Matrix isolation ESR and theoretical investigations of 11B14N11B and

Lon B. Knight Jr., Devon W. Hill, T. J. Kirk, and C. A. Arrington. J. Phys. Chem. , 1992, 96 ... Vladimir E. Bondybey, Alice M. Smith, and Jürgen Agr...
0 downloads 0 Views 868KB Size
J. Phys. Chem. 1992,96, 555-561

555

Matrix Isolation ESR and Theoretical Investigations of l1Bl4Nl1Band loBl4Nl1B: Laser Vaporization Generation Lon B. Knight, Jr.,* Devon W. Hill, T. J. Kirk, and C. A. Arrington Chemistry Department, Furman University, Greenville, South Carolina 2961 3 (Received: June 25, 1991)

The BNB radical has been generated by the laser vaporization of solid boron nitride and isolated in neon, argon, and krypton matrices at 4 K for detailed electron spin resonance (ESR) investigations. These studies indicate that BNB is a significant vapor species above solid boron nitride. The ESR experimental results and theoretical calculations conducted at the CI level show that BNB has an X *Zground electronic state. Calculated nuclear hyperfine interactions (A tensors) for llB and 14N agree closely with the experimental results. A population analysis of the calculated wave function is compared with valence region orbital characters obtained from the commonly applied free atom comparison method used to relate A values to electronic structure. The unpaired electron is located primarily in sp orbitals on each boron atom with only a small amount of spin density on nitrogen. The magnetic parameters for BNB were practically identical in the three different rare gas hosts. The neon results (MHz) at 4 K are gll 2.0020 (4), g, = 2.0011 (3); ("B), A, = 434 (l), All = 485 (2); (14N) lAlll5 3, lAll = 20 (1).

Introduction Information concerning the vapor-phase composition above solid boron nitride under a variety of experimental conditions is significant since chemical vapor deposition and thin film formation of this material have many practical applications.' The ability to calculate and characterize the electronic structure and geometry of simple BxNytype molecules is crucial in understanding such potentially important processes. Experiments originally designed to produce BN+ for a matrix isolation ESR (electron spin resonance) investigation have yielded instead the first spectroscopic characterization of the symmetric BNB neutral radical. The BN+ molecular ion is predicted by theory to have a 4Z ground state.2 It would be especially interesting to study experimentally since it is isoelectronic to neutral BC which theory and matrix ESR results have shown to be X 4Z.3 The combination of pulsed laser vaporization and photoionization at 16.8 eV has been employed in our laboratory for previous rare-gas matrix ESR studies of several small radical ions? including AlF+, BF+,5 C2+,6SiO+, Si2+,' GaAs+,* AlH+,9 and PdH2+.I0 Sufficient vapor-phase amounts of BN,@,were apparently not produced under our experimental conditions since no BN+ ESR signals were detected. Mass spectral studies of laser-vaporized BN,,, under supersonic expansion conditions indicate the presence of BNB and larger BN clusters but practically no diatomic BN." However, different ionization efficiencies (1) Gmelins Handbook o/ Inorganic Chemistry, 8th ed.; Boron Compounds, 3rd supplement, Vol. 3; Springer: Berlin, 1988. See also references contained in: Matrin, J. M. L.;Francois, J. P.; Gijbels, R. J . Chem. Phys. 1989, 90,6469. (2) Karna, S. P.; Grein, F. Mol. Phys. 1985, 56, 641. (3) Knight, Lon B., Jr.; Cobranchi, S. T.; Petty, J. T.; Earl, E.; Feller, D.; Davidson, E. R. J . Chem. Phys. 1989, 90, 690. (4) Knight, L. B., Jr. Radical Ionic Systems; Lund, A., Shiotani, M., Eds.; Kluwer Academic: New York, 1991; p 73-97. Knight, L. B., Jr. Chemistry and Physics of Matrix-Isolated Species; Andrews, L.,Moskovits, M., Eds.; North-Holland: New York, 1989; Chapter 7. (5) Knight, L. B., Jr.; Earl, E.; Ligon, A. R.; Cobranchi, D. P.; Woodward, J. R.; &tick, J. M.; Davidson, E. R.; Feller, D. J. Am. Chem. SOC.1986,108, 5065. Knight, L. B., Jr.; Ligon, A.; Cobranchi, S. T.; Cobranchi, D. P.; Earl, E.; Feller, D.; Davidson, E. R. J . Chem. Phys. 1986, 85, 5437. (6) Knight, L. B., Jr.; Cobranchi, S.T.; Earl E. J . Chem. Phys. 1988,88, 7348. ( 7 ) Knight, L. B., Jr.; Herlong, J. 0.;Babb, R.; Earl, E.; Hill, D. W,; Arrington, C. A. J . Phys. Chem. 1991, 95, 2732. Knight, L. B., Jr.; Ligon, A.; Woodward, R. W.; Feller, D.; Davidson, E. R. J . Am. Chem. SOC.1985, 107, 2857. (8) Knight, L. B., Jr.; Petty, J. T. J . Chem. Phys. 1988, 88, 481. (9) Knight, L. B., Jr.; Cobranchi, S. T.; Gregory, B. W., Earl, E. J . Chem. Phys. 1987,86, 3143. (IO) Knight, L. B., Jr.; Cobranchi, S.T.; Herlong, J.; Kirk, T.; Balasubramanian, K.; Das, K. K. J . Chem. Phys. 1990, 92, 2721.

