Reactions of laser-ablated beryllium atoms with hydrogen cyanide in excess argon. FTIR spectra and quantum chemical calculations on BeCN, BeCN, HBeCN, and HBeNC

Article abstract of DOI:10.1021/ja970740g

Laser-ablated beryllium atoms have been reacted with hydrogen cyanide (H¹²CN, H¹³CN, and D¹²CN) during condensation in excess argon at 6-7 K. In the matrix infrared spectrum, the major products observed are BeNC, BeCN, HBeNC, and HBeCN. Consistent with typical beryllium bonding, these new beryllium species are linear molecules. Density functional theory calculations on these products with the BP86 functional and 6-31 1G* basis sets predict vibrational frequencies extremely well, even for HBeCN where mixing between the nearly isoenergetic Be-H and C≡N stretching modes causes significant complications in the spectra. Although B3LYP and MP2 calculations are more sophisticated than the BP86 method, they do not predict the vibrational frequencies of these products nearly as well. More important is the carbon 12/13 isotopic frequency ratio as a description of the normal modes, and the BP86 method generates 12/13 ratios much closer to observed values for HBeCN than frequency ratios from the more time-consuming CISD method.

Full text of DOI:10.1021/ja970740g

6392
J. Am. Chem. Soc. 1997, 119, 6392-6398
Reactions of Laser-Ablated Beryllium Atoms with Hydrogen
Cyanide in Excess Argon. FTIR Spectra and Quantum
Chemical Calculations on BeCN, BeNC, HBeCN, and HBeNC
Dominick V. Lanzisera and Lester Andrews*
Contribution from the Department of Chemistry, UniVersity of Virginia,
CharlottesVille, Virginia 22901
ReceiVed March 7, 1997X
Abstract: Laser-ablated beryllium atoms have been reacted with hydrogen cyanide (H12CN, H13CN, and D12CN)
during condensation in excess argon at 6-7 K. In the matrix infrared spectrum, the major products observed are
BeNC, BeCN, HBeNC, and HBeCN. Consistent with typical beryllium bonding, these new beryllium species are
linear molecules. Density functional theory calculations on these products with the BP86 functional and 6-311G*
basis sets predict vibrational frequencies extremely well, even for HBeCN where mixing between the nearly isoenergetic
Be-H and CtN stretching modes causes significant complications in the spectra. Although B3LYP and MP2
calculations are more sophisticated than the BP86 method, they do not predict the vibrational frequencies of these
products nearly as well. More important is the carbon 12/13 isotopic frequency ratio as a description of the normal
modes, and the BP86 method generates 12/13 ratios much closer to observed values for HBeCN than frequency
ratios from the more time-consuming CISD method.
Introduction
the linear species BeCCH and HBeCCH.12 Because acetylene
is isoelectronic with hydrogen cyanide, these products can be
viewed as analogues of BeCN and HBeCN, respectively. This
study substitutes HCN for C2H2 in an attempt to observe and
characterize beryllium cyanides with and without hydrogen and
to determine their structures. The BeCCH and HBeCCH
molecules were characterized by DFT/B3LYP13-15 frequency
calculations to support the vibrational assignments. Here, we
will compare the pure DFT functional, BP86,16,17 with the hybrid
B3LYP functional as well as the MP218-20 and CISD21-23
methods, to help identify these new molecules.
A previous study of reactions of laser-ablated boron with
hydrogen cyanide produced the linear products BNC and BCN,
as well as the nonlinear products HBNC and HBCN.24 Also
present in the product spectra was the cyclic molecule HB(CN),
where the boron is bonded to both the carbon and nitrogen.
Because Be orbitals do not usually adopt sp2 or sp3 hybridiza-
tion, all products, in contrast to B compounds, should be linear.
The results presented in this article demonstrate that four new
beryllium cyanide products are formed and trapped in solid
argon, and all of these products have linear geometries.
In studies of alkali cyanides, the potential energy surface was
computed to be flat in the coordinate describing the metal
position around the CN.1-5 The bonding between the metal
and cyanide was almost entirely ionic, and the ground state
structure for these molecules was described as T-shaped, not
linear. By contrast, Bauschlicher et al. calculated alkaline earth
cyanides and found that all such molecules prefer the linear
isocyanide structure.6 A barrier to interconversion to the linear
cyanide structure exists for the smaller molecules, BeNC and
MgNC, but not for CaNC and BaNC. Although this barrier is
only 0.23 eV at the SCF-CI level for MgNC, the barrier is much
more substantial for BeNC at 0.71 eV. These calculations
suggest that beryllium cyanides exist in a linear form as either
the cyanide or isocyanide with no stable intermediate structure.
The first laser ablation matrix infrared experiments with
beryllium uncovered several new compounds. Reactions of Be
with H2, O2, and H2O generated molecules such as BeH2,
ArBeO, OBeO, and HBeOH.7-11 For each of these products,
the geometry about Be was linear with sp hybridization, the
preferred geometry for alkaline earth species.
(12) Thompson, C. A.; Andrews, L. J. Am. Chem. Soc. 1996, 118, 10242.
(13) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(14) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett. 1989,
157, 200.
(15) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(16) Becke, A. D. Phys. ReV. A 1988, 38, 3098.
(17) Perdew, J. P. Phys. ReV. B 1986, 33, 8822.
(18) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett. 1988,
153, 503.
In a matrix isolation study of the reactions of laser-ablated
Be with C2H2, the major new products were determined to be
X Abstract published in AdVance ACS Abstracts, June 15, 1997.
(1) To¨rring, T.; Bekooy, J. P.; Meerts, W. L.; Hoeft, J.; Tiemann, E.;
Dymanus, A. J. Chem. Phys. 1980, 73, 4875.
