cobaltous stearate Nickel alkyl salicylaldimines
may be used as highly selective column liquids for gas chromatography. The gas chromatographic retention data also enable one to calculate rather precisely and rapidly equilibrium constants, free energies, heats, and entropies for the reaction between these column liquids and a wide variety of ligand molecules n-hich are passed as vapors through the column. Inorganic cheniists have been interested for many years in the details of such ligand-metal interactions. Gas chromatography provides a very simple way of investigating such effects. Moreover it is capable of measuring much weaker interactions than have hitherto been studied, and is particularly suited to nonaqueous systems (and hence hydrolyzable ligands) over a wide temperature range, while the more traditional complex ion studies have bem largely concentrated on aqueous solutions around 25” C. TT7ith elution gas chromatography one measures of course the equilibrium under conditions in which the metal atom is always in excess relative to the ligand (otherwise peaks with sharp fronts are observed), but by use of displacement or frontal analysis the complete step-wise metalligand interaction may be studied. In the displacement of amines on a metal
510
ANALYTICAL CHEMISTRY
stearate column we have incidentally been able to fo1loJv the separation as a series of colored bands (6), so that in one sense we have observed true gas chromatography. So far as we are aware this is the first recorded case. It is also probable that the kinetics of metal-ligand interactions could also be studied by gas chromatography using the theory developed by Klinkenberg iIO), for it sometimes happens that there is a very considerable broadening of ligand peaks as opposed to nonligand peaks. Thus with the Pt analogue of the Pd complex mentioned above, r e have found that olefins are not only very markedly retarded relative to paraffins, but their peaks become so spread out that the column material is quite unsuitable for olefin separation. A normal sharp paraffin peak will occur in the middle of a n enormously wide peak, which corresponds to an olefin with some six carbon atoms less in its molecule. As a further variation on the active inorganic column, C. G. Scott (13) has recently been experimenting in our laboratory with aluminas which are heavily coated with inorganic solids such as XaC1, NaBr, SaT, and CuC1. The resulting materials behave as though the alumina was acting merely as an inactive but large surface area support for the inorganic solids. At low vapor concentrations these treated aluminas give rise to nice symmetrical peaks, but with a very high selectivity. Thus at 100’ C. the retmtion time of benzene relative to n-heutane can be varied from 1.0 (SaOH-treated alumina), through 2.0 (NaCl), 3.0 (NaBr), to 3.9 (SaI). On lowering the temperature to 50” C. the ratio increases to 7 n-ith KaI. With the CuC1-modified alumina, the retention time ratio of propene to propane a t 100’ C. is of the order of 500 to 1, while butadiene appears to be irreversibly adsorbed at temperatures as high as 300’ C. Less than 1 gram of this material placed a t the inlet to an analytical column has proved to be an efficient subtractor of olefins from a Fischer Tropsch Clo - CI6fraction. Such highly selective columns may be used for special analytical separations or for the identification of compound types ( 2 , 5 , IW), but it is our
belief that their chief value is likely to be found in preparative gas chromatography, particularly as they are n-ell suited to displacement analysis with high vapor concentrations. ACKNOWLEDGMENT
We thank Imperial Chemical Industries, and the Gon-Mac Instrument Co. for the provision of apparatus, and the Germanium Information Center for samples of high purity germanium. One of us (P. L. T.) is indebted to the Department of Scientific and Industrial Research for a maintenance grant. LITERATURE CITED
(1) Andrews, T. D., Part I1 Thesis,
Oxford, 1962. (2) Barber, D. W., Phillips, C. S. G., Tusa. G. F.. Verdin., A.., J . Chem. Soc. 1959,‘ 18. ‘ (3) Borer, K., D. Phil. Thesis, Oxford, 1960. __..
(4) Borer, K., Phillips, C. S. G., Proc. Chem. SOC.1959, 189. (5) Crtrtoni, G. P., Lowrie, R. S., Phillips, C. S. G., Venanzi, L. M., “Gas Chromatography 1960,” Scott, ed., p. 273, Butterworths, London, 1960. (6) Clayfield, G. W., Greene, P. D., Part I1 Theses, Oxford, 1962. (7) Drake, J. E., Jolly, W. L., Proc. Chem. SOC.1961, 379. (8) James, A. T., Martin, A. J . P., J. d p p l . Chem. 6,105 (1956). (9) Keulemans, A. I . RI., Perry, S. G., “Gas Chromatography 1962,” van Swaay, ed., Butterworths, London, in press. (10) Klinkenberg, A., Chem. Eng. Sci. 15, 255 (1961). (11) Phillips, C. S. G., Powell, P., Semlyen, J. A., J . Chem. Soc., in press. (12) Phillips, 0 . S. G., “Gas Chromatography,” Coates, Noebels, Fagerson, e&., p. 51, Academic Press, New York, 1958. (13) Scott, C. G., “Gas Chromatography 1962.” van Swaav, ed., Butterworths. London, in press. (14) Simmons, M. C., Richardson, D. B., Dvoretzky, I., “Gas Chromatography 1960,” R. P. W. Scott, ed., Butterworths, London, 1960. (15) Stock, A., “Hydrides of Boron and Silicon.” Cornel1 U. P., Ithaca, N. Y.. 1933. (16) Stokland, K., Kgl. Norske Videnskab. Selskabs Skrifter 1950, No. 3, p. 1. C.A. 46, 5412c (1952). (17) Timms, P. L., and Phillips, C. S. G., J . Chrom. 5, 131 (1961). RECEIVED for review Sovember 5 , 1962. Accepted February 15, 1963. Presented a t the International Symposium on Advances in Gas Chromatography, Cniversity of Houston, Houston, Texas, January 31-24, 1963. I
.
1
