6294
J. Phys. Chem. 1995, 99, 6294-6300
Gas Phase Synthesis, Structure, and Dissociation of Boron Triazide R. L. Mulinax, G. S . Okin, and R. D. Coombe* Department of Chemistry, University of Denver, Denver, Colorado 80208 Received: December 1, 1994; In Final Form: January 31, 1995@
Gas phase B(N3)3 is readily prepared by a spontaneous room temperature reaction between HN3 and BC13. The infrared spectrum of the reaction products exhibits strong absorptions at 2163 and 1360 cm-' and a weaker absorption near 1100 cm-'. These features are assigned to B(N3)3 by comparison with results from ab initio calculations of the geometry and vibrational frequencies of the molecule. B(N3)3 has a strong UV cmz. The density of gaseous B(N3)3 prepared in these absorption at 230 nm, with 0230 = 1.3 x experiments decays with a time constant near 30 min. This decay is attributed to dissociation at the vessel walls, a process which leaves a visible film. Infrared spectra of films produced by either room temperature dissociation or UV photodissociation of B(N3)3 exhibit features attributable to BN, likely in the presence of excess nitrogen.
Introduction Nitrogen-rich molecules are well known for their ability to store energy. Recent research on such species has focused on small molecules with extremely high proportions of nitrogen, such as N4, Ng, or azide-substituted compounds.'-3 In this paper, we present new information about azide compounds of boron. Apart from their ability to store large amounts of energy, these species hold additional interest as possible precursors for boron nitride thin films. Both hexagonal and cubic BN films have many important applications as either wide band gap semiconductors or as tribological coating^.^ Boron azides have been known for more than forty years. In 1954, Wiberg and Michaud5 synthesized boron triazide, B(N3)3, by the reaction of diborane with HN3 in an ether solution at low temperature. The reaction evolved H2 to leave B(N3)3, and the conditions were adjusted to produce an HZ yield approximately 95% of that expected from the 3:l stoichiometry of the reaction. These authors used similar methods to produce Al(N3)3, as well as extremely energetic adducts of B(N3)3 and Al(N3)3 with NaN3 and LiN3. In 1963, Paetzold6 reported the synthesis of ClzBN3 from the reaction of BC13 with LiN3 in a CH2C12 solution. The product was obtained as a crystalline solid identified as trimeric dichloroboron azide, (BC12N3)3. Upon heating to 200 "C, the compound converted to hexachloroborazole by loss of NZ from the trimer and migration of the chlorine atoms. The crystal structure of (ClzBN3) was determined by Mueller7 in 1971, the results confirming the trimeric structure. Dichloroboron azide was also studied by Wiberg and Michaud8 in 1972. These authors produced it from the reaction of BC13 with trimethylsilyl azide in CH2C12. In 1978, Dehnickeg reported the generation of azides of aluminum, gallium, and boron from reactions of the triiodides of these metals with iodine azide in benzene. He recorded the infrared spectra of the monoazide products, I2MN3, and observed the formation of oligimers of these species. Alkylboron azides have been generated using similar methods. In 1966, Paetzoldloproduced (CH3)2BN3 from the reaction of (CH3)zBBr and tri-n-butylsilyl azide, obtaining the product as an explosive liquid. Recently, ab initio computational methods have been used" to determine the structure of (CH3)2BN3. Similar reactions of organic azides with halogenated boron compounds have been used by organic chemists for the synthesis of heterocyclic ring compounds of @Abstractpublished in Advance ACS Abstracts, April 1, 1995.
nitrogen. Very recently, Boehmer and co-workers3reported the synthesis of tetraazidodiborane from the gas phase reaction between HN3 and diborane at 400 K. This species may well be an intermediate in the low-temperature, condensed phase synthesis of boron triazide described by Wiberg and Michaud5 in their seminal 1954 paper. To our knowledge, boron triazide has not been explicitly studied since Wiberg and Michaud. In this paper, we describe a simple method for preparing B(N3)3 in the gas phase by the stoichiometric reaction of BCl3 with HN3. The IR and UV absorption spectra of the molecule are reported, and the assignment of infrared absorption features is supported by the results of ab initio calculations of its structure and vibrational frequencies. As expected from the calculated heat of formation, gas phase B(N3)3 is found to be rather short lived, and we report measurements of its time decay at room temperature. Thermal and photolytic dissociation of B(N3)3 at room temperature produce thin films with spectroscopic features similar to those of boron nitride. The results suggest the utility of this novel species as either a high-energy material or as a precursor for BN films.