(2) van Vaals, J. J.; Meerts, W. L.; Dymanus, A. Chem. Phys. 1984, 86,
147.
(3) Wormer, P. E. S.; Tennyson, J. J. Chem. Phys. 1981, 75, 1245.
(4) Klein, M. L.; Goddard, J. D.; Bounds, D. G. J. Chem. Phys. 1981,
75, 3909.
(19) Frisch, M. J.; Head-Gordon, M.; Pople, J. A. Chem. Phys. Lett. 1990,
166, 275.
(20) Frisch, M. J.; Head-Gordon, M.; Pople, J. A. Chem. Phys. Lett. 1990,
166, 281.
(5) Marsden, C. J. J. Chem. Phys. 1982, 76, 6451.
(6) Bauschlicher, C. W., Jr.; Langhoff, S. R.; Partridge, H. Chem. Phys.
Lett. 1985, 115, 124.
(21) Pople, J. A.; Seeger, R.; Krishnan, R. Int J. Quantum Chem. Symp.
1977, 11, 149.
(22) Krishnan, R.; Schlegel, H. B.; Pople, J. A. J. Chem. Phys. 1980,
72, 4654.
(23) Raghavachari, K.; Pople, J. A. Int J. Quantum Chem. 1981, 20,
167.
(24) Lanzisera, D. V.; Andrews, L.; Taylor, P. R. J. Phys. Chem. A In
press.
(7) Tague, T. J., Jr.; Andrews, L. J. Am. Chem. Soc. 1993, 115, 12111.
(8) Thompson, C. A.; Andrews, L. J. Am. Chem. Soc. 1994, 116, 423.
(9) Thompson, C. A.; Andrews, L. J. Chem. Phys. 1994, 100, 8689.
(10) Andrews, L.; Chertihin, G. V.; Thompson, C. A.; Dillon, J.; Byrne,
S.; Bauschlicher, C. W., Jr. J. Phys. Chem. 1996, 100, 10088.
(11) Thompson, C. A.; Andrews, L. J. Phys. Chem. 1996, 100, 12214.
S0002-7863(97)00740-3 CCC: $14.00 © 1997 American Chemical Society

Reactions of Laser-Ablated Be with HCN in Solid Ar
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6393
Table 1. Observed Frequencies (cm-1) of Products from
Beryllium-Hydrogen Cyanide Reactions
Experimental Section
The apparatus for pulsed laser ablation, matrix isolation, and FTIR
spectroscopy has been described previously.24-26 Mixtures of 0.3%
HCN, H13CN or DCN in Ar codeposited at 3 mmol/h for 2 h onto a
6-7 K cesium iodide window react with beryllium atoms ablated from
a target source (Johnson-Matthey, lump, 99.5% Be) rotating at 1 rpm.
The fundamental 1064-nm beam of a Nd:YAG laser (Spectra Physics
DCR-11) operating at 10 Hz and focused with a f ) +10 cm lens
ablated the target using 20-30 mJ per 10 ns pulse. The procedure for
preparing HCN was described previously.24,27 For DCN, we added D2-
SO4 (Aldrich) and D2O (Aldrich) to solid KCN (Aldrich) and, because
of prior HCN passivation of the manifold, the H/D ratio approached
unity. For H13CN, we added concentrated HCl to K13CN (Cambridge
Isotopes) and obtained a 13C enrichment of greater than 80%. Following
deposition, a Nicolet 550 Fourier transform infrared (FTIR) spectrom-
eter collected infrared spectra from 4000 to 400 cm-1 by using a liquid
nitrogen cooled MCT detector; the resolution was 0.5 cm-1 with a
frequency accuracy of ( 0.2 cm-1. After sample deposition, annealing
to 15 K followed by broadband (240-580 nm) mercury arc photolysis
(Philips 175 W) produced changes in the FTIR spectra. Further
annealings to 25 and 35 K also changed some of the spectral features.
12/1a
13/1a
12/2a
phot/annb
identity
2202.8
2183.1
2159.4
2153.2
2128.6
2088.7
2085.7
2044.3
1971.0
938.4
912.4
805.8
800.1
530.1
2165.6
2134.6
2159.4
2149.5
2116.7
2050.8
2050.9
2002.0
1971.0
932.0
906.0
802.2
795.3
528.3
2194.2
2183.1
1674.2
1631.3
1642.1
2088.7
2098.6
2044.3
1477.3
938.4
895.6
805.8
750.9
436.1
+35/-35
+620/-60
+5/-5
+20/-10
+35/-35
-45/-5
+20/-10
-10/-60
+10/+10
-45/-5
+20/-10
+620/-60
+35/-35
+35/-35
+35/-35
+20/-10
+20/-10
HBeCN
BeCN
BeH2
HBeNC
HBeCN
BeNC
HBeNC
CN
BeH
BeNC
HBeNC
BeCN
HBeCN
HBeCN
HBeCN
HBeNC
HBeNC
431.5
422.1
418.0
525.4
521.7
545.4
521.7
a Isotopes for carbon/hydrogen. b Percent increase or decrease on
photolysis/annealing to 25 K.
We perfomed density functional theory (DFT) and Hartree-Fock
(HF) calculations on potential product molecules using the Gaussian
94 program package.28 Calculations used either the BP86 pure DFT
functional, the B3LYP hybrid DFT functional, the MP2 method, or
configuration interaction with singles and doubles (CISD). All
calculations employed the 6-311G* basis sets for each atom29,30 or for
comparison in BP86 calculations, Dunning’s correlation consistent
double ú-basis set (cc-pVDZ)31,32 for the H, C, and N atoms, plus the
D95* basis sets for Be atoms.33 The geometry optimizations used
redundant internal coordinates and converged via the Berny optimization
algorithm,28,34 and the program calculated vibrational frequencies
analytically. Of the four methods, the least expensive BP86/6-311G*
calculations provided the best results, as will be described.
Results
Matrix infrared spectra for various isotopic combinations are
reported as well as the relative change in intensity of product
peaks following broadband photolysis and subsequent annealing
to 25 K. Besides the beryllium product peaks mentioned in
this section, we observed bands for HNC and CN, apparently
formed from radiation in the ablation process.35 Also, oxides
such as the (BeO)2 ring at 866.3 and 1131.2 cm-1, BeOBe at
1412.4 cm-1, and Ar-BeO at 1526.1 cm-1 are formed8-10 due
to oxides on the target surface and reaction of beryllium with
residual H2O or D2O from the synthesis of HCN or DCN.
Figure 1. Matrix infrared spectra in the 2220-1960-cm-1 Be-H and
CtN stretching regions following pulsed laser ablation of Be atoms
codeposited with Ar/HCN (300/1) samples on a CsI window at 6-7
K: (a) Be + H12CN, (b) Be + H13CN, and (c) Be + D12CN.
1960-cm-1 range for reactions of Be with H12CN with use of
an ablation energy near 30 mJ. This frequency range represents
the majority of the Be-H and CtN stretching regions, and the
presence of BeH, BeH2, and CN indicates that both Be-H and
CtN vibrational modes of products may be present in this
spectrum.7,32 The strongest absorption, labeled A, at 2088.7
cm-1 decreases by nearly half on photolysis. A band to the
red of this absorption, labeled C, increases a small amount on
photolysis and declines by a smaller amount on annealing; the
peak at 2153.2 cm-1, also labeled C, exhibits similar photolysis
and annealing behavior. Near the high-energy end of the
spectrum, a band labeled D increases 35% on photolysis and
decreases by a similar percentage on annealing. A weak band
also labeled D tracks with the 2202.8-cm-1 peak on photolysis
and annealing and suggests the possibility that these bands
represent different vibrational modes of the same product.
Another weak absorption labeled B at 2183.1 cm-1 broadens
and increases enormously on photolysis, then sharpens with a
moderate decrease in total intensity on 25 K annealing.
Be + H12CN. Table 1 lists all the observed frequencies as
well as the photolysis and 25 K annealing behavior for all
frequencies. Figure 1a presents the best spectrum in the 2220-
(25) Lanzisera, D. V.; Andrews, L. J. Phys. Chem. A 1997, 101, 824.
(26) Lanzisera, D. V.; Andrews, L. J. Phys. Chem. A 1997, 101, 1482.
(27) Bohn, R. B.; Andrews, L. J. Phys. Chem. 1989, 93, 3974.
(28) Gaussian 94, Revision B.1: Frisch, M. J.; Trucks, G. W.; Schlegel,
H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.;
Keith, T.; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.;
Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.;
Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng,
C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E.
S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.;
Baker, J.; Stewart, J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A.;
Gaussian, Inc.: Pittsburgh, PA, 1995.
(29) McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639.
(30) Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys.
1980, 72, 650.
(31) Kendall, R. A.; Dunning, T. H., Jr.; Harrison, R. J. J. Chem. Phys.
1992, 96, 6796.
The Be-C and Be-N regions of the spectrum from 950 to
740 cm-1 are presented in Figure 2a. The largest absorptions
occur in the higher energy region of this spectrum and are
labeled A and C. The A band declines 45% on photolysis, while
the 912.4-cm-1 peak increases a small amount. A weaker band
at 805.8 cm-1 increases tremendously on photolysis then
sharpens on annealing, thereby indicating that this band, labeled
(32) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007.
(33) Dunning, T. H., Jr.; Hay, P. J. In Modern Theoretical Chemistry;
Schaefer, H. G., III, Ed.; Plenum: New York, 1976; pp 1-28.
(34) Schlegel, H. B. J. Comput. Chem. 1982, 3, 214.
(35) Milligan, D. E.; Jacox, M. E. J. Chem. Phys. 1967, 47, 278.

6394 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Lanzisera and Andrews
In Figure 2b, identifying 12C/13C partners is fairly straigh-
forward. Both the A band and C band exhibit a 6.4 cm-1
isotopic carbon shift with respect to their natural isotopic
counterparts, while the lack of a HCN cluster band near 800
cm-1 makes for clearer observation of the product peaks in this
region of the spectrum. The B and D peaks shift 3.6 and 4.8
cm-1, respectively, upon carbon-13 substitution.
In the bending region (Figure 3b), the large cluster of peaks
is, with one exception, unshifted given the resolution of the
instrument. However, the blue shoulder at 530.1 cm-1 in Figure
3a shifts to 528.3 cm-1 in Figure 3b, indicating that this cluster
of peaks is due to at least two product molecules.
Be + DCN. Figure 1c presents the best Be + DCN product
spectrum for the CtN stretching region. Some of the peaks
are about half as large as in the HCN experiments, while others
are even weaker. Because beryllium can also react with the
D2O in the sample, the yield of products from the DCN reaction
was reduced.11 Bands in Figure 1c that are about half the
intensity of those in Figure 1a belong to products without
hydrogen. The weaker bands are those of residual products with
hydrogen. For example, the strong absorption at 2088.7 cm-1
belongs to a product without hydrogen, but the C band at 2153.2
cm-1 corresponds to a molecule with hydrogen. The strongest
new peak, at 2098.6 cm-1, appears to track with the C peaks of
Figure 1, parts a and b. At the high-energy end of the spectrum,
an absorption labeled D tracks with the 2202.8-cm-1 band from
Figure 1a and is about the same size as this band in the DCN
experiment.
Figure 2. Matrix infrared spectra in the 950-740-cm-1 Be-N and
Be-C stretching regions following pulsed laser ablation of Be atoms
codeposited with Ar/HCN (300/1) samples on a CsI window at 6-7
K: (a) Be + H12CN, (b) Be + H13CN, and (c) Be + D12CN.
In the Be-D stretching region, two new absorptions ac-
company BeD2 at 1674.2 cm-1. The first peak, at 1631.3 cm-1
,
tracks with the upper C bands. The other peak, at 1642.1 cm-1
,
increases more on photolysis and decreases more on annealing,
thereby tracking with the D bands of Figure 1, parts a and b.
In Figure 2c, the intensity of the 938.4-cm-1 absorption
indicates that this peak, like the other bands labeled A, possesses
no hydrogen atoms. By contrast, the C band at 912.4 cm-1 is
much smaller in this figure and has a tracking deuterated
counterpart at 895.6 cm-1. A similar contrast occurs near 800
cm-1, where the 805.8-cm-1 band does not shift, but the 800.1-
Figure 3. Matrix infrared spectra in the 540-400-cm-1 bending region
following pulsed laser ablation of Be atoms codeposited with Ar/HCN
(300/1) samples on a CsI window at 6-7 K: (a) Be + H12CN, (b) Be
+ H13CN, and (c) Be + D12CN. The cluster near 450 cm-1 is due to
DBeOD complexed with D2O from the exchange reaction (see ref 11).
cm-1 band shifts to 750.9 cm-1
.
B, tracks with the band labeled B in Figure 1a. The HCN cluster
peak near 800 cm-1 contains a weak product band at 800.1
cm-1, labeled D, which increases about a third on photolysis
and decreases about a third on 25 K annealing.
In Figure 3c, the cluster of peaks highlighted by a doublet at
525.4 and 521.7 cm-1 is much weaker, and two new clusters
appear, indicating that the original cluster consisted of two sets
of product bands which have different hydrogen isotopic shifts.
In the bending region (Figure 3a), the linear bend of HNC at
477.1 cm-1 can be observed, but the spectrum is dominated by
a cluster of peaks near 525 cm-1. These absorptions, at 530.1,
526.9, 525.4, 523.2, and 521.7 cm-1, all increase modestly upon
broadband photolysis and decrease on 25 K annealing.
Be + H13CN. Figure 1b presents the spectrum for the Be-H
and CtN stretching regions for products of the reaction of Be
with isotopically enriched H13CN. The largest absorption,
labeled A, decreases about 35% on photolysis and tracks
reasonably well with the A band in Figure 1a. Because of the
residual H12CN in this experiment, a trace of the 2088.7-cm-1
band remained. Notable is that there is no red shoulder to the
large band in Figure 1b as there was in Figure 1a. The next
largest absorption, labeled C, tracks with the 2153.2- and 2085.7-
cm-1 bands from the Be + H12CN experiment, and the small
isotopic carbon shift from the 2153.2-cm-1 peak suggests a
Be-H stretch. A weaker absorption at 2165.6 cm-1, labeled
D, tracks with the peaks labeled D in Figure 1a. Among the
weaker bands, a peak at 2134.6 cm-1 exhibits the same large
intensity increase upon photolysis as the 12C counterpart at
2183.1 cm-1, while a faint peak at 2116.7 cm-1 tracks with the
other absorptions labeled D.
The largest of these new doublets, at 422.1 and 418.0 cm-1
,
increases 20% on photolysis and decreases 10% on annealing,
while the doublet at 436.1 and 431.5 cm-1 increases 35% on
photolysis and is reduced by a similar amount on annealing.
The cluster of peaks near 450 cm-1 is due to DBeOD complexed
with D2O from the exchange reaction.11
Calculations. Table 2 lists the results for BP86 calculations
for four possible product molecules in the Be + HCN reaction.
Because beryllium is an alkaline earth element, only those
products with linear geometries are included. The spectra give
no evidence of products with two cyanide groups, so these
calculations were not performed.
Frequencies calculated by different methods for comparison
with experiment are presented in Table 3. Calculated frequen-
cies for BeCN and BeNC show that the BP86 approach
generally predicts lower vibrational energies than B3LYP
calculations. The MP2 level is better than the B3LYP approach
and only slightly worse than the BP86 method for BeNC, but
yields unreasonable frequencies for BeCN. With the BP86
method, use of the more inclusive basis sets, cc-pVDZ with
D95*, tends to generate slightly higher vibrational frequencies

Reactions of Laser-Ablated Be with HCN in Solid Ar
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6395
Table 2. DFT/BP86 Calculations for Products of Reactions of Be with HCN
b
species
energy (au)
bond lengths (Å)a
freq in cm-1 (int in km mol-1
)
HBeNC
HBeCN
BeNC
-108.20564
-108.20091
-107.55429
-107.54833
rHBe ) 1.33; rBeN ) 1.53; rCN ) 1.19
rHBe ) 1.33; rBeC ) 1.66; rCN ) 1.17
rBeN ) 1.54; rCN ) 1.19
2179.2 (305), 2066.9 (217), 918.9 (90), 565.0 (214 × 2), 172.1 (4 × 2)
2223.1 (167), 2147.4 (61), 799.5 (77), 554.7 (203 × 2), 232.5 (13 × 2)
2067.1 (368), 938.2 (129), 173.1 (0 × 2)
BeCN
rBeC ) 1.67; rCN ) 1.17
2199.3 (83), 806.6 (114), 243.6 (4 × 2)
a All products are linear molecules. b For the major stable isotope.
Table 3. DFT Calculations of Vibrational Frequencies and Intensities for BeCN and BeNC with Different Combinations of Functionals and
Basis Sets
method/basis set
BP86/6-311G*
BP86/cc-pVDza
B3LYP/6-311G*
MP2/6-311G*
BeCN
CtN stretchb
2199.3 (83) T 2149.5 (76)
1.02317
806.6 (114) T 804.1 (114)
1.00311
2208.4 (85) T 2158.1 (77)
1.02331
833.4 (132) T 831.0 (132)
1.00289
2291.0 (72) T 2239.3 (65)
1.02309
827.7 (120) T 825.1 (120)
1.00315
4090.4 (2036) T 3999.9 (1966)
1.02263
832.4 (134) T 829.4 (132)
1.00362
12C/13C ratioc
Be-C stretchb
12C/13C ratioc
Be-C-N bendb 243.6 (2 × 4) T 237.4 (2 × 4) 261.4 (2 × 2) T 254.7 (2 × 2) 260.7 (2 × 4) T 254.0 (2 × 3) 258.4 (2 × 6) T 251.7 (2 × 5)
12C/13C ratioc
1.02612
1.02631
1.02638
1.02662
BeNC
CtN stretchb
12C/13C ratioc
Be-N stretchb
12C/13C ratioc
2067.1 (368) T 2029.0 (367)
1.01878
938.2 (129) T 931.2 (124)
1.00752
2083.1 (317) T 2045.1 (318)
1.01858
944.0 (147) T 936.9 (142)
1.00758
2146.0 (396) T 2106.3 (395)
1.01885
969.6 (139) T 962.5 (134)
1.00738
2083.6 (283) T 2045.5 (285)
1.01863
947.0 (155) T 939.8 (150)
1.00873
Be-N-C bendb 173.1 (2 × 0) T 171.7 (2 × 0) 152.4 (2 × 1) T 151.1 (2 × 1) 186.8 (2 × 0) T 185.3 (2 × 0) 167.0 (2 × 0) T 165.5 (2 × 0)
12C/13C ratioc
1.00815
1.0086
1.00863
1.00906
a cc-pVDZ basis set on C and N atoms and D95* basis set on Be. b Vibrational frequencies in cm-1 and intensities in km mol-1 in parentheses
for 12C T 13C isotopes. c Ratio of 12C isotope frequency to 13C isotope frequency.
Table 4. Observed and BP86/6-311G* Calculated Vibrational
than with the 6-311G* basis sets. The ratios of the carbon-12
Frequencies (cm-1) for BeCN and BeNC
and carbon-13 isotopic frequencies, as a characteristic of the
normal modes, are effectively independent of the method used
for these molecules.
Be12CN
Be13CN
BeN12C
BeN13C
obsd
2183.1
2199.3
1.007
2134.6
2149.5
1.007
2088.7
2067.1
0.990
2050.8
2029.0
0.989
calcd
scale factora
Discussion
obsd
805.8
806.6
1.001
802.2
804.1
1.002
938.4
938.2
1.000
932.0
931.2
0.999
Four products have been identified in these experiments by
using isotopic data and BP86 calculations. Unlike with reactions
of B with HCN, no cyclic compounds with a Be(CN) ring
structure are observed, and this results from Be preferring linear
geometries and a coordination number of no more than 2.
Species A: BeNC. The strong absorptions at 2088.7 cm-1
in Figure 1, parts a and c, and 2050.8 cm-1 in Figure 1b
correspond to the CtN stretching motion of beryllium isocya-
nide, BeNC. The BP86 calculations predict this product to be
3.7 kcal mol-1 more stable than BeCN, while no stable cyclic
Be(CN) compound would converge. Table 3 lists the observed
frequencies for BeNC, as well as the results of these calculations.
The strong absorptions at 938.4 and 932.0 cm-1, for BeN12C
and BeN13C, respectively, are nearly identical with the predicted
values, while calculations of the CtN stretching frequencies
are still only about 1% lower than the observed values. The
calculated intensities for both these vibrational absorptions
(Table 2) are more than 100 km mol-1, with the CtN stretch
nearly three times more intense than the Be-N stretch. For
both peaks, the intensities for the DCN experiments are
competitive with those for the HCN experiments, suggesting
an absence of hydrogen in the molecule.
calcd
scale factora
calcd
243.6
237.4
173.1
171.7
a Calculated frequency divided by observed frequency.
the CtN stretching region retains a large fraction of its intensity
in the DCN experiments. According to Table 2, the Be-C
stretch should be the strongest absorption for this molecule and
occur near 800 cm-1. The absorptions observed at 805.8 and
802.2 cm-1 are also strong in the DCN experiments and are
nearly identical with the predicted values. Table 4 also presents
observed and calculated frequencies for BeCN. The photolysis
and annealing behavior of these absorptions are unique in this
experiment and are an identifying characteristic of BeCN peaks.
The CtN stretching modes that mimic this behavior are at
2183.1 and 2134.6 cm-1 for Be12CN and Be13CN, respectively.
As with the Be-C stretching mode, these computationally
inexpensive DFT calculations predict vibrational frequencies
within 1% of their experimental values. These absorptions are
not among the most intense in Figure 1, but the BP86
calculations predict them to be less than one quarter as intense
as the BeNC bands given equal concentrations of both products.
Compared to boron isocyanide,24 beryllium isocyanide has a
CtN stretch about 20-30 cm-1 higher in energy. This value
is much lower than the 2171-cm-1 value for BCN and confirms
the assignment of these strong bands to BeNC. Also, the B-N
stretch in BNC lies higher than the B-C stretch of BCN, and
the presence of strong bands at 938.4 and 932.0 cm-1 in the
high-energy region of Figure 2 suggests that these absorptions
correspond to Be-N stretching vibrations.
On the basis of a comparison with BCN and BNC, one
expects the CtN stretching frequency of BeCN to be about
100 cm-1 higher in energy, and that is the case in these
experiments. For the isoelectronic BeCCH molecule,12 the
Be-C stretching motion absorbs at 855.0 cm-1 when both
carbons are 12C. This frequency changes to 851.8 cm-1 when
the first (i.e., bonded to Be) carbon is changed to 13C. The
small carbon isotopic shift in BeCCH and BeCN, especially
relative to BeNC, suggests a mode that behaves like a “sym-
Species B: BeCN. Although BeCN is calculated to be of
nearly the same energy as BeNC, no other strong absorption in

6396 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Lanzisera and Andrews
Table 5. Experimental, BP86/6-311G*, and CI/6-311G* Vibrational Frequencies (cm-1) for HBeCN and HBeNC
HBe12CN
HBe13CN
DBe12CN
HBeN12C
HBeN13C
DBeN12C
obsd
calcd-BP86
calcd-CI
2202.8
2223.1
2350.4
2165.6
2187.5
2301.2
2194.2
2212.5
2345.6
2153.2
2179.2
2258.7
2149.5
2177.7
2253.8
1631.3
1646.4
1702.9
obsd
calcd-BP86
calcd-CI
2128.6
2147.4
2223.9
2116.8
2132.7
2220.0
1642.1
1627.8
1680.5
2085.7
2066.9
2190.5
2050.9
2030.3
2154.3
2098.6
2078.4
2207.6
obsd
calcd-BP86
calcd-CI
800.1
799.5
820.9
795.3
796.8
818.0
750.9
760.0
780.8
912.4
918.9
953.1
906.0
911.7
945.9
895.6
867.4
899.6
obsd
calcd-BP86
calcd-CI
530.1
554.7
579.7
528.3
552.7
577.5
436.1/431.5
464.8
485.8
525.4/521.7
565.0
599.1
525.4/521.7
565.0
599.0
422.1/418.0
464.1
491.3
metric” stretching mode with little carbon motion. A similar
situation exists when comparing BCN and BNC.24
also be detected. Because the C bands increase in intensity
and because the BP86 calculations predict the frequency of
HBeN13C to be near this absorption, this A band must consist
of two components: a strong BeNC absorption and a weaker
HBeNC absorption. Given the resolution of the instrument, it
is impossible to determine the exact frequency of HBeN13C,
but deconvolution of the spectra following photolysis gives a
HBeNC frequency of 2050.9 cm-1 versus 2050.8 cm-1 for
BeNC.
HBeNC has not only a CtN stretch of nearly the same
frequency as that of BeNC but also a Be-N stretch close in
energy to that of BeNC. The C band in Figure 2a is much less
intense in Figure 2c and tracks with the C bands of Figure 1.
These peaks, then, are assigned to HBeNC. The BP86 calcula-
tions listed in Table 5 are in excellent agreement with experi-
ment. For DBeNC, the band in Figure 2c at 895.6 cm-1 clearly
tracks with the 912.4-cm-1 band of residual HBeNC. Although
the calculations predict a larger deuterium isotopic shift, the
experimental evidence indicates that deuteration shifts the Be-N
stretching mode a relatively small amount. Because this isotopic
shift is so small and the frequency shift of this mode from that
of BeNC is also quite small, hydrogen does not appear to affect
this vibrational mode of the isocyanide molecules.
Species C: HBeNC. The strongest band with HCN (Figure
1a) that is weaker in DCN experiments (Figure 1c) is at 2153.2
cm-1 with a 13C counterpart only 3.7 cm-1 to the red. The
small carbon isotopic shift and the drop in intensity in a partially
deuterated sample suggest a Be-H stretching mode. Because
no calculation of a cyclic product molecule converged with a
potential minimum, the only possible products for this band are
HBeCN and HBeNC. Table 5 gives the BP86 and CISD
calculated frequencies for isotopic product molecules. For
HBeCN, the Be-H stretch is calculated by the BP86 method
to shift 14.7 cm-1 due to mixing with the CtN stretch, but
HBeNC is calculated to shift only 1.5 cm-1, in much better
agreement with experiment. Also, the calculated intensities for
the Be-H stretches are 305 and 275 km mol-1 for HBeN12C
and HBeN13C and 61 and 9 km mol-1 for HBe12CN and HBe13-
CN, further suggesting that the 2153.2/2149.5-cm-1 bands are
due to the Be-H stretching absorptions of HBeNC. The
calculations predict the corresponding DBeNC absorption at
1646.4 cm-1. The observed 1631.3 cm-1 band agrees well with
this calculation and tracks with the other Be-H absorptions on
photolysis and annealing.
In Figure 1a, the BeNC peak at 2088.7 cm-1 has a shoulder
band labeled C. The corresponding BeN13C absorption in
Figure 1b has no such shoulder. Also, the intensity of this band
is much reduced in Figure 1c, suggesting not only that this peak
corresponds to a product distinct from BeNC, but also that this
species contains hydrogen. Because frequencies in this region
typically are due to CtN stretches of isocyanide complexes,
such as BeNC, this band must arise from HBeNC. According
The final observable mode for HBeNC is the doubly
degenerate linear hydrogen bending mode predicted at 565.0
cm-1 for both 12C and 13C with a hydrogen isotopic shift of
approximately 100 cm-1. The doublet at 525.4 and 521.7 cm-1
among a cluster of peaks does not shift with carbon isotopic
substitution and is assigned to this mode with the splitting
possibly due to degeneracy breaking by the matrix environment.
In Figure 3c, the stronger doublet at 422.1 and 418.0 cm-1 tracks
with the HBeNC bands and exhibits a reasonable deuterium
isotopic shift based on the calculations. Although the BP86
calculations are less accurate for these bending motions than
for stretching vibrations, they are still nearly always within 10%
of the experimental values.