0022-365419212096-555$03.00/0

might influence these observations since photoionization at 7.9 eV using an F2 excimer laser was employed for the mass spectral observations. These ESR observations in three different matrices show that BNB is a significant vapor-phase species above laservaporized BN(s). No previous experimental studies have been reported for BNB except for a recently conducted matrix isolation FTIR investigation.I2 Previous theoretical calculations reveal that BNB is an extraordinarily stable species with an estimated dissociation energy of 265 kcal mol-' and a symmetric X2Z,+ ground state.13 This theoretical study also finds an extremely low bending frequency of 73 cm-' and a low vertical ionization potential of 6.75 eV. Such a low bending frequency has also been reported for the 11-valence-electron C3+molecular ion which is isoelectronic to BNB and CBC.I4 Theoretical and experimental matrix ESR investigations conducted in our laboratory for C3+and CBC will be described in a subsequent report.15 It is interesting that the theoretical results obtained thus far support a nonlinear ground-state structure for C3+and CBC while the results of this study clearly show BNB to be linear. The limited amount of experimental evidence available tends to favor a bent C3+ structure.I6 These ESR results reveal that the unpaired electron in BNB is located predominantly in sp orbitals on the boron atoms. Calculated isotropic and dipolar nuclear hyperfine interactions (A tensors) for IlB and I4N show reasonable agreement with the experimental A tensors which are virtually the same in neon, argon, and krypton matrices. Geometric optimization conducted in association with our hyperfine calculations yield a B-N bond length of 1.338 A. At the CI level of theory, our calculations also find a symmetric linear ground state (X 2Z) in agreement with the earlier theoretical study.I3 Small molecules containing boron recently studied by the rare-gas matrix isolation method include FTIR studies of B-H20 reactions" and ESR investigations of BCO which was found to (1 1) Morse, Michael D.; Chemistry Department, University of Utah, private communication. ( 1 2) Andrews, Lester, Chemistry Department, University of Virginia, private communication. (13) Martin, J. M. L.; Francois, J. P.; Gijbels, R. J . Chem. Phys. 1989, 90, 6469. Martin, J. M. L.;Francois, J. P.; Gijbels, R. Chem. Phys. Lett. 1990, 172, 354. (14) Grev, Roger S.; Alberts, Ian L.; Schaefer, Henry F., 111. J . Phys. Chem. 1990,94,3379. Raghavachari, K. Chem. Phys. Lett. 1990,171,249. (15) Knight, Lon B., Jr. To be published. (16) Faibis, A.; Kanter, E. P.; Tack, L.M.; Bakke, E.; Zabransky, B. J. J. Chem. Phys. 1987,91,6445. Vager, 2.; Kanter, E. P. J. Phys. Chem. 1989, 93, 7745. (17) Andrews, L.; Burkholder, R. G. J . Phys. Chem., in press. Geong, G. H.; Boucher, R.; Klabunde, K. J. J . Am. Chem. SOC.1990, 112, 3332.

0 1992 American Chemical Society

Knight et al.

556 The Journal of Physical Chemistry, Vol. 96, No. 2, 1992 HELIUM

REFRIGERATOR

IA

LIQ. N2

4K\

LASER _____ ----

i’v

ESR CAVITY

Figure 1. ESR matrix isolation apparatus used for trapping laser-vaporized species in rare gas h a t s at 4 K. The sample, solid boron nitride in these studies, is mounted on the end of a 6 mm diameter rod which can be admitted into the matrix apparatus through a direct insertion valve assembly labeled DIV without venting the vacuum system to air. See text. Following deposition onto a 4 K copper surface, the matrix sample is lowered into the X-band ESR cavity and an electromagnet (9 in. diameter pole face) with a 4 in. air gap is rolled into position for recording

ESR spectra. have a 4X ground state,Is BH2,I9HBBH,4 B2,20BC? and BNH.lS Experimental Section

The laser vaporization-matrix isolation ESR apparatus used in our laboratory has been described p r e v i ~ u s l y . ~ Its * ~ ,basic ~ features are shown in Figure 1. A closedcycle helium refrigerator (APD-HS304) was used to cool the copper deposition target to 4 K. The output of a Nd:YAG laser was frequency doubled to 532 nm for pulsed vaporization of the boron nitride sample (Alfa). A Scientech calorimeter (Model 3801) was employed to measure the laser energy which was typically in the range of 5-10 mJ/pulse at 10 Hz. Direct measurement of the laser vaporization power proved useful in reproducing experimental conditions and in carefully incrementing the laser power among a series of separate matrix deposition experiments. The laser beam was focused to approximately 0.5 mm and moved across the 7 mm diameter target by means of an external lens whose focal length was 40 cm. Numerous deposition were conducted at various laser powers, degrees of focusing (spot size), rates of rastering motion across the target, and rare gas matrix flow rates which were typically 6 sccm. The background pressure as measured by an ionization gauge in the vicinity of the matrix deposition target was approximately 8 X Torr before matrix gas flow; during neon deposition, this pressure increased to approximately 2 X Torr. Following a typical 30-min deposition the cryostat was lowered 6 cm by a hydraulic system so that the matrix sample was positioned inside the X-band TEIo2ESR cavity which remained at ambient temperature. The ESR cavity, employing 100-kHz modulation, was mounted on its connecting waveguide inside the high-vacuum system. Two new features have been added to the apparatus to reduce background impurities and facilitate changing the sample to be laser vaporized without venting the matrix vacuum chamber. A liquid nitrogen containing “can” or cold trap was mounted in direct line of sight a few inches from the matrix deposition target. This cylindrical cold trap had an open-ended pipe (19 mm diameter) welded into its center. It was through this pipe that the laser vaporization sample was admitted by means of a direct insertion valve assembly labeled DIV in Figure 1 and an antechamber-type (18) Hamrick, Y.M.; Van Zee, R. J.; Godbout, J. T.; Weltner, W., Jr.; Lauderdale, W. J.; Stanton, J. F.; Bartlett, R. J. J. Phys. Chem. 1991, 95, 2840. (19) Knight, L. B., Jr.; Winiski, M.; Miller, P.; Arrington, C. A.; Feller, D. J. Chem. Phys. 1989, 91. 4468. (20) Knight, L. B., Jr.; Gregory, B. W.; Cobranchi, S. T.; Feller, D.; Davidson, E. R. J. Am. Chem. SOC.1987, 109. 3521.

pumping arrangement (not shown). The sample was pressed inside a small copper cup. The backside of the copper cup was then epoxied to the end of a 30 cm long, 6 mm diameter brass rod which was vacuum sealed through an O-ring fitting. Hence new samples could be introduced and positioned right at the front surface of the liquid nitrogen trap without venting the entire matrix apparatus. We have found this cryogenic type pumping and sample handling arrangement to be highly efficient and helpful in reducing the level of background impurity radicals which are always present in such high-energy-type matrix trapping experiments. Experimental Results

On several laser vaporization matrix isolation ESR studies of this type, we have found that the trapping of molecular radicals is quite sensitive to the laser power employed and the degree of focusing. Obviously a certain minimum surface temperature is required to vaporize the solid sample. However, at excessive laser powers and/or too small a spot size (high degree of focusing), the ESR intensity of molecular radicals is observed to decrease. Apparently under such extreme laser bombardment, the major species vaporizing from the surface are atoms and even ions in so” cases. Hence there is an optimum range of laser energy and focuSing which produces the highest quality and most intense ESR signals of molecular radicals generated in this manner. The ESR spectra analyzed below and assigned to the BNB radical were obtained under such optimized experimental conditions. A total of 48 separate matrix depositions were conducted in the course of these BNB studies. The ESR spectra observed for neon, argon, and krypton matrices were virtually identical, with detailed analyses yielding the same magnetic parameters for the A and g tensors well within the experimental uncertainty. The two highest and lowest field ESR transitions observed for neon and argon matrices at 4 K are compared in Figure 2. The line shape is characteristically that of a randomly oriented, axially symmetric powder pattern sample. Except for BO, CH3, and HCO radical impurity lines?] no other molecular radical ESR signals were detected in the 0-8000 G range. On a few depositions involving intense laser vaporization energies, the B2(X3Z)radical was observed.20 The B2 molecule probably resulted from B atom recombination reaction during the condensation process. The overall nuclear hyperfine pattern shown in Figure 3 is a septet of triplets. An analysis of the higher order splittings, Mj states having different J origins, shows that the large septet hfi results from two equivalent Z = 3/2 nuclei. The coupling of such equivalent nuclei can yield Jvalues of 3, 2, 1, and 0. The MJ= f 3 lines can only arise from the overall J = 3 coupling, while MJ = 2 can arise from both J = 3 and J = 2, etc. Hence, if the A value is sufficiently large, the MJ= f 2 lines, for example, will exhibit such a higher order splitting as shown in Figures 2,3, and 5. The observed lines and quantum integer assignments for this ESR spectrum assigned to the l1BI4N1lBradical in a neon matrix at 4 K are listed in Table I. The hyperfine splittings show that the species observed is the symmetric BNB radical rather than the unsymmetric BBN molecule. The more intense features are the transitions which occur perpendicular to the molecular symmetry axis. These are labeled B = 90° lines in Table I and denoted by the symbol I in the various figures. The IlB isotope has Z = 3/2 and its natural abundance is 80%. The small triplet splitting of about 7 G on the perpendicular lines having a 1:l:l intensity ratio results from I4N(Z = 1) nuclear hyperfine interaction (hfi). Note that the relative phases of the first-derivative absorption pattern for the parallel and perpendicular lines at low field are reversed compared to the analogous features at high field. The I4N hfi along the molecular symmetry axis (A,,) was too small to be fully resolved. Based upon shoulder features observed under high-resolution conditions and overall line width, an upper limit of approximately 1 G can be established for the nitrogen A,, parameter. (21) Weltner, W., Jr. Magnetic Aroms and Molecules; Van Nostrand Reinhold: New York, 1983.

ESR Studies of IlB14N11Band IoB14BI'B

The Journal of Physical Chemistry, Vol. 96, No. 2, 1992 557

I

-3

MJ= 3 I

I

2918

3018

3118

3688

I

3788

3888

Figure 2. The highest and lowest field portions of the ESR spectra for IlBl4NILBtrapped in solid neon and argon matrices are compared. The two equivalent "B(Z = 3/2) nuclei produce a widely space septet pattern which shows higher order splittings for the inner Mj groups as explained in the text. EacH of these MItransitions perpendicular to the molecular symmetry axis are further split into smaller triplets by I4N(I=l). H atom impurity lies are quite common in such high energy trapping experiments. Some of the transitions arising from the impurity radical BO(X22) are also indicated. The magnetic field position corresponding to g, is 3418 G.

SIMULATED

3000

3200

I

3400

3600

38bO

Figure 3. An overall simulated ESR spectrum for the septet of triplets assigned to IIBI'NIIB obtained from an exact diagonalization calculation is presented in the lower trace and compared with a neon matrix sample in the upper trace. The dashed portion of a line in the M, = 2 group indicates overlap between a "BO line and the ugphase parallel component of a ILB1'NIIBline. See Figure 5. While I4Natom lines (triplet with A = 4 G near g,) are denoted in the experimental spectrum, other lines arising from impurities such as H atoms, HCO,"BO, and l0BO are not labeled. See text. The magnetic parameters for I'B14NIIB used to generate the simulated spectrum are listed in Table 11.

The magnetic parameters for 11B14N11B in all three rare gas hosts are listed in Table 11. The extremely large magnitude of the boron A values (A, = 434 and A,, = 485 MHz) clearly indicate that they are positive. However, the experimental results alone cannot determine the absolute signs of the small nitrogen A values. A detailed comparison with ab initio theoretical calculations discussed below indicates that both A,, and A, for 14N are most probably negative. The fact that all three rare-gas matrices yield virtually the same g and A values is most unusual. Ordinarily, a small shift of 1-3% in A values is observed as the host material is changed. Strong internal bonding in the BNB molecule and the small polarizability of the atoms involved might account for its environmental insensitivity.

An analysis was also conducted for the isotopic combination of 1oB14N11B whose ESR spectrum consisted of a large IlB quartet of I0B(I = 3) septets with each of the perpendicular components further split into the 7 G I4N triplets. The ratio of A values obtained from this analysis for IlB and 1°B showed agreement with the known ratio of nuclear g factors (p,/l). For example the A, parameter for 1°B in argon was 145 (1) MHz compared to 433 (1) MHz for IlB. This A value ratio of 2.99 (2)agrees within the experimental uncertainty to the known gN ratio of 2.986.*l Of course the individual 10B14N11B ESR lines were significantly less intense than those of llB14NllB given the smaller natural abundance of loB (20%) and its larger nuclear spin of I = 3. The consideration of the natural abundances of the two isotopes and

Knight et al.

558 The Journal of Physical Chemistry, Vol. 96, No. 2, 1992 TABLE 1: Observed and Calculated ESR Line Positions (gauss) for 11B14N11B in a Neon Matrix at 4 K"

perpendicular

(e = 9 0 0 ) ~

IJ;Mj)b 3;3 3;2 2;2

3;1 2; 1

1;l 3;O 2;o 1 ;o

obsd

calcd

268 2946 2953 3074 308 1 3088 3098 3105 3113 3217 3225 3232 3241 3248 3256 3258 3264 3272 3367 3374 3381

2938 2946 2953 3074 308 1 3088 3098 3105 3113 3217 3224 3232 3241 3248 3255 3257 3264 327 1 3367 3374 3381 3392 3399 3406 3408 3415 3422 3416 3423 3430 3528 3535 3542 3552 3559 3566 3567 3574 3581 3695 3702 3709 3719 3726 3733 3870 3877 3884

3399

d

0;O

3;-1 2;-1 1;-1 3 ;-2

2;-2 3:-3

3551 3558 3565 3567 3574 3582 3694 3701 3708 3718 3726 3733 3870 3877 3884

parallel

TABLE II: Observed Magnetic Parameters (MHz) for the Linear "BI4N"B Radical in Its X *EGround Electronic State in Rare-Gas Matrices at 4 K

(e = 0 0 ) c

calcd

298 1

2981

3047

3047

3069

3068

3208

3209

A,

g,

gll

All

IALI I 4

neon 2.0020 (4) 2.0011 (3) 434 ( 1 ) 485 (2) 20 (1) 53 argon 2.0025 (6) 2.0015 (4) 433 (1) 485 (1) 20 (2) 1 3 krypton 2.002 (1) 2.002 ( 1 ) 434 (2) 485 (3) 20 (2) 5 3 "The unusually large magnitude of the IlB A values indicates that they are positive. The sign of the I4N A values can not be determined from these experimental results alone. The best agreement with the theoretical calculation is obtained if both All and A , are taken as negative. The I4N hfs for the parallel (e = 0') lines could not be fully resolved. I1BI4N "B

3230

I4N

IlB

obsd

:

NEON 4 K

3231 1 .

3245

3245 3379 3400 3422 3422

1 'y e ZJq.10 I

I

2880

3860 3940 Figure 4. An expanded scale presentation of the highest and lowest field ESR transitions (MJ = 1 3 ) shows the axial nature of the g and A tensors in the 11B14NIIB radical. See overall spectrum in Figure 3. Note that

the parallel (11) lines, transitions along the molecular symmetry axis, do lines show 14Nnot show resolved I4N hfs while the perpendicular (I) ( I = l ) triplets with A , = 7 G.

3556 I1

3577

3577

3592

3591

3739

3739

3762

3761

393 1

3931

'Spectrometer frequency is 9587.1 (3) MHz. bAll of the J and M, combinations for two equivalent 11B(I=3/2)nuclei are listed. The smaller triplets ( A , z 7 G) for each of these IJ;MJ) boron hyperfine perpendicular lines results from I4N(I=l) hyperfine interaction. The I4N hfs is not fully resolved for the parallel lines. See Figures 2-5. Calculated line positions were obtained from an exact diagonalization analyses of the spin Hamiltonian using the magnetic parameters listed in Table 11. The direction 0 = 90' is perpendicular to the molecular axis of the linear BNB radical. The experimental uncertainty for the observed lines is i l G. dBNB lines in this region are obscured by common background impurities such as N atoms, H atoms, and CH, radicals. the number of hyperfine lines would predict an intensity ratio for the highest field I4Ntriplet in the two radicals of 8:l for 11B14N11B relative to IoB14N11B or IlBl4NIoB.The experimentally observed peak heights showed reasonable agreement with this predicted ratio which does not consider line-width differences and anisotropic effects. The final set of magnetic parameters for BNB listed in Table I1 for neon, argon, and kypton matrices was extracted from the observed ESR line positions given in Table 1 by an exact diagonalization analysis using the standard spin Hamiltonian for an axially symmetric radical with S = 1/2.5,21 The close agreement

I

2960

' 1 B 1 4 N 1 1: BNEON 4 K

Figure 5. This expanded scale presentation for the MJ = i 2 ESR lines of l1Bl4Nl1Bshows the higher order splitting that occurs when M, can originate from two different J values produced by the various coupling nuclei, J = 2 and 3 in this combinations of the two equivalent 11B(I=3/2) region. This splitting occurs for both the parallel and perpendicular lines, but only the perpendicular lines show resolved I4N(I=l)triplet hfs. See overall ESR spectrum for I'Bl4N"B in Figure 3.

with the calculated line positions is also apparent from Table I. Hyperfine interactions for all three magnetic nuclei in the 1oB14N11B and 11B14N*1B radicals were computed exactly, producing spin determinants of dimensions 168 X 168 and 96 X 96, respectively. A line fit within the experimental error of f l G was achieved without the inclusion of quadrupole interactions. The 0 vs H R output ~ of this exact diagonalization approach was coupled with a hrentzian line-shape program for generating simulated spectra. The exact diagonalization routine and lineshape program, written in our laboratory, have been described previo~sly.~" A comparison between the simulated overall spectrum for llBl4N1lBand that observed in a neon matrix at 4 K is presented in Figure 3. The agreement between the observed and simulated spectra is excellent in every detail, including relative phases, intensities, line positions, and the different J origins of the MJ components as discussed earlier. Expanded scale spectra for the Mj = f3 and M j = f2 tranare presented in Figures 4 and 5 , respectively. sitions of 11B14N11B The Occurrence of only two g and A components in these spectra provides strong experimental evidence that the radical is linear. A tensor anisotropy from two different nuclei environments are clearly recognizable in the parallel (11) and perpendicular (I)

ESR Studies of 11B14N11B and IoB14BIIB

The Journal of Physical Chemistry, Vol. 96, No. 2, 1992 559

Aimc

obsd"

calcdb

obsd'

451 (1) 17 (1)

418 17

-14 (1) 6 (1)

(All

- AL)/3

=

!&&N&3N((3

--*

-17

7

features of the spectrum. The absence of a third hyperfine component in the spectrum virtually eliminates the possibility of a nonlinear radical. This conclusion is supported by the theoretical calculations discussed below. Recording spectra a t elevated temperatures of 8, 20, and 40 K for neon, argon, and krypton matrices, respectively, did not significantly alter the spectral line shapes. Visible light irradiation of the matrices did not affect the intensity of the ESR signals assigned to BNB. For isolated ion radicals such photobleaching is known to cause a large decrease is signal intensity.4d,22 Hence this negative response to photobleaching provides supporting evidence for the assignment of the observed ESR spectrum to a neutral radical. Spectra recorded at various angles between the matrix trapping surface and the external magnetic field indicated no preferential orientation of BNB in any of the rare gas lattices. Electronic Structure: From Experimental Results. Using the commonly applied approximations involved in the free atom comparison method (FACM)?l the distribution of the unpaired electron in the ground X 2Z state of BNB can be estimated from the observed magnetic parameters. The results of this simple approach will be followed by an extensive theoretical calculation of the hyperfine interactions. This direct comparison with theory can be used to test the reasonableness of the FACM approach. Such a rigorous hyperfine comparison between experiment and theory has been conducted for only a few polyatomic radicals, including C202+and N4+.23*24It is well known that nuclear hyperfine interactions (hfi) are extremely sensitive to electronic structure and thus provide a rigorous test for theoretical calculations. This topic has been discussed in detail in the analysis of experimental/theoretical results for SiO" and BC.3 However, nuclear hfi need not be as sensitive to the electronic structure of the trapping medium, as observed in this case. The molecular gl and g , values for BNB do not deviate sufficiently from the k e e spin value (g, = 2.0023) to warrant a detailed analysis of spin-orbit coupling to excited states. The observed g tensor does support a ground 2Z state assignment. However, even if low-lying excited zII states were present, their effect on the g tensor would likely be small since the unpaired electron is predominantly located on boron in the ground state and the spin-orbit parameter for a boron 2p orbital is extremely small. The molecular Ah and Adip values for IlB and I4N in BNB can be obtained from the experimental AI,and A, parameters listed in Table I1 by the following relations:*'

=

-103.990

CI

calcdb

Experimental parameters are the same within the experimental uncertainty in neon, argon, and krypton matrices. The I4NA,,and A, parameters were taken as negative. Their absolute signs cannot be determined from experiment alone. See text. bThecalculated parameters were obtained at the five reference CI level containing 28 121 configurations. see Table V. 'The Ai, and parameters are the standard terms used in the analysis of ESR hyperfine data. See text for definitions.