Experimental Methods Boron triazide was generated by the spontaneous room temperature reaction between gaseous BCl3 and gaseous HN3. Samples of HN3 were produced by the well-known reaction12 of NaN3 with excess stearic acid near 90 "C. The HN3 was stored in Pyrex bulbs and diluted with He to produce mixtures which were about 8% HN3. The purity of the HN3 was examiced by FTIR spectroscopy, and the absolute amount of HN3 present in the mixtures was determined from UV absorption spectra. BCl3 (CP grade) was obtained from Matheson Gas Products and was not purified further before its use in these experiments. The gases were handled with a MoneYstainless steel vacuum system evacuated by a mechanical pump. Pressures in the system were measured with capacitance manometers. The reagents were mixed in small 10 cm cells used for recording IR and UV spectra or in a large 1 L cell used for photolysis of samples of B(N3)3. Two different small cells were used. One of these was a stainless steel cell with its internal walls coated with halocarbon wax (Halocarbon Inc.) to provide an easily passivated, nonreactive surface. This cell could be fitted with
0022-365419512099-6294$09.00/0 0 1995 American Chemical Society
Boron Triazide
J. Phys. Chem., Vol. 99, No. 17, 1995 6295
either IR (KC1) or UV (fused silica) windows. The other cell was fabricated from Pyrex, had fused silica windows, and was used primarily for UV spectra. The large cell was constructed from 10-cm-diameter Pyrex pipe and was equipped on one end with a 10-cm-diameter fused silica window and on the other with an aluminum plate. As noted below, consistent results were obtained with all three cells. IR spectra were recorded with a Nicolet 5DXC FTIR spectrometer (2 cm-’ resolution). UV spectra were recorded with a Milton-Roy 3000 diode array spectrometer. Ab initio calculations were performed using the GAUSSIAN 92 family of programs13 on an IBM RS-6000 701 1-25T work station.
Results and Discussion 1. Reaction of HN3 with BCb. Boron compounds are well known for their ability to form adducts with electron donors.14 For example, planar BX3 compounds form adducts with NH3 via donation of the lone pair electrons on the nitrogen to the unoccupied out-of-plane orbital localized on the boron atom. In HN3, both the N atom bound to hydrogen and the terminal N atom have lone pairs and are electron rich, so the reaction with BCl3 is expected to proceed by formation of a transient “adduct” produced by donation of these electrons to the boron atom:
BC13
+
4000
8
HN3
(la)
cl\B/
I
c1 ,-;N
9
/ ” ’
BCI,
+
HN,
-
H,N2 Ci
u
(1b)
\B/cl
I
c1 Of the two possible configurations for the adduct, that labeled a above will likely dissociate without reaction whereas l b can react by elimination of HCl to leave an azide-substituted boron compound. This process is expected to be roughly thermoneutral; presumably, a similar process does not occur for the BC13/ NH3 adduct because of the substantially stronger N-H bond in NH3 relative to that in HN3. The chloroboron azide compound thus formed can undergo two more addition-elimination processes with HN3, generating two more HCl molecules and producing a fully azidified boron molecule. Hence, the overall process would be as follows: BC1,
2250
500
Frequency (cm-l)
+ 3HN3 - B(N3)3+ 3HC1
(2)
The maximum amount of boron triazide is therefore expected to be produced by reaction of one unit of BCl3 with three units of HN3. Gaseous BCl3 was allowed to react at room temperature (295 K) with an HN3/He mixture (8.5% HN3) in either a wax-coated
Figure 1. (a) Infrared spectrum of gaseous BCl3. (b). Infrared spectrum of gaseous H N 3 . (c) Infrared spectrum of the products of the reaction between gaseous BCl3 (0.3 Torr) and gaseous H N 3 (0.9 Torr). Vertical bars in c indicate the results of an MP2/6-31G(d) calculation of the infrared frequencies of B(N3)3.
stainless steel cell or a Pyrex glass cell as noted above, and the products were interrogated with IR and UV absorption spectroscopy. A typical reaction mixture included 0.30 Torr of BC13 and varying partial pressures of the HN3/He mixture, up to about 1.2 Torr. Figure l a shows an IR spectrum of 0.30 Torr of BC13 alone, and Figure l b shows an IR spectrum of 10.61 Torr of the HN3/He mixture alone, which contained 0.90 Torr of HN3. Figure IC shows the IR spectrum of products of the reaction between these two species (10.91 Torr), taken 5 min after mixing (to allow for purging of the sample compartment of the FTIR). Features attributable to the BCl3 and HN3 reactants are clearly absent, indicating complete reaction. A number of new peaks corresponding to reaction products are evident. The spectrum is dominated by intense peaks at 2163 and 1360 cm-’, and includes a smaller group of peaks near 1100 cm-’. Although the 2163 cm-’ peak is close to that corresponding to the N3 asymmetric stretch15 in HN3, it is clearly different as the latter is a doublet as shown in Figure lb. Further, the other HN3 features (the symmetric stretch near 1160 cm-’, another doublet, and the NH stretching feature near 3340 cm-’) are no longer present. Tentatively, we assign the 2163 cm-’ peak to the N3 asymmetric stretch in azide groups bound to boron. This assignment is reasonable since the 1360 cm-’ peak is clearly assigned to the B-N stretching vibration.I6 The peaks near 1100 cm-’ lie close to those of the N3 symmetric stretch in HN3 and are assigned to a similar vibration in the new boron azide molecule. The group of lines between 2600 and 3200 cm-’ clearly corresponds to HC1, an expected product of the reaction. The fact that the spectrum in Figure IC is so simple suggests that the absorber responsible for the peaks at 2163,
6296 . I . Phys. Chem., Vol. 99, No. 17, 1995
Mulinax et al.
