Electron spin resonance matrix isolation studies of 27Al16,17O, 69,71Ga16,17O and 115In16,17O: Observed hyperfine interactions compared with ab initio theoretical results

Article abstract of DOI:10.1063/1.475164

Electron spin resonance (ESR) studies are reported for Al^16,17O, Ga^16,17O, and In^16,17O isolated in neon matrices at 4 K. Except for Al¹⁶O, no previous ESR measurements have been reported for these X ²Σ diatomic radicals. The pulsed laser vaporization of the metals in the presence of ¹⁶O₂ and ¹⁷O₂ produced high quality ESR spectra of these metal oxide radicals whose nuclear hyperfine interactions (A tensors) were fully resolved for both the metal and oxygen nuclei. An analysis of the experimental spin densities in combination with different types of theoretical calculations provided detailed information concerning the electronic structure trends going down this metal oxide group. Increased p-orbital spin density on oxygen was observed for the heavier metal oxide radicals. Nonrelativistic ab initio calculations with an extended basis set and the UB3LYP method reproduced the trends in the isotropic and dipolar hyperfine interactions. All-electron CI calculations, restricted open-shell Hartree-Fock (ROHF) wave functions, and unrestricted Hartree-Fock wave functions gave results very different from experiment and from each other for the isotropic interaction. All calculations were in fair agreement with each other for the dipolar interaction and provided an assignment of the sign for that term.

Full text of DOI:10.1063/1.475164

Electron spin resonance matrix isolation studies of 27 Al 16,17 O , 69,71 Ga 16,17 O and
115 In 16,17 O : Observed hyperfine interactions compared with ab initio theoretical
results
Lon B. Knight Jr., Thomas J. Kirk, John Herlong, John G. Kaup, and E. R. Davidson
Citation: The Journal of Chemical Physics 107, 7011 (1997); doi: 10.1063/1.475164
View online: http://dx.doi.org/10.1063/1.475164
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/107/18?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Electron spin resonance g tensors for complexes of Ne and Ar with AlO:Theoretical studies related to the large
matrix effect observed for AlO
J. Chem. Phys. 122, 124504 (2005); 10.1063/1.1866095
Matrix-isolation investigation of the diatomic anion radicals of aluminum and gallium ( Al 2 − and Ga 2 − ): An
electron spin resonance and ab initio theoretical study
J. Chem. Phys. 115, 4632 (2001); 10.1063/1.1389841
Ab initio and molecular-dynamics studies on rare gas hydrides: Potential-energy curves, isotropic hyperfine
properties, and matrix cage trapping of atomic hydrogen
J. Chem. Phys. 110, 11814 (1999); 10.1063/1.479173
Electron spin resonance studies of 45 Sc 17 O , 89 Y 17 O , and 139 La 17 O in rare gas matrices: Comparison
with ab initio electronic structure and nuclear hyperfine calculations
J. Chem. Phys. 110, 5658 (1999); 10.1063/1.478464
Neon and argon matrix ESR and theoretical studies of the 12 CH 3 Cd , 12 CD 3 Cd , 13 CH 3 Cd , 12 CH 3 111
Cd , and 12 CH 3 113 Cd radicals
J. Chem. Phys. 110, 3398 (1999); 10.1063/1.478206
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

27
Electron spin resonance matrix isolation studies of Al16,17O,
69,71Ga16,17O and 115In16,17O: Observed hyperfine interactions compared
with ab initio theoretical results
Lon B. Knight, Jr., Thomas J. Kirk, John Herlong, and John G. Kaup
Chemistry Department, Furman University, Greenville, South Carolina 29613
E. R. Davidson
Chemistry Department, Indiana University, Bloomington, Indiana 47405
͑Received 10 June 1997; accepted 6 August 1997͒
Electron spin resonance ͑ESR͒ studies are reported for Al16,17O, Ga16,17O, and In16,17O isolated in
neon matrices at 4 K. Except for Al16O, no previous ESR measurements have been reported for
2
these X ⌺ diatomic radicals. The pulsed laser vaporization of the metals in the presence of 16O2 and
17O2 produced high quality ESR spectra of these metal oxide radicals whose nuclear hyperfine
interactions ͑A tensors͒ were fully resolved for both the metal and oxygen nuclei. An analysis of the
experimental spin densities in combination with different types of theoretical calculations provided
detailed information concerning the electronic structure trends going down this metal oxide group.
Increased p-orbital spin density on oxygen was observed for the heavier metal oxide radicals.
Nonrelativistic ab initio calculations with an extended basis set and the UB3LYP method
reproduced the trends in the isotropic and dipolar hyperfine interactions. All-electron CI
calculations, restricted open-shell Hartree–Fock ͑ROHF͒ wave functions, and unrestricted Hartree–
Fock wave functions gave results very different from experiment and from each other for the
isotropic interaction. All calculations were in fair agreement with each other for the dipolar
interaction and provided an assignment of the sign for that term. © 1997 American Institute of
Physics. ͓S0021-9606͑97͒01942-9͔
I. INTRODUCTION
lighter MO radicals but it was more problematic for the
heavier GaO and InO molecules. However, the isotopic
(Aiso) and dipolar (Adip) components of the hyperfine inter-
actions reveal interesting qualitative trends going down this
MO series. In fact, this set of MO radical measurements
͑BO, AlO, GaO, and InO͒ represents the most complete ver-
tical profile of spin density results reported to date.
Understanding the electronic structures of ‘‘simple’’
metal oxides is important for several reasons, especially
given their involvement in superconducting materials and
thin film insulators. For this vertical series of increasingly
2
complex and isovalent X ⌺ metal oxide diatomic radicals,
the nuclear hyperfine interactions reveal interesting trends
for the valence electronic structure region. It is especially
significant to develop computational methods which can be
tested directly on such prototypical species, thus increasing
the reliability of theoretical predictions for the more complex
materials for which direct hyperfine and other measurements
cannot readily be made. Metal oxide ͑MO͒ formation is often
a critical factor in chemical vapor deposition ͑CVD͒ pro-
cesses which are used to produce a wide variety of coatings
for microelectronic devices.
This study reports and analyzes the first electron spin
resonance ͑ESR͒ measurements of any kind for gallium ox-
ide ͑GaO͒ and indium oxide ͑InO͒. Also, the first 17O (I
ϭ5/2) hyperfine structure ͑hfs͒ measurements for aluminum
oxide (Al17O) are reported, although the 27Al hfs for 27Al16O
has been fully resolved in previous matrix studies and in
high resolution gas phase experiments.1,2 Matrix and gas
phase studies have also been reported for the lightest mem-
ber of this series, namely the boron oxide ͑BO͒ molecule.3,4
A complete resolution of the hyperfine interactions ͑A ten-
sors͒ for the metal atoms and 17O in this MO series was
possible. As might be expected, the free atom comparison
method ͑FACM͒ seemed to be a reasonable approach for the
Comparisons are made between the experimental Aiso
and Adip values and those calculated from unrestricted
Hartree–Fock ͑UHF͒ and density functional theory ͑DFT͒
methods. The unusually large divergence in the A values
predicted by these two theoretical methods is discussed in
terms of special electronic structure features of these metal
oxide radicals. It is most difficult to properly balance contri-
butions to the overall electronic structure from the two ionic
extremes of MϩOϪ ͑unpaired electron formally resides on
oxygen͒ and MϩϩOϪϪ ͑where it resides on the metal atom͒.
Previous dipole moment calculations have encountered a
similar problem for the AlO radical.5 This treatment found a
gradual transition from AlϩOϪ to AlϩϩOϪϪ in the internu-
clear distance range of 3 to 4 a.u. The observed nuclear hy-
perfine properties of these MO radicals is extremely sensitive
to the balance of these two different ionic descriptions, thus
making ESR measurements especially appropriate. The re-
sults of this study should provide a better understanding of
ionic metal oxide bonding, especially since only a few cases
of this type have been studied previously and none have had
their 17O hyperfine properties investigated. Also in several
cases where metal oxides have been observed, for example
ScO and YO,6 the metal hyperfine interactions were not re-
J. Chem. Phys. 107 (18), 8 November 1997
0021-9606/97/107(18)/7011/9/$10.00
© 1997 American Institute of Physics
7011
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

7012
Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
solved into parallel and perpendicular components, thus pre-
venting a direct determination of the separate ‘‘ns’’ and
‘‘np’’ or ‘‘nd’’ orbital contributions.
ent experimental procedures. One procedure involved the di-
rect passage of pure O2(g) over the metal surface ͑Al, Ga, or
In͒ which was undergoing pulsed laser vaporization. ͑See
Fig. 1 of Ref. 28.͒ The second harmonic ͑532 nm͒ of the
Nd:YAG pump laser was focused to a spot size of approxi-
mately 0.3 mm. The focused beam was slowly rastered over
the surface by means of an external focusing lens. The va-
porization laser was operated at 10 Hz and pulse energies of
approximately 15 mJ for aluminum, 10 mJ for gallium, and 4
mJ for indium. The high purity metal targets were obtained
from Alfa. In this arrangement, the laser ablated reaction
products encountered the condensing rare gas in the vicinity
of the copper matrix surface which was located 6 cm from
the laser vaporization target. A typical rare gas flow rate of 5
sccm ͑std. cm3 minϪ1͒ was employed. This flow rate for
neon caused the hot-filament type ionization gauge ͑cali-
brated for N2͒ in the matrix deposition vicinity to indicate a
pressure of approximately 5ϫ10Ϫ5 Torr.
A considerable number of gas phase electronic transi-
tions have been analyzed for this group IIIA diatomