2683
TRIMETHYLAMINE ALANE
vation is a limiting value, we can only suggest that AS*,, has a magnitude which would be expected for an activated complex in a gaseous transition state. However, we wish to emphasize the speculative nature of the foregoing comments which are based upon quantitative treatment of the surface vaporization data. The result obtained in this work for the relative ionization cross sections for gaseous Zn and P,, UZ,,/UP, = 0.49, may be compared to the value 0.29 which is obtained from the ionization cross sections for Zn and P of Otvos and Stevensonz6and the assumption of crosssection additivity which is common in high-temperature mass spectrometry. The discrepancy may be attributed in part to an error in the electron multiplier detection efficiency correction or to the use of incorrect cross sections for atomic zinc and phosphorus, but a particularly likely source of error resides in the use of the rule of additivity to obtain molecular cross
sections by summing contributions from the constituent atoms. I n a critical review, Pottie2' has pointed out that summation of constituent atomic cross sections generally leads to over estimation of the magnitude of a molecular cross section. At the present time there is no reliable empirical correlation or theoretical approach to calculation of cross sections for ionization of molecules by electron impact.
Acknowledgment. R. C. Schoonmaker wishes to thank the National Science Foundation for a fellowship and Professor C. A. Coulson, Mathematical Institute, Oxford University, for providing facilities and generous hospitality during a sabbatical leave when this article was prepared. (26) V. W. Otvos and (1956).
D. P. Stevenson, J . Am. C h m . Soc., 78, 640
(27) R. F. Pottie, J . Chem. Phya., 44, 916 (1966).
Trimethylamine Alane, Its Molecular Weight, Dipole Moment, and Structure'
by D. G. Hendricker and C. W. Heits&* Contribution No. 1786 from the Ames Laboratory of the Atomic Energy Comrnhabn, Iowa State University, Ames, Iowa (Received February 14, 1967)
Trimethylamine alane, AIH3.N(CH3)3, was studied with respect to its molecular weight, dipole moment, and nmr spectrum. All evidence indicates a monomer under the conditions used here. Old evidence for a dimer which is said to exist in benzene solution is reexamined and rejected.
I n one of the earliest reports on trimethylamine alane, AIH3*N(CH3)3, Wiberg, Graf, and Uson describe the compound as being partly associated in freezing benzene but monomeric in boiling ether.3 The association in benzene has also been observed by other worker^.^-^ This apparent association is attributed by some to hydrogen bridges between monomer unit^^^^^^ while others ascribe it to dipole-dipole interaction.* Some spectral studies of the solid, solutions, and vapor phase have led to the conclusion that
the material is completely dimerized in the crystal, completely dissociated into AlH3 N(CHJ3 molecules +
(1) Work persormed at the Ames Laboratory of the Atomic Energy Commission. (2) To whom all correspondence should be addressed at E. I. du Pont de Nemours and Co., Explosives Department, Experimental Station, Wilmington 98, Del. (3) E. Wiberg, H. Graf, and R. Uson, 2.Anorg. Allgem. Chem., 272, 221 (1953).
(4) G. Schomburg and E. G. Hoffmann, 2. Elektroohem., 61, 1110 (1957).
Volume 71r Number 8 July 1967
D. G. HENDRICKER AND C. W. HEITSCH
2684
in the vapor, and that an equilibrium exists between dimer and monomer in inert solvents.sJ Other workers favor only a monomer on the basis of spectral eviden~e.~,~ The molecular weight measurements must be considered anomalous for several reasons. First, the degree of association in benzene, as reported by four independent l a b ~ r a t o r i e s ~and - ~ verified in this work, is a nonintegral constant (-1.4) over a wide range of concentration (0.05-0.5 F ) . Second, the complex behaves as a monomer in boiling ether.3 Finally, similar compounds such as BH3.N(CH3)3,9 GaH3.N(CH3)3,10 are all AlH, 2N (CH&, 3,' and A1H3.N(C2H5)35*12 monomers. We wish to report some additional evidence on the molecular weight, dipole moment, and nmr spectrum of this compound.