0'3
a, 0
C
.E m c
F
I0.0
1.o
2.0 3.0 [HN3]/[BC13]
4.0
5.0
Figure 2. Stoichiometryof the BC13/HN3 reaction, shown as a plot of the intensities of IR features corresponding to BC13 (triangles),the N3 group in B(N3)3 (closed squares), the B-N stretch in B(N3)3 (asterisks), and H N 3 (open squares).
1360, and 1100 cm-' is a highly symmetric molecule, as might be expected to be the case for B(N3)3. This issue, as well as the tentative assignment of the features in the spectrum, was investigated by ab initio computations of the geometry and IR frequencies of the molecule, as discussed below. The stoichiometry of the reaction was probed by measuring the absorbance of the BC13 peak at 950 cm-', the HN3 peak at 3340 cm-', and the product peaks at 2163 and 1360 cm-' as the HN3/BC13 proportion was varied from 0 to 4. Data were recorded from many different experiments, with several different batches of HN3. The experiments were performed with the waxcoated metal cell, and in every case the reagents were allowed to react for precisely 5.0 min before the IR spectrum was recorded. In fact, the reaction was very much faster than this, but the delay was necessary to allow for complete purging of the sample chamber of the FTIR spectrometer. These data are shown in Figure 2 as plots of absorbance vs the HN3/BC13 proportion. It is apparent that the absorbance of the product peaks grows linearly with proportion up to HNflCl3 = 3.0, at which point the intensities level off and decline. The HN3 absorbance is zero until this proportion is reached, beyond which unreacted HN3 appears in the spectra. These results are strongly indicative of the 3:l reaction stoichiometry suggested by eq 2 above. Although the HC1 produced by the reaction is clearly evident in the IR spectrum of the products (Figure IC), its absorbance was not a good diagnostic of the progress of the reaction. We believe that this result may be attributed to adherence of the HC1 to the wax-coated cell walls. The HCl absorbance might well be a good diagnostic for reactions in a Pyrex cell. The absorbance of the BC13 feature at 950 cm-' declines linearly with increasing HN3/BC13 proportion up to about 1.O, beyond which this feature is clearly absent from the spectrum. This result shows that the BCl3 is completely removed by 1 equiv of HN3 but also implies that the frequency of the B-C1 stretch is shifted considerably in the product (presumably ClzBN3). The data shown in Figure 3 show that this is indeed the case. The figure shows the spectral region around the 950 cm-" peak in greater detail. As the HN3/BC13 proportion increases, the absorbance of the 950 cm-' peak declines as noted above, but a new peak appears to grow in at about 975 cm-" and is very clearly evident for HN*C13 > 1.O. As HNflCl3 increases still further, the intensity of this peak declines, and the feature is nearly absent by HN3/BC13 = 3.0. We tentatively assign this transient peak to the B-Cl stretch in the intermediate Cl2BN3.
0.504 1040
9 5 Frequency (cm-')
Figure 3. Infrared spectra in the 915 to 1040 cm-' range of the products of the BC13 + H N 3 reaction at different initial HNflC13 ratios: (a) [HN~I/PC131 = 0; (b) [HN#[BC131 = 1.37;(c) p-IN31/[BC131 = 2.07; (d) [HNs]/[BClsl = 3.03.
Another factor may also be at work here. ClzBN3 has been shown to t r i m e r i ~ e ,forming ~.~ a crystalline solid. The trimers would not have the ability to react with additional HN3. The data in Figure 2 show that excess HN3 (Le., beyond an HN3/ BCl3 proportion of 1.0) is completely reacted up to the stoichiometric proportion of 3.0. If, however, 1 equiv of HN3 is mixed with BCl3 and time is allowed for trimers to form, additional HN3 added at that point should not react. This was found to be the case. If a 5.0 min delay was allowed between addition of the first and the second equivalent of HN3, unreacted HN3 was clearly evident in the IR spectrum. 2. Calculations of the Structure and Frequencies of Boron Azides. Ab initio calculations of the structure and frequencies of boron azides were performed in an effort to corroborate the tentative IR assignments described above. The calculations were performed with the GAUSSIAN 92 family of programs13 and made use of several different basis sets at both the HartreeFock and MP2 levels of theory. First, a geometry optimization was performed at the HartreeFock level for B(N3)3. Because of hardware limitations, full optimizations were not done; instead, partial optimizations were performed which assumed the similarity of the N3 groups in the molecule (C3 symmetry). This assumption seemed reasonable in light of the simplicity of the IR spectrum shown in Figure IC above. Variables were defined which described the planarity of the central BN3 part of the molecule, the B-Na-Np bond angles, the Na-NB-N, bond angles, the dihedral angles along the B-Na-Np-N, linkage, the angles between the azide groups, and the B-Na, N,-Np, and Np-N, bond lengths. Hartree-Fock calculations were performed with 3-21G, 6-3 1G(d), and 6-31+G(d) basis sets, and results for the latter two cases are shown in Table 1. In general, the molecule is found to be a nearly planar pinwheel structure with the azide groups arrayed symmetricallyabout the boron atom, as shown in Figure 4. The B-Na-Np angles near 118" and the Na-Np-Ny angles near 175" are very similar to those found for other azides.17 The bond lengths in the system are in good agreement with those found for (CH&BN3 by Hauser-Wallis and co-
Boron Triazide
J. Phys. Chem., Vol. 99, No. 17, 1995 6297
TABLE 1: Calculated Geometries for B(N& parameter' HF/6-31G(d) HF/6-31+ G(d) MP2/6-31G(d) B-N 1.440 1.439 1.442 1.240 1.241 1.241 Na-Np 1.092 1.092 1.166 NI3-W AI =LXI -B -Na 90.0 90.2 90.2 AZ=LB-Na-Np 117.5 117.7 120.4 A3= LXi-B-Na-Nb 90.0 90.0 90.0 & = LNa-Np-Xz 89.0 89.8 90.7 = LXz-Np-X3 89.0 89.8 90.7 = 1x2-Np-N, 89.0 89.8 90.7 As= LB-Na-Np-Xz 93.5 94.2 79.1 A6 = LNa-Np-X3 87.4 87.4 86.5 87.4 = LX3-Nb-N, 87.4 86.5 A7= LNa-B-Na 120.0 120.0 120.0 Parameters defined as in Figure 4. Distances in angstroms, angles in degrees. (I
N
IY N 19
TABLE 2: Calculated Frequenciee for B(N3)3 HFl6-3 1G(d) HFl6-31+G(d) MP2/6-31G(d) experiment
Vl71V18
V21h2
v23IvZ4
1209 (134) 1202 (144) 1145 (181) 1100
1435 (1556) 1426 (1583) 1451 (862) 1360
2583 (1094) 2575 (1152) 2202 (841) 2163
Frequencies in cm-', followed by relative IR intensities in parentheses. Remaining 18 frequencies have significantly smaller intensities. spectrum (Figure IC) in the sense that only three doubly degenerate frequencies have significant IR intensity and that among these three there are two strong bands corresponding roughly to asymmetric stretches in the N3 groups and along the BN bonds and one weaker band involving more symmetric stretches in the N3 groups. Reasonable agreement with the experimental frequencies is obtained if the calculated results are multiplied by a scaling factor of 0.89, as is normally the case for HF frequencies. The MP2 frequencies show much better agreement. The results of the MP2 calculation are shown in Figure IC as a stick spectrum. Here again only 6 of the 24 frequencies (3 groups of two very nearly degenerate frequencies) have significant IR intensity, and the relative intensities among these features agree very well with experiment. The frequencies themselves are also in reasonable agreement with Figure IC, although all three are somewhat high. This agreement would appear to be such as to confirm our tentative assignment of the IR spectrum in Figure ICto B(N3)3. A much more definite assignment might be made from observations of isotopically mixed samples at low temperatures, Le., in a matrix. Although we hope to perform such experiments in the future, they were not within the scope of the present work. The heat of formation of B(N3)3 was determined from the calculated MP2/6-3 1G(d) energy of this molecule (accounting for the zero point energy) relative to the energy of B 9/2N2 calculated at the same level of theory. This treatment yielded a result AHf.298 = 21.2 kcal/mol, a value comparable to that of BH3.18 From this result and the known heat of formation of N3 radicals,19 the B-N bonds in the molecule are calculated to have a dissociation energy near 108 kcal/mol. From heats of formation,'* the enthalpy change for the overall generation of B(N3)3 (reaction 2) is calculated to be -156 kcal/mole. As expected, this process appears to be driven by the instability of HN3. From the strength of the B-C1 bond18 in BC13 and the average B-N bond strength in B(N3)3, however, the first step in the process (reaction lb) is expected to have an exothermicity of only about 11 kcal/mol. We note also that decomposition of B(N3)3 to solid hexagonal BN and four N2 molecules is quite energetic, liberating roughly 8 1 kcal/mol. Calculations were also performed for BC13, C12BN3, and C1B(N3)2 in an effort to understand the shift in the B-Cl stretching frequency among these molecules. Geometry optimizations and frequency calculations at the HF/6-3 1+G(d) level were performed for all three molecules, and calculations at the MP2/ 6-31G(d) level were performed for BCl3 and ClzBN3. Although comparison of the results for BC13 with the experimental spectrum shown in Figure la indicated that the calculations were accurate to within only about 50 cm-', it was apparent that the B-C1 stretching vibration does indeed shift about 25 cm-I to the blue on going from BCl3 to C12BN3, and shifts still further for ClB(N3)z. While the IR intensity of this feature is still moderate (Le., relative to the azide stretch) for C12BN3, it is considerably weaker for ClB(N3)z. Hence, these results support the idea that the transient feature observed near 975 cm-l in the data shown in Figure 3 is a chloroboron azide intermediate, probably C12BN3.