Table 5 also reports CISD calculations with the 6-311G* basis
set on HBeNC and HBeCN. Note that the isotopic ratios are
similar to those of the BP86 method, but further displaced from
the observed values. Although CISD calculations are more
sophisticated than BP86 calculations and require much greater
computation time, the BP86 calculations better enable identi-
ficaiton of products in these experiments.
to Table 5, BP86 calculations predict this mode at 2066.9 cm-1
,
in excellent agreement with experiment. For DBeNC, this
calculated frequency shifts to the blue at 2078.4 cm-1. For
HBeNC, mixing between the Be-H and CtN stretching modes
forces the CtN frequency to the red, but such mixing is absent
in DBeNC. The peak in Figure 1c which best matches the
calculation is at 2098.6 cm-1, a band which tracks with the
HBeNC absorption.
For HBeN13C, there is no obvious CtN stretching absorption,
but one expects an isotopic shift similar to that for the other
isocyanide product, BeNC. The BP86 calculations predict this
frequency to be 2030.3 cm-1
. Because the other isotopic
calculations are approximately 20 cm-1 too low, the CtN
stretch of HBeN13C should occur at about 2050 cm-1, near the
BeN13C peak, but no separate intense absorption can be observed
in the spectrum. Upon photolysis, the A band of Figure 1b,
which is noticeably larger than the A band of Figure 1a prior
to photolysis, decreases in intensity, but to a lesser degree than
for the 12C experiments. A slight shift in the band profile can
Species D: HBeCN. The absorptions labeled D in Figure 1
also belong to a product with hydrogen. Although the frequen-
cies of these peaks are consistent with both Be-H and CtN
stretches, the large carbon isotopic shift and relatively small
hydrogen isotopic shift indicate that these are CtN stretches.
These frequencies are also more consistent with a cyanide
product than with an isocyanide product. Calculations on

Reactions of Laser-Ablated Be with HCN in Solid Ar
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6397
HBeCN (Table 5) give a CtN stretching frequency ap-
proximately 20 cm-1 higher than the observed values for each
of the isotopomers, in excellent agreement with experiment.
Calculated intensities of these absorptions are 167, 213, and
86 km mol-1 for HBe12CN, HBe13CN, and DBe12CN, respec-
tively, suggesting mixing with the nearby Be-H stretching
mode. Because all of the bands labeled D track with each other
and because the Be-H stretching frequency predictions are only
1% higher than the frequencies at 2128.6 and 2116.7 cm-1 for
12C and 13C, these absorptions must be due to the Be-H stretch
of HBeCN. The calculated intensities for these peaks are 61
and 9 km mol-1 for 12C and 13C, suggesting preferential mode
mixing for one of the isotopes. Since the HBe13CN bands are
much closer in energy, the mode mixing must be greater in
HBe13CN than in HBe12CN. The CtN stretching intensity of
HBe13CN is greatest and the Be-H stretch intensity is lowest
because mode mixing has transferred intensity from the latter
to the former. Similarly for DBeCN, the CtN stretch intensity
is lowest because there is no mode mixing. The Be-D
and be trapped in the argon matrix or form other products. For
example, BeCN can form via H atom elimination, HBeNC via
rearrangement, and BeNC via rearrangement and H atom
elimination.
The most intense absorptions in the spectra of the Be + HCN
reaction products correspond to the radical BeNC. Although
the calculated intensity of this absorption is also the strongest
among all the products observed, this band is too intense for
BeNC to be anything other than a major product, if not the
most abundant new product. The insertion mechanism probably
yields more BeCN than BeNC because of a more direct path to
product formation, but far more BeNC is observed. This
suggests a possible alternate mechanism similar to a SN2
reaction:
Be* + NCH f [BeNC]† + H f BeNC + H
Another possibility is Be bonding to a CN fragment formed
from the radiation in the ablation process, but the “SN2”
mechanism would yield preferentially more BeNC compared
to BeCN. Such disproportionation in the formation of BeNC
might explain why these peaks are reduced so drastically on
photolysis while those of BeCN increase tremendouslysthe
photolysis initiates an equilibration process between the two
products to bring their relative concentrations closer. The
ultraviolet radiation apparently has sufficient energy to overcome
the barrier to interconversion:
vibration, predicted at 1627.8 cm-1, appears at 1642.1 cm-1
,
with again approximately 1% difference between experiment
and theory. For this peak, the lack of mode mixing makes it
more intense than its HBeCN counterparts.
For the mixed CtN and Be-H stretching modes in HBeCN,
the carbon 12/13 ratios (observed, BP86, CISD) are (1.0172,
1.0163, 1.0214) and (1.0055, 1.0069, 1.0018), respectively.
Clearly, the BP86 calculation more accurately reflects the
observed mode mixing.
Like BeCN, HBeCN has a Be-C stretching vibration near
800 cm-1. As with HBeNC and BeNC, this Be-C mode shifts
a small amount upon addition of hydrogen to the beryllium atom.
As shown in Table 5, calculations predict these frequencies
extremely well. Unlike in HBeNC, the hydrogen isotopic shift
in HBeCN is significant and agrees very well with the BP86
predictions. For the three stretching modes in HBeCN, the BP86
calculated frequencies deviate no more than 1.2% from experi-
ment for any isotopic combination.
UV
BeNC y z BeCN
The high yield of HBeNC relative to HBeCN may result from
rapid rearrangement of the insertion product to the more stable
isomer, but another “SN2” reaction may be the reason for the
relatively high yield. First, Be abstracts an H atom to form
BeH, which is also observed in the product spectrum, then the
following reaction can take place:
[HBe]† + NCH f [HBeNC]† + H f HBeNC + H
The linear bending mode for HBeCN is predicted to appear
near that of HBeNC, but with a small carbon isotopic shift.
With carbon three atoms away from the hydrogen, as in HBeNC,
isotopic substitution does not noticeably affect the frequency,
but with carbon only two atoms away, as in HBeCN, a small
shift should be observed. On the blue shoulder of the cluster
of peaks comprising the HBeNC bands, a peak shifts from 530.1
to 528.3 cm-1 upon 13C substitution. The presence of a second
doublet other than one due to HBeNC in the DCN experiments,
which tracks with the bands labeled D, indicates that this second
linear bend is due to HBeCN. As with HBeNC, the calculations
do not predict these frequencies as well as for the stretching
modes, but are still quite good nonetheless. Because the
degeneracy of this linear H bend is broken in DBeCN, it is
likely that the degeneracy is also broken for HBeCN, but the
HBeNC bands prevent the observation of the other member of
this doublet to the red of the observed peak.
Comparison of Frequency Calculations. The apparent
superiority of the BP86 functional over more sophisticated
methods can be explained by recognizing that all of these
methods calculate harmonic vibrational frequencies. Anhar-
monicity in the products lowers the experimental vibrational
frequencies and is the reason the B3LYP, MP2, and CISD
methods generally overestimate the observed frequencies. In
a comparison of various Hartree-Fock and DFT methods, Scott
and Radom showed that the BP86 calculations tend to overes-
timate bond lengths, thereby underestimating harmonic stretch-
ing frequencies.36 This shortcoming of the BP86 method
actually enables these calculations to predict vibrational fre-
quencies of the beryllium-hydrogen cyanide products with
excellent precision. That this result is fortuitous should be
noted.
Nevertheless, HBeNC and especially HBeCN required more
precise calculations than usual because of mixing between the
Be-H and CtN stretching modes. Whereas the BP86 calcula-
tions with 6-311G* basis sets were accurate in predicting the
isotopic shifts observed in the experiments, the other methods
did not fare as well because of this mode mixing. With use of
the BP86 functional with the cc-pVDZ plus D95* basis sets,
the predicted carbon-13 shifts of the Be-H and CtN stretches
in HBeCN are 20.3 and 30.2 cm-1, which is somewhat less
accurate than the values determined with the 6-311G* basis sets.
As Table 3 shows, the difference when using different basis
Reaction Mechanisms. Given the relative simplicity of the
Be/HCN system and the lack of observed dicyanide complexes,
the major pathway to product formation is via insertion:
Be* + HCN f [HBeCN]†
The growth of some products on photolysis and the general
decrease of product absorptions on annealing suggest that
energetic Be atoms are required for the insertion reaction.
3
Recent studies have provided convincing evidence that P Be
atoms from the ablation process are the reactive species in these
experiments.7-12 The excited HBeCN molecule can either relax
(36) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.

6398 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Lanzisera and Andrews
sets is relatively small for BeCN and BeNC, and the same can
be said for HBeCN and HBeNC. With the B3LYP method,
however, the differences are more pronounced, with carbon-13
shifts of 7.6 and 44.5 cm-1 predicted by this method. Although
the B3LYP, MP2, and CISD methods are probably superior to
the BP86 method for calculating geometries and harmonic
frequencies, the BP86 functional provides clear advantages for
identifying new reaction products from their experimental (i.e.,
anharmonic) frequencies.
as mode mixing in HBeCN and the coincidence of absorptions
in BeN13C and HBeN13C, the BP86 calculations with 6-311G*
basis sets have enabled the identification of all four major
products with confidence. Of particular importance is the
comparison of isotopic frequency shifts (or ratios) as a descrip-
tion of normal modes37 to determine the degree of agreement
between theory and experiment. The BP86 calculations with a
more inclusive basis set, cc-pVDZ plus D95*, and the more
sophisticated B3LYP and MP2 calculations are inferior at
predicting the product vibrational frequencies. Although all
methods predict harmonic frequencies, the BP86 calculations
tend to overestimate bond lengths in the geometry optimization,
leading to lower energy predictions for stretching vibrations.
These predictions are, therefore, closer in energy to the observed
(i.e., anharmonic) values. The CISD method succeeds at
predicting vibrational frequencies nearly as well as the BP86
method, but the increase in computation time makes using this
method impractical for the purposes of this study.
Conclusions
The major products of the reaction of laser ablated beryllium
atoms with hydrogen cyanide are BeNC, BeCN, HBeNC, and
HBeCN. No products with a Be(CN) ring are observed.
Because of the low concentration of HCN in the Ar mixture,
no products with two or more CN groups were detected. The
CtN and Be-N stretching frequencies are similar for both
BeNC and HBeNC, indicating that the presence of the hydrogen
atom has little effect on these modes. The same is true for the
Be-C stretches of HBeCN and BeCN, but not for the CtN
stretch. In HBeCN, mixing between the nearly isoenergetic
Be-H and CtN stretching modes tends to push the CtN
vibrational frequencies higher and the Be-H frequencies lower.
The effect is most pronounced for HBe13CN, leading to an
unusually large 11.8 cm-1 carbon isotopic shift for the Be-H
stretch and a loss of intensity in this mode relative to that of
HBe12CN.
The density functional calculations show that the BeNC,
BeCN, HBeNC, and HBeCN molecules are linear. These
compounds further illustrate the role of beryllium bonding in
small linear molecules and complement earlier studies with
HBeH (the textbook example of sp hybridization), HBeCCH,
BeCCH, and HBeOBeH.7,11,12
Acknowledgment. This work was supported by the Air
Force Office of Scientific Research. Calculations were per-
formed on the University of Virginia SP2 machine. K13CN was
kindly donated by Cambridge Isotope Laboratories.
For the products observed in these experiments, BP86
calculations have provided an excellent guide to identification
of the product bands. For most of the absorptions, these
calculations predict vibrational frequencies within 1% of the
experimental values. Despite complexities in the spectra, such
JA970740G
(37) Martin, J. M. L.; Taylor, P. R.; Hassanzadeh, P.; Andrews, L. J.
Am. Chem. Soc. 1993, 115, 2510.