Experimental Section Chemical Preparations. The compound AlH3. N(CH& was prepared by the method of Ruff and Hawthorne from LiAIH4 and (CH3)3NHC1 in ether.5 After isolation and sublimation as described by these authors, the mixture of AlH,*2N(CH& and AlH3.N(CH& was exposed to vacuum a t room temperature until one-fourth to one-half of the original bulk had sublimed away. Since the diamine is more volatile than the monoamine, it sublimes more rapidly. The composition of the sublimate was checked periodically by the pyrolysis technique described below. When the ratio of H2/N(CH3)3 reached 1.5, the product was considered pure. All solvents used in this study were refluxed over and distilled from LiA1H4. Trimethylamine was dried and stored over CaHz in a steel cylinder. (CHa)3NHCl was prepared by mixing the amine and anhydrous HCI (cylinder gas purified by bubbling through concentrated H2S04) in anhydrous ether. LiAlH4 used in the synthesis of the complex was taken from a fresh can of 95% lithium aluminum hydride (Metal Hydrides, Inc.). The deutero complex A1D3.N(CH& was prepared in the same fashion from LiAID4. Because of the necessity of checking the hydrogenamine ratio frequently in the final stages of purification of the complex, the following rapid pyrolytic technique was devised. A sample, either weighed or unweighed, was sublimed through a heated (200") glass coil consisting of several turns of 6-mm tubing about 1 in. in diameter. The vapor decomposed as AlH,.N(CH3)3 +A1
+ ,/2H2 + N(CH&
The aluminum deposited as a mirror in the coil while the gases were separated by passage through a trap at - 196". The hydrogen and trimethylamine were The Journal of Physical Chemistry
measured separately in a gas buret. The pyrolytic method was compared with conventional wet chemical methods on an authentic sample of A1H3.N(CH&. A n d . Calcd for A1H3.N(CH3)3: H-, 3.36; Al, 30.3; N, 15.7; N(CH3)3, 66.3; H2/N(CH3)3, 1.50. Found (wet chemical): H-, 3.32; Al, 30.7; N, 15.6; (pyrolytic): H-, 3.35; N(CH&, 66.1 (N, 15.7); H2/N(CH3), 1.50. Solution Measurements. Cryoscopic molecular weights were determined with a thermistor (Sargent, Model 5-81630), a bridge (Sargent, Model 5-81600), and a Bristol recorder. The cell was equipped with stopcocks and standard taper joints for attachment to the vacuum line and control of the atmosphere over the solution. Samples were weighed under a dry helium atmosphere and sublimed into the cell. Solvents were purified by recrystallization before distillation from LiAlH4. Solvents, samples, and solutions for the dipole moment measurements were prepared in the same fashion as those used in the molecular weight determinations. The dielectric instrument, its calibration, and the treatment of data have been described elsewhere.', A portion of each solution was analyzed for Al, N(CH3)3, and H- by wet chemical methods. The results of these analyses agreed well with the composition data derived from the weights of the components except in a few cases where some decomposition had led to the presence of diamine as indicated by low values for Hand Al. I n these cases, the polarization data were corrected for the presence of the diamine using data published earlier." In no case did the correction amount to more than 5% of the total concentration. Samples for nmr were prepared in sealed tubes on a vacuum line.