+
Figure 4. Calculated geometry of the calculation.
B(N3)3,
showing the variables of
workers." There is a slight twist along the Na-Np bond which moves the terminal nitrogen atom (N,) very slightly out of plane. The dipole moment of the molecule is still very nearly zero, however. The HF calculations with the 6-31G+G(d) basis set gave results which were very similar to those obtained with the 6-31G(d) basis. Geometry optimizations were also done at the MP2 level of theory with 3-21G and 6-31G(d) basis sets. The MP2/6-31G(d) results are shown in Table 1. The MP2/ 3-21G results were in substantial agreement with the HartreeFock data with the exception of the B-Na-Np bond angle, which is determined to be much larger (127" rather than 118") by the MP2 calculation. The MP2 calculation with the larger 6-31G(d) basis set finds 120" for this angle, a more reasonable value. These geometry optimizations appear to have established at least the general features of the B(N3)3 structure reasonably well. The central question is whether the infrared spectrum for this structure agrees with that shown in Figure IC. Frequency calculations were carried out at both the HF and MP2 levels of theory. Table 2 shows the major frequencies and IR intensities calculated at the HF/6-3 lG(d), HF/6-31+G(d), and MP2/63 1G(d) levels, based on analogous geometry optimizations. There is very little difference between the results of the two HF calculations, as was the case for the geometry optimizations at this level. The HF results agree well with the experimental
Mulinax et al.
6298 J. Phys. Chem., Vol. 99, No. 17, 1995
d
a,
a, 0
0
.i! E I/)
!.-4
c
c
E VJ
C
C
e
I 1
+e
O 4000 0.55
*
2250
!-
500 d
L
0.754 . 2000
Wave numbers (cm- 1)
0.8 A 0.7-
0.6-
C
0.5-
0.4-
0,
.->
-
2
0.3-
Figure 7. Infrared spectra in the 500-2000 cm-' region of films deposited by the dissociation of B(N3)3. (a) Film from thermal decomposition at 298 K, on KCl. (b) Film from photodissociation, on MgF. Arrows indicate frequencies ascribed to hexagonal BN (1390 cm-l) and cubic BN (1080 cm-') (ref 20).
-
0.2-
B(N3)3
0.1 -
or 04 0
5
for the decay. It seems that the loss of B(N3)3 from the system is due to either self-reaction (either a unimolecular or multimolecular process) or reactions at the wall. Two observations favor the latter possibility. First, the observed time decay (Figure 6) did not vary substantially with variations in the amount of B(N3)3 initially present, as it would if multimolecular processes (e.g., trimerization) contributed substantially to the decay. Further, no new IR features grow in as the decay proceeds, as shown in Figure 5. This result suggests that dissociation of the molecule produces N2 and species which are deposited on the walls, Le.,
0
e 98
'
Wavenumbers (cm-1)
Figure 5. Infrared spectra of the products of the stoichiometric reaction between BC13 and HNs, at different times after mixing: (a) 3 min, (b) 13 min, (c) 32 min, and (d) 62 min.
a
t
1250
10
20
30
40
50
60
Time (min) Figure 6. Time decay of B(N3)3: squares from infrared spectra; triangles from UV spectra. 3. Time Decay of B(N3)3. Boron triazide was observed to be a relatively short-lived species at room temperature, with a lifetime quite sensitive to impurities in the gas mixture. Figure 5 shows the changes in the IR absorption spectra of a sample of B(N3)3 over a period of 62 min. The loss of B(N3)3 from the gas mixture was monitored by observation of the decay of the IR absorptions of the molecule at 2163 and 1360 cm-', and these data are shown in Figure 6. The data shown in the figure represent the best conditions (Le., best gas purity and smallest cell leak rate) achieved with our apparatus. It is clear that even under these circumstances virtually all of the azide is gone within 1 h. A single-exponential fit to the data (shown in the figure), indicates the e-l time constant of the decay to be about 30 min. Measurements of the leak rate of the cell (in this case the wax-coated steel cell) indicated that far less than 10% of the azide originally present might be removed by presumed reaction (see below) with 0 2 or H20 from a leak duri,ng this period, and hence this possible mechanism is not responsible
-
B(N3)3
+ 3N,
(34
BN -t- 4N2
(3b)
BN,
where either the BN3 from reaction 3a or the BN from 3b would be taken up as a film. Indeed there was some evidence of film formation on the cell windows. After several experiments involving repetitive production of B(N3)3 in the cell, recording of spectra, and evacuation, features at 1260 and 720 cm-' began to grow in the IR spectra of the evacuated cell. These absorptions are shown in Figure 7. These frequencies are close to those reportedZofor hexagonal BN, i.e., 1390 and 800 cm-', and the intensities are in the appropriate proportion. The lower frequencies in the spectum of Figure 7 may well indicate weaker bonds in a BIN, film, perhaps because of an excess of nitrogen (see the discussion below). Indeed, processes such as 3a above might be expected to lead to films with such an excess. The features shown in the figure were very reproducible and appeared on the windows of both the waxed steel and glass cells at room temperature. Gaseous B(N3)3 was found to decay much more rapidly than indicated in Figures 5 and 6 when a small leak of laboratory air was introduced into the cell. In this case, both the 2160 and 1360 cm-' features of spectra such as Figure ICwere rapidly
Boron Triazide
J. Phys. Chem., Vol. 99, No. 17, 1995 6299 12.91
I
A
0.8 '*O:
12.8
6
h
12.7-
t 2 12.62
v)
E v)
12.5-
m + 12.4-
12,3LL--12.20
0.04
200
250 300 350 Wavelength (nm)
400
Figure 8. Ultraviolet absorption spectrum of B(N3)3 at a pressure of 0.3 Torr.
lost, and HN3 absorptions (see Figure lb) grew in. No 1260 cm-I peak associated with window deposits appeared. In keeping with the known proclivity for other boron compounds to react in air,4 we postulate that the H20 present adds to the boron atom via the unpaired electrons on the oxygen atom, forming an adduct which eliminates HN3: B(N3),
+ 3/,H,0 - 3HN3 + '/,B203
(4)
We presume that the B203 formed as a coproduct is rapidly lost on the walls. 4. Ultraviolet Absorption of B(N3)3. Ultraviolet and visible absorption spectra of B(N3)3 were recorded with a Milton Roy 3000 diode array spectometer, using a 10 cm Pyrex cell equipped with fused silica windows. The samples of B(N3)3 were prepared as described above, from 0.3 Torr of BCl3 and a 3:l stoichiometric excess of HN3. Spectra were recorded within 2 min of initial mixing of the reagents. No absorption was observed in the region from 300 to 900 nm. A very strong absorption was recorded at 230 nm, and this spectrum is shown in Figure 8. A plot of the absorbance at 230 nm vs the B(N3)3 density (the latter assumed to be equivalent to the initial BC13 density in the reaction mixture) was linear, and the slope indicated an absorption cross section 0230 = (1.3 i0.1) x lo-'' cm2. The uncertainty noted represents the precision of the data. The largest source of error is of course the assumption of 100% conversion of the BC13 initially present to B(N3)3. Taking this into account, the absorption cross section noted above is a lower limit. The magnitude of the 230 nm absorption was used to observe the time decay of the B(N3)3. Here again, the absorption was found to decay over a few tens of min. These data are plotted in Figure 6 along with the IR data. Clearly, the decay rates agree well with one another, supporting the assignment of the UV absorber as B(N3)3. Many azides show an absorption onset in the 200 to 300 nm region,21,22although few have absorptions as strong as that shown in Figure 8 at wavelengths longer than the vacuum UV. The longer wavelength absorptions are thought to correspond to n z* transitions in the electrons of the N3 chain,z3leading to dissociation of the weak RN-N2 bond. Since the B(N3)3 molecule has three such weak RN-N, bonds, a number of interesting reactive species might be formed depending on how many of these bonds are ruptured by photolysis, in either a concerted or sequential manner.
-
2
4
8 1 0 1 2 1 4
6
Time (min) Figure 9. Pressure rise from photolysis of a sample initially containing B(N& at 0.3 Torr, the remainder HCl (0.9 Torr) and He. The sample was irradiated between 3 and 11.5 min.
For continuous illumination, it is expected that all three bonds will break, leading to three NZ molecules and a BN3 radical:
-
+ hv NB(N3), + N, NB(N3), + hv - N,B(N3) + N, B(N3)3
N,B(N3)
+ hv - BN, + N,
(5) (6)
(7)
Although the energy of a single 230 nm photon (124 k c d mol) is likely sufficient to break all three azide bonds at once, this would be unusual (although there are some well known examplesz4of such multiple bond rupture). In any case, either sequential or concerted rupture of the three bonds will generate three N2 molecules, resulting in a pressure rise in the system. If it is assumed that the BN3 radicals end up as a boron-nitrogen deposit on the vessel walls, then the pressure rise expected from complete photolysis of 0.3 Torr of B(N3)3 would be 0.6 Torr:
+
0.3 Torr of B(N3)3 hv
-
BN3(s)
+ 0.9 T of N,
(8)
If on the other hand the BN3 fragment also absorbed the UV radiation or dissociated on the walls to BN, then the pressure rise would be 0.9 Torr: 0.3 Torr of B(N3)3
+ hv - BN(s) + 1.2 Torr of N,
(9)
Hence these possibilities, as well as others in which fewer than three NZ molecules are evolved per B(N3)3 molecule, are readily distinguished by monitoring the pressure rise upon complete photolysis of a known sample of B(N3)3. These experiments were performed by using the large 1 L cell described above. This cell and the smaller 10 cm IR cell were attached to the vacuum line via a tee, such that IR spectra of gases in the smaller cell could be used to ascertain the production of B(N3)3 upon the mixing of BCl3 and H N 3 in the system. Pressures in the large cell were measured with a capacitance manometer attached directly to the cell. Irradiation of the gases in the large cell was accomplished with a Varian R300-2 xenon lamp (50 W), with the output expanded to effectively fill the volume of the cell. Figure 9 shows a plot of the total pressure in the cell (including Ar diluent) vs time for irradiation over 8.5 min. The experiment was performed several times, with consistent results. As shown, the pressure rises with time and levels off well before the lamp was turned off. The
Mulinax et al.