Adip

-

*I03874

TABLE III: Comparison of Experimental and Theoretical Nuclear Hyperfine Panmeters (MHz) for BNB in Its X *EGround State I'B I4N

cos2 8 - l i p 3 )

where all symbols have their standard meanings and the averages are taken over the spin density. For BNB in neon, Ai, = 45 1 (1) (22) Auk, B. S.; Andrews, L. J . Chem. Phys. 1975.63, 1411. Bondybey, V. E.; English, J. H. J. Chem. Phys. 1979, 71, 777. (23) Knight, L. B., Jr.; Steadman, J.; Miller, P. K.; Bowman, D. E.; Davidson, E. R.; Feller, D. J. Chem. Phys. 1984, 80, 4593. (24) Knight, L.B., Jr.; Johannessen, K. D.; Cobranchi, D. C., Earl, E. A,; Feller, D.; Davidson, E. R. J . Chem. Phys. 1987, 87, 885.

L

0

I

I

\

1-103.998

c -104.001

4

-103.684 60

100

140

180

B-N-B Angle

Figure 6. Calculated energy as a function of B-N-B angle using MELDF. Both CI and SCF energies are displayed on different energy scales.

MHz and Adip= 17 (1) MHz for I'B; Aiso= -14 (1) MHz and = 6 (1) MHz for I4N provided both A, and All for nitrogen are negative. The same hyperfine values are obtained regardless of which matrix host was employed. These experimental A values and those calculated from the highest levels of theory used in the A tensor computations are compared in Table 111. For all four parameters the agreement is unusually good. The experimental molecular Aisoand Adipresults can be converted to orbital characters by taking simple ratios with a set of commonly employed free atom parameters.25 Using atomic parameters for IlB of Ah = 2547 MHz and Adip= 64 MHz; and Ai,= 1811 MHz and &, = 56 MHz for I4N the following results are obtained. The 2s a n i 2p, characters for a single boron atom in BNB are 0.18 and 0.27, respectively; for 14N, 2s = -0.01 and 2p, = 0.1 1. An independent check on the reasonableness of this FACM method and the experimental assignment is that the sum of these characters is essentially unity. For one unpaired electron this is certainly the expected qualitative result but the quantitative value of 1.0 must be somewhat fortuitous since the FACM approach involves several approximations. For example, overlap populations are ignored and the atomic parameters employed are calculated, rather than experimental parameters. Theoretical Calculations of the Nuclei Hyperfine Interactions (A Tensors). The initial geometry determination of BNB was carried out using GAUSSIAN 8626 with the 6-3 lG* basis set at the SCF, MP/2 and MP/4 levels. The result of these calculations are summarized in Table IV. At the highest level of theory, MP4/SDTQ, the molecule is linear with a 22,ground state. The E N bond distance is 1.338 A. The calculated charge distribution is -0.457 e on nitrogen and 0.229 e on boron. Both the SCF and MP2 levels of theory give a double-minimum energy surface with one minimum near a bond angle of 80' and a second at 180'. The MP4 calculation showed only the single minimum for the linear geometry. Molecular properties were calculated using the MELDF suite of programs developed by Davidson and ~o-workers.~'The optimum geometry was also examined using MELDF in a series of single-point energy calculations. The Dunning double zeta basis set with polarization functions was used in these calculations.28 The variation of energy with bond angle and bond length is shown in Figures 6 and 7. These potential energy surfaces display a double minimum with variation of bond angle. At the S C F level the Adip

(25) Morton, J. R.; Preston, K. F. J. Magn. Reson. 1978, 30, 577. (26) Frisch, M. J.; Binkley, J. S . ; Schlegel, H. B.; Ragahavachari, K.; Melius, C. F.; Martin, R. L.; Stewart, J. J. P.; Bobrowicz, F. W.; Rohlfing, C. M.; Kahn, L. R.;Defrees, D. J.; Seeger, R.; Whiteside, R. A,; Fox,D. J.; Fleuder, E. M.;Pople, J. A. GAUSSIAN 86; Carnegie-Mellon Quantum Chemistry Publishing Unit: Pittsburgh, 1984. (27) MELDF is a suite of programs that was used for the hyperfine calculations. It was developed and written by E. R. Davidson and colleagues. (28) Dunning, T. H., Jr. J . Chem. Phys. 1970.53, 2823.

560 The Journal of Physical Chemistry, Vol. 96, No. 2, 1992

Knight et al. ~~

SCF/UHF start linear start bent energy, au B-N bond length, 8, B-N-B angle

-103.68985 1.325 81.5

-103.67387 1.309 180.0

~

MP/4 SDTQ

MP/2

start linear

start bent

start linear

bent. fixed at 84.3O

-103.98456 1.328 180.0

-103.97932 1.338 84.3

-104.00985 1.338 180.0

-103.99814 1.338 84.3

TABLE V Nuclear Hyperfim Parameted (MHz) Calculated for BNB Using MELDFb IlB I4N no. of -

configurn A ,

five reference CI’s, threshold 1 X lod four reference CI’s, threshold 1 X 10” three reference CI’s, threshold 1 X 10” two reference Cl’s, threshold 1 X lod one reference CI, threshold 1 X linear one reference CI, 100’ threshold 1 X

28121

418

Adip A, 17.1 -17.4

Adip 6.9

27052

417

17.2 -17.2

6.9

20331

412

17.5 -19.2

6.9

10671

402

18.3 -23.3

6.9

5979

402

18.3 -22.0

6.9

9350

481

17.1 -14.9

6.2

-e z Q)

3

U

A comparison with the experimental A values is presented in Table 111. bSee ref 27 for the MELDF program. ‘ A , and Adipare defined in the text.

400 60

‘-30 100

140

180

6-N-6 ANGLE

Figure 8. Calculated isotropic hyperfine coupling constants using MELDF as a function of B-N-B angle. Calculationswere done at a bond length

of 1.326 A with one reference CI state and an energy threshold of 1 X au. This is a different calculationalmethod than the five-reference CI approach whose results for the A values are listed in Table 111.

I

ii

D

.lW.W02

+

1.30

,

I

1.31

.

D

I

I

1.32

1.33

.

I

1.34

.

I 1.35

B-N bond length In Angstroms

Figure 7. Calculated CI energy (au) using MELDF for linear BNB as a function of the B-N bond length.

lowest energy is for a bent molecule with an angle of looo, but for the CISD calculations the linear structure clearly has the lowest energy. The hyperfine parameters were calculated for the linear geometry with a bond length of 1.325 A at several levels of configuration interaction and at several angles with a single reference CI. The best values reported in Table V were determined for a multireference codiguration interaction calculation which included five reference configurations. Ai, for boron varies from 497 MHz at an angle of SOo to 402 M H z at 180°, while thc nitrogen Ais,, varies smoothly from -8.0 MHz a t 80° to -22.0 MHz at 180’. The dipolar hyperfine coupling constants exhibited negligible dependence on angle. The systematic inclusion of larger numbers of excited configurations in the C I calculation by lowering the energy threshold and increasing the number of reference configurations had little effect on the dipolar term but increased Airo for B by 4% and decreased the magnitude of the negative Ai,for nitrogen by 20%. The highest occupied orbital containing the unpaired electron is a a, orbital with most of the electron density on the boron atoms. Symmetry constrains this orbital to have only up orbital contributions from nitrogen with no s character so that the isotropic hyperfine parameter from nitrogen is small and arises mostly from spin polarization effects. On the boron atoms the orbital is best

described as an sp hybrid directed away from the nitrogen atom but bonding with respect to the up orbital on the nitrogen atom. The best Lewis structure representation is the pair of resonance structures: &==N=B: :B==N=B. Such a structure has a formal +1 charge on nitrogen and a formal -I/* charge on each boron, but the electronegativity difference gives rise to a net negative charge density on the nitrogen atom. FACM vs CIllCuLted Orbitalcharacters. A conceptually useful option in the MELDF suite of programs is the capability of projection of a particular property from the generalized M O wave function onto an atomic orbital basis set described in terms of the original basis functions (Figure 8). Use of this feature in the OCCUP program of MELDF gives a set of A0 spin densities that can be compared with the FACM analysis of the experimental Ai, and Adipparameters presented above. This analysis indicates an unpaired spin density of 0.17 in one boron 2s orbital, 0.31 in one boron 2pu orbital, 0.03 in the nitrogen 2pu orbital, and practically no nitrogen 2s character. It is important to recognize that these calculated orbital characters for the unpaired electron in BNB were obtained (or projected) from the calculated CI wave function based solely on the energy criteria. The Ai, and Adipparameters calculated from this same wave function show excellent agreement with experiment. Hence the close agreement between these calculated orbital characters, from such a rigorously tested wave function, and those obtained from the approximate FACM analysis procedure commonly applied to experimental ESR data to extract electronic structure information serves to validate the basic FACM approach. Only in a very few cases has it been possible to duectly test the FACM approximations despite its frequent application.‘,21 The boron FACM results of x = 0.18 and x2pcr = 0.27 are remarkably close to the calculated values of 0.17 and 0.3 1, respectively. The greatest deviation occurs in the nitrogen 2p, FACM result of 0.1 1 compared to the calculated value of 0.03.

-

J. Phys. Chem. 1992,96, 561-566 Based upon a previous detailed analysis for several diatomic molecules, this overage for the nitrogen 2p, character could have been anticipated.' With FACM, the nitrogen anisotropy is attributed solely to local 2p character while there are clearly anisotropic contributions from orbitals centered on boron, especially the large amount of spin density in the boron 2p, orbitals directed toward nitrogen. Even though there is an f 3 dependence, this nonlocal contribution will not be negligible since the B-N bond distance is small.

Conclusion The BNB radical has been identified by matrix isolation ESR spectroscopy as a vapor species above laser-vaporized solid boron nitride. The same experimental ESR results were obtained in neon, argon, and krypton matrices. Supporting evidence that BNB is a direct vapor species rather than a recombination reaction product formed during condensation is the fact that it was observed in a krypton matrix. The heavier rare gases under these experimental conditions are known to hinder or prevent such deposition or recombination reactions such as B + BN BNB or B2 N BNB. ESR and theoretical calculations indicate that the electronic ground state of BNB is linear X %"+. The large boron isotropic hyperfine interaction and the properties of the g tensor eliminate the possibility of a ground 211 state. A detailed analysis of the orbital containing the unpaired electron in BNB shows most of the spin density located in boron sp, orbitals. A test of the free atom comparison method (FACM) revealed that this approach yields orbital characters similar to those obtained directly from a calculated CI wave function that closely reproduces the observed hyperfine parameters, namely Ai, and AdiY The geometric de-