Results and Discussion In benzene, over a concentration range from 0.04 to 0.187 F , the molecular weight was determined as 125 f 2, in cyclohexane, over a range from 0.050 to 0.076 ~
_
_
_
~
~
( 5 ) J. K. Ruff and M. F. Hawthorne, J . Am. Chem. Soc., 82, 2143
(1960). (6) R. Dautel and W. Zeil, Z. Elektrochem., 64, 1234 (1960). (7) H . Roszinski, R. Dautel, and W. Zeil, 2. Physik. Chem. (Frankfurt), 36, 26 (1963). (8) G. W. Fraser, N. N . Greenwood, and B. P. Straughan. J . Chem. Soc., 3742 (1963). (9) R. W. Parry, G . Kodama, and D . R. Schultz, J . Am. Chem. Soc., 80, 24 (1958). (10) D. F. Shriver and R. W. Parry, Inorg. Chem., 2, 1039 (1963); D . F. Shriver and C. E. Nordman. ibid., 2, 1298 (1963). (11) C. W. Heitsch, C. E. Nordman, and R. W. Parry, ibid., 2, 508 (1963). (12) E. Wiberg and H. Noth, Z . Naturforsch., 106, 237 (1955). (13) C. W. Heitseh and J. G. Verkade, in or^. Chem., 1, 302 (1962).
TRIMETHYLAMINE ALANE
iI
I ,5 14.
I
2685
0
1
I
I
K
I
I
1
I
X 0
-1 u)
a Q
= RUFF & HAWTHORNE-BENZENE
* = BENZENE A = CYCLOHEXANE D = CYCLOHEXANE 8 BENZENE X = WIBERG ,GRAF USON BENZENE
-
0.8F 1
1
I
I
.05
,I
.2
1
I
.4 MOLALITY OF ALUMINUM
.3
1
1
.5
.6
I
Figure 1. The cryoscopic molecular weight of AlHg .N(CH3)Ias determined in this work (benzene, cyclohexane, and cyclohexane -1 benzene) and by others (Ruff and Hawthorne6 and Wiberg, Graf, and Uson3).
F , S9.5 f 1.1, and in cyclohexane plus 0.13% benzene, at 0.050 F , 105. This can be compared with 89.1 for the formula weight. These results are presented graphically in Figure 1 along with those of Wiberg, et u L . , ~ and Ruff and H a ~ t h o r n e . ~ The anomaly of the measurements taken in the presence of even a small amount of benzene is a t once obvious. The compound behaves as a simple monomer in cyclohexane. The constancy of the molecular weight in benzene over a large range of concentration would seem to exclude two of the previous models of the apparent association. Both the equilibrium between a dimer and monomer, first proposed by Wiberg13 and the dipole-dipole interaction model of Schomburg and Hoffmann4 require the behavior of a monomer a t infinite dilution. That such is not the case may be inferred from the essentially zero slope of the degree of association vs. concentration plot.
In order to be certain that high molecular weights were not the result of accidental hydrolysis of part of the sample with the subsequent formation of diamine AlH3 2N(CH3)3, three of the cryoscopic molecular weight solutions were analyzed. The concentrations by analysis of the 0.125 m benzene and the 0.050 m cyclohexane solutions were less than 2% lower than the concentrations by weight of the original components. In the case of the 0.064 m benzene solution, the discrepancy was 6%, which would explain the marked deviation of this point from the more general 1.4 value for the apparent association. In spite of our best efforts, some hydrolysis of the samples had occurred. However, this decomposition cannot account for the marked deviation from monomeric behavior of the compound in benzene. The behavior of the AlH3 adduct can be contrasted with the alkylaluminum hydride adducts. Peters, Volume 7 1 , A'umber 8 J u l y 1967
D. G. HENDRICKER AND C. W. HEITSCH
2686
Table I : Dipole Moment of AIHJ N( CH& e
Xa
e
nD
Benzene Solutions
0.0664 0.0542 0.0332 0.0239 0.0214 0.0140 0.00982 0.00645 0.00395 0.00258 0
3.709 3.548 3.093 2.901 2.819 2.550 2.494 2.432 2.314 2.333 2.274
1.4925 1.4941 1.4959 1 .4968 1.4965 1.4968 1.4978 1.4974 1.4980 1 ,4979 1 ,4980
An explanation for the anomalous molecular weight in benzene is a breakdown of one or more of the fundamental conditions for cryoscopic molecular weight measurements such as the formation of a solid phase that is pure solvent. However, without additional data it is difficult to assign a cause to the anomaly observed here. Nevertheless, there can be little doubt that the cryoscopic molecular weight in benzene is an artifact and that trimethylamine alane is composed principally of a normal monomeric species under the conditions of this study. The relative polarity of the atoms and groups in AlHI. N(CH& may be assigned
H-
Cyclohexane Solutions
0.00251 0.00585 0.00381 0.00215 0
2.121 2.109 2.062 2.030 2.017
----
1,4229 1 ,4230 1.4234 1 ,4235 1.4235
Solvent-----
dt/hX bnDlhX Po,b cc D. C(jb
CeHs
C6Hl2
22.740 -0.0791 346.6 4.22 i0.05
15.229 -0.0996 345.0 4.11 f 0.08
’Mole fraction of
*
(CHJ)JN.AIHJ. Calculated by the method of P. Cohen-Hendriquez and C. J. F. Bottcher, “Theory of Electric Polarization,” Elsevier Publishing Co., Amsterdam, The Netherlands, 1952, p 302.