6300 J. Phys. Chem., Vol. 99, No. 17, 1995 total pressure rise is 0.6 Torr for an initial B(N3)3 pressure of 0.3 Torr, indicating dissociation to BN3 and 3N2. The presumption that the BN3 produced by the photolysis was removed as a solid was supported by observation of a milky white deposit on the cell walls and window. In appearance, these deposits were much like those produced by thermal dissociation of B(N3)3 as described above. An IR spectrum of such a film deposited by photolysis of B(N3)3 over a 2.54-cmdiameter MgF2 substrate is shown in Figure 7, along with the spectrum of the film deposited by thermal decomposition. Both spectra show absorptions in the region characteristic of BN. The film deposited from photolysis has a broad absorption with a maximum near 1400 cm-', near the frequency expected for hexagonal BN (1390 cm-'), but also shows a weaker peak with a maximum at 1100 cm-', near the frequency characteristic of cubic BN (1080 cm-'). The 1260 cm-' feature observed from thermal decomposition of B(N3)3 is also present. Clearly photolysis of B(N3)3 produces a film somewhat different from that produced by thermolysis. The photolysis pressure rise data (Figure 9) indicate that only three of the possible four molecules of N2 are liberated by photodissociation, strongly suggesting that the resulting B-N film is nitrogen rich. This is likely to be the case for thermolysis as well. The differences in the spectra may be reflective of differences in the distribution of this excess nitrogen in the films. The spectrum of the thermolytic film shows features much like those of hexagonal BN, but shifted about 100 cm-' to the red. This might be interpreted as resulting from uniform distribution of the excess nitrogen through the film. On the other hand, the photolytic film appears to be a mixed phase entity, with IR frequenciesquite close to those which are nominal for hexagonal and cubic BN, as well as the red-shifted features of the thermolytic film. The excess nitrogen may not be uniformly distributed in this case. This interpretation is surely speculative. In any event, these first results certainly bode well for the prospect of making much purer BN films from experiments in which finer control is exerted over conditions like the substrate temperature, photolysis geometry, and gas purity.
Summary and Conclusions The stoichiometric gas phase reaction of HN3 with BC13 at a 3:l mixing ratio would appear to be a facile means for the generation of B(N3)3, an energetic molecule potentially useful for the generation of boron nitride films. Ab initio calculations of the structure of the molecule indicate that it should have a C3 near-planar pinwheel geometry. Calculated IR spectra based on this geometry are in good agreement with experimental results. The calculations also support the assignment of IR features observed for HN3:BC13 mixing ratios less than 3:l to partially azidified boron species (e.g., C12BN3), in agreement with a hypqthetical reaction mechanism in which B(N3)3 results from three sequential additiodelimination steps. B(N3)3 has an intense UV absorption at 230 nm, with ~ 2 3 0= 1.3 x lo-'' cm2. Pressure rise data indicate that, under continuous irradiation in this wavelength region, all three weak azide bonds are broken to yield three N2 molecules and a reactive BN3 species. The latter is lost on the vessel walls, presumably as a nitrogen-rich boron-nitrogen film. Similar films are produced by the thermal decomposition of B(N3)3 at room temperature, a process which occurred with an exponential time constant near 30 min in our apparatus. IR spectra of these films indicated features very similar to hexagonal BN, with frequencies shifted slightly to the red (likely due to the presence of excess nitrogen).