-

+

-

561

pendence of the calculated A values was conducted as a function of bond length and bond angle in BNB. Chemical vapor deposition (CVD) of the type that is typically used for thin film formation applications has apparently not employed the direct pulsed laser vaporization of a refractory material such as boron nitride. A homogeneous gas-phase method for forming BN films involves the heating of NH,(g) and B2H6(g).29 Film formation by the laser decomposition of a gas-phase compound in contact with a substrate surface has been studied in detail.30 An extensive review of laser-assisted deposition of thin films from gas-phase and surface-adsorbed molecules has been presented by Herman.31

Acknowledgment. Project support from the National Science Foundation (L.B.K., CHE-9019511) and the donors of The Petroleum Research Fund, administered by the American Chemical Society, is gratefully acknowledged. Undergraduate students were supported by an NSF-REU grant and a Duke Endowment grant to Furman University. A Du Pont College Science Grant to Furman's chemistry department provided valuable equipment support. The authors are indebted to Professor E. R. Davidson for the use of his MELD program for calculating nuclear hyperfine interactions and to Dr. David Feller for helpful discussions concerning various features of these programs. We are also indebted to Dr. Michael Morse at the University of Utah for kindly supplying the TOF-MS data on laser-vaporized boron nitride. (29) Adams, A. C.; Capio, D. C. J. Electrochem. SOC.1980, 127, 399. (30) Copley, S. M. J . Appl. Phys. 1988, 64, 2064. (31) Herman, I. P. Chem. Rev. 1989, 89, 1323.

Nuclear Magnetic Relaxation in Cyclopropenone M. T. Cbenon,*Vt C. Coupry,+and L. G. Werbelow*.t LASIR, CNRS, 94320 Thiais, France, and Laboratoire des Mgthodes Spectroscopiques, Centre St. JZrame, URA CNRS 773, Case 541, 13397 Marseille, Cedex 13, France (Received: June 28, 1991)

The nuclear spin relaxation characteristics of the oleftnic carbon in cyclopropenone were investigated over a range of temperatures. When relaxation-induced multispin order is monitored rather than the decay of athermal one-spin order, molecular detail otherwise obscured is rendered accessible. Known dipoledipole interaction anstants provide the basic description of molecular dynamics. With knowledge of the relevant molecular dynamics, it was possible to determine the orientation and antisymmetry of the carbon chemical shielding tensor.

Introduction Nuclear multispin relaxation is rapidly gaining deserved recognition as an important physicochemical probe which details the anisotropy of interaction, the modulation of interaction, and, most importantly, the temporal correlation between interactions. Although certain features of multispin relaxation are seeded in the works of Abragam, Solomon, and Lurqat' dating back to the 1950s, it was Redfield's development of a perturbative treatment of spin relaxation2 that provided direction for further exploration in this field. With a sound theoretical basis for describing the spin relaxation process, Hubbard, Blicharski, Shimizu, Schneider, McConnell, Freed, and others investigated 'crosscorrelation", 'anomalous relaxation", 'differential relaxation", and other embryonic concepts during the 1 9 6 0 ~ . The ~ arrival of the 1970s saw operator descriptions of the relaxation experiment being popularized by P ~ p e r . ~ Using these pioneering studies as a foundation, it has been established that temporal correlations between various time-deLASIR.

* Laboratoire des Methodes Spectroscopiques.

pendent spin interactions play a central role in multispin relaxation and are responsible for interconversions between various forms of spin order.5 Of course, during the 1980s, operator (multispin) descriptions of the multipulse 2D NMR experiment also became (1) Abragam, A.; Pound, R. N. Phys. Rev. 1953, 92, 943. Solomon, I. Phys. Rev. 1955,99, 559. LurGat, F. C. R. Hebd. Seances Acad. Sci. 1955, 240, 2402. (2) Redfield, A. G. Adu. Magn. Reson. 1965, I , 1 . (3) For listings of older literature in this field, see: Werbelow, L. G.; Grant, D. M. Adu. Magn. Reson. 1977, 9, 189. Werbelow, L. G.; Grant, D. M. J . Magn. Reson. 1915, 20, 554. (4) Pyper. N. C. Mol. Phys. 1971, 21, 1; 1972, 22, 433. (5) For listings of newer literature in this field see: Canet, D. Prog. NMR Specrrosc. 1989,21,237. Hartzell, C. J.; Stein, P. C.; Lynch, T. J.; Werbelow, L. G.; Earl, W. L. J . Am. Chem. SOC.1989,111, 5114. Also see: Elbayed, K.; Canet, D. Mol. Phys. 1989,68, 1033. Chang, W. T.; Wang, P. L.; Duh, D. M.; Hwang, L. P. J. Phys. Chem. 1990,94, 1343. Konrat, R.; Sterk, H. J . Phys. Chem. 1990.94, 1291. Foucat, L.; Chenon, M. T.; Werbelow, L. G. J . Phys. Chem. 1990,94,6663. Bull, T. E. J . Magn. Reson. 1988,80,480. Bull, T. E. J . Chem. Phys. 1990,93, 6824. Decatur, J. D.; Farrar, T. C. J . Phys. Chem. 1990, 94,7395. Krishnan, V. V.; Kumar, A. J . Magn. Reson. 1 9 9 1 , 92, 293. Kontaxis, G.; Muller, N.; Sterk, H. J . Magn. Reson. 1991, 92, 332. Fuson, M. M.; Anderson, D. J.; Liu, F.; Grant, D. M. Macromolecules 1991, 24, 2594.

0022-365419212096-561$03.00/00 1992 American Chemical Society