et al., report values of degree of association of 1.181.95 in freezing cyclohexane for R,H3-,Al*N(CH3)3 where R is methyl or ethyl and z is 1or Z.14 The similarities of the values for the dipole moment, 4.22 D. in benzene and 4.11 D. in cyclohexane, argue for the similarity of the species in both cases. The same conclusion can be drawn from the similarities between the spectra in the same two solvent^.^^^ The cryoscopic data favor the monomeric molecular weight of 89 observed in cyclohexane over the value of 125 measured in benzene. The infrared spectrum of the vapor of this compound has been interpreted in three different laboratories on the basis of a m ~ n o m e r . ” ~The ,~~ between the vapor andsolutionspectraimplies that the same species, the monomer, predominates under both conditions. The Of 4*1-4*2 for the moment Of A1H3*N(CH3)3 Can be ‘Ompared to the Of 4’69 observed for BH3.N(CH&.16 One would anticiDate .~ a slightly lower moment for the alane compared to the borane‘ This correspondence indicate that the alane, like t,he borane, is a monomer in solution. The Journal of Physical Chemistry
+
-
+/
(CH3),N-AI-H
-
\
H-
The total moment, p , can be considered as the vector sum of the individual link moments, p i j , that exist between the individual atoms i and j. Treating the sum of the H-C and C-N link moments as a group we may say p = pAl--N
-
llN(CHa)a
- 3 cos e p A l - H
where 8 is the N-AI-H bond angle. To a rough a p proximation, ~ N ( C H ~may ) ~ be taken as the value for free trimethylamine, 0.64 D.” For values of 0 close to a tetrahedral angle, 3 cos e will approximate unity. The value of p A 1 - ~ will undoubtedly be less than 1 D. and greater than that of a C-H bond, usually taken as 0.4 D.17318 The value of 0.6 D. is assumed here. These quantities in the above equation give 5.3 D. for an approximation of the AI-N link moment. This is in good agreement with the value of 5 D. predicted for this q ~ a n t i t y . A ~ large moment for the AI-N link had been postulated for the compound AIH3*2N(CH3)3.11*1sThe high polarization of this compound can be seen as arising from the flexing of the two symmetrically opposed but highly polar AI-N links, a form of atomic polarization. An alternative (14) F. M.Peters, B. Bartocha, and A. J. Bilko, Can. J . Chem., 41, 1051 (1963). (15) R. C.Taylor,D. G. Hendricker. and C. W. Heitsch, unpublished work. (16) J. R. Weaver and I t . W. Parry, Znorg. Chem., 5, 713 (1966). (17) C. P. Smyth, “Dielectric Behavior and Structure,” McGrawHill Book Co., Inc., New York, N. Y . , 1955. (18) R. J. Le Fe’vre, “Dipole hloments,” Methuen and Co. Ltd.. London, 1948. (19) C. W.Heitsch and R. N. Kniseley, Speclrochim. Acta, 19, 1385 (1963).
TRIMETHYLAMINE ALANE
explanation of the high polarization of the diamine adduct would be a flexing of the A1-H bonds. At this point there is relatively little evidence to permit a choice of one over the other, but there seems to be little doubt that one or the other or both of these forms of atomic polarization must be responsible for the large polarization of A1H3.2N(CH3)3. The nmr spectrum of AIH3.N(CH3)3 was observed in several solvents. Using Si(CH3)4 as an internal standard and reporting high field shifts as positive, the following results were obtained for the methyl (and A1H3) protons: C6H6, -1.87 ppm (-3.70); C6H12, - 2.39 ppm (- 3.42) ; 1,4-dioxane, - 2.29 ppm (Al-H absorption obscured by solvent) ; Si(CH3)4, -2.40 ppm (-3.48). I n those solvents, including benzene, in which both signals were free and unobscured by solvent adsorbtion, the ratio of the methyl to signal was 3.0 f O . l : l , consistant with the 9:3 ratio anticipated for this molecule. The values for the cyclohexane solutions can be compared with those of Peters, et al., who reported -3.54 ppm for AI-H proton absorption in both CH3AlHz.N(CH3)3 and C2H5A1H2. N(CH3)3.14 The same authors also give values around --2.0 ppm for the N-methyl absorption in these compounds. In all solvents, there seems to be only fair agreement with the results of Dautel and Zei120and with Ruff and Ha~thorne.~ This is probably attributable in large part to the choice of standards by these authors-a mixture of benzene and acetic acid for the first and water for the second. However, the same basic pattern, two peaks in a 3 : 1 ratio, was seen by these authors as well as in the present work. It will be noted that all of the shift values reported here for both signals are relatively constant in saturated solvents. In benzene, the methyl proton signal moves to high field relative to the saturated solvents in a fashion anticipated for the change in bulk susceptibility, while the AIHI signal moves in the opposite direction. A very similar situation is observed for (CH3)3NBH3where the methyl protons (and BH3 protons) absorb at -2.20 ppm (-2.13) in benzene and a t -1.6 ppm (-2.6) in 1,4dioxane and other saturated solvents.21 This is an-
2687
other striking similarity between the aluminum complex and the very normal, monomeric boron analog. In conclusion, we would like to point out that a monomeric structure for (CH3)3N-AIH3 is indicated by several pieces of data. These include the cryoscopic molecular weight in cyclohexane, the very normal dipole moment observed in both cyclohexane and benzene, the chemical shift values when compared to other similar aluminum compounds, the simplicity of the nmr spectrum, the similarity of behavior with trimethylamine borane of the nmr signals in benzene solutions, the ebullioscopic molecular weight in ether reported by Wiberg,3 and the interpretation of the vapor-phase infrared spectrum by Fraser and coworkers.8 Some of these by themselves carry little weight. Thus, the two-peak nmr spectrum would be anticipated also for a system of a dimer in a facile equilibrium with a monomer; the strongly basic ether would be expected to cleave a double hydrogen bridge in a dimer to give two monomers and the monomer would be expected to predominate in the vapor phase. Nevertheless, the collective evidence for the monomer is considerable. On the other hand, aside from a single very weak infrared band reported by Zeil and co-~orkers,~" the only evidence for a higher molecular weight is the cryoscopic determination in benzene. While this admittedly strange result has been observed in five different it can hardly be held as evidence for a dimer in equilibrium with a monomer. Such an equilibrium should be evident as a concentration-dependent variation in the cryoscopic data. Yet the combined work of three laboratories shows a constant degree of association over a tenfold range of concentration22(see Figure 1). It must then be concluded that evidence for a dimer is very weak and that the existing data are most consistent with a monomeric structure. (20) R. Dautel and W. Zed, 2. Elektrochem., 62, 1139 (1958). (21) C.W.Heitsch, Inorg. Chem., 4, 1019 (1965). (22) The reports of Schomburg and Hoffmann' and Dautel and Zeils did not give details with regard to the concentration behavior of the molecular weight. They did report the same degree of association observed here and elsewhere.
Volume 71, Number 8 July 1967