It seems clear from the experiments and the associated calculationsthat B(N3)3, ClB(N32, C12BN3, and probably radical derivatives of these species like BN3 can be made and their spectroscopy studied in much greater detail, perhaps by lowtemperature matrix isolation methods. Such work would surely add to that reported recently on other BxNyspecies.25 It would also be interesting to continue the study of films generated by the decomposition of boron azides. Most BN films produced to date have been rich in boron, not n i t r ~ g e n .It~ may be that the excess nitrogen in films produced from compounds like B(N3)3 could be removed by gentle annealing, yielding much higher quality BN films. Finally, we note that similar chemistry might well be employed to produce azide compounds of aluminum, gallium, or indium, some species of which are already known.26
Acknowledgment. This work was supported in part by grants from the U.S. Air Force Office of Scientific Research (Grant No. F49620-93-1-063 1) and the National Science Foundation (Grant No. CHE-9300383). References and Notes (1) Lauderdale, W. J.; Stanton, J. F.; Bartlett, R. J., J. Phys. Chem. 1992, 96, 1173. (2) Benard, D. J. J. Appl. Phys. 1993, 74, 2900. (3) Benard, D. J.; Boehmer, E; Michels, H. H.; Montgomery, J. H. J. Phys. Chem. 1994, 98, 8952. (4) Synthesis and Properries ofBoron Nitride, Materials Science Forum, Vol. 54/55: Pouch. J.. Alterovitz. A,. Eds.: Trans Tech: Aedermannsdorf. Switzerland, 1990). (5) Wibera. E.; Michaud. H. Z. Naturforsch. 1954. 96. 497. (6) Paetzoid, P. I. Z. Anorg. Allg. C h k 1963, 326, 47, 53, 58, 64. (7) Mueller, U. Z. Anorg. Allg. Chem. 1971, 382, 110. (8) Wiberg, N.; Joo, W. C.; Schmid, K. H. Z. Anorg. Allg. Chem. 1972, 394, 197. (9) Kreuger, N.; Dehnicke, K. Z. Anorg. Allg. Chem. 1978, 444, 71. (10) Paetzold, P. I. Z. Anorg. Allg. Chem., 1966, 345, 79. (11) Hauser-Wallis, R.; Oberkammer, H.; Einholz, W.; Paetzold, P. 0. Inorg. Chem., 1990, 29, 3286. (12) Schlie, L. A.; Wright, M. W. J. Chem. Phys. 1990, 92, 394. (13) Frisch, J. M.; Trucks, G. W.; Head-Gordon, M.; Gill, P. M.; Wong, M. W.; Foreman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M. A,; Replogle, E. S.; Gomperts, R.; Andres, J. L.; Raghavachari, K.; Binkley, J. S.; Gonzalez, C.; Martin, R. L.; Fox, D. J.; Defrees, D. J.; Baker, J.; Stewart, J. J.; Pople, J. A. GAUSSIAN 92; Gaussian, Inc.: Pittsburgh, PA, 1992. (14) Brinck, T.; Murray, J. S.; Polizer, P. Inorg. Chem. 1993,32,2622. (15) Dows, D. A.; Pimentel, G. C., J. Chem. Phys. 1955, 23, 1258. (16) See for example: Bellamy, L. J.; Gerard, W.; Lappert, M. F.; Williams, R. L. J. Chem. SOC.1958, 2412. Also: Dagard, J.; Mongeot, H.; Alchekzai, H.; Tachagnes, J. P. Can. J. Chem. 1976, 54, 2135. (17) Treinen, A. The Chemistry of the Azido Group; Patai, S., Ed.; Interscience: London, 1971; Chapter 1. (18) Schlegel, H. B.; Harris, S. J. J. Phys. Chem. 1994,98,11178. Chase, M. W.; Davies, C. A.; Downey, J. R.; Frurip, D. J.; McDonald, R. A.; Szverud, A. N. JANAF Thermochemical Tables, 3rd ed.; J. Phys. Chem. Re$ Data 1985, 14. (19) Continetti, R. E.; Cyr, D. R.; Osbom, D. L.; Leahy, D. J.; Neumark, D. M. J. Chem. Phys. 1993, 99, 2616. (20) Cholet, V.; Vandenbulke, L.; Rouan, J. P. J. Mater. Sci. 1994, 29, 6 . Ishihara, R.; Sigiura, 0.; Matsumura, M., Appl. Phys. Lett. 1992, 60, 3244. (21) Coombe, R. D.;Patel, D.; Pritt, A. T., Jr.; Wodarczyk, F. J. J. Chem. Phys. 1981, 75, 2117. (22) Okabe, H. J. Chem. Phys. 1968, 49, 2726. (23) Baronavski, A. P.; Miller, R. G.; McDonald, J. R. Chem. Phys. 1978,30,119. McDonald, J. R.; Rabalais, J. W.; McGlynn, S. P. J. Chem. Phys. 1970, 52, 1332. Clossen, W. D.; Gray, H. B. J. Amer. Chem. Soc. 1963, 85, 290. (24) See for example: Nathanson, G.; Gutlin, B.; Rosen, A. M.; Yardley, J. T. J. Chem. Phys. 1981, 74, 361. Yardley, J. T.; Gutlin, B.; Nabanson, G.; Rosen, A. M., J. Chem. Phys. 1981, 74, 370. (25) See for example Hassanzadeh, P.; Andrews, L. J. Phys. Chem. 1992, 96, 9177. JP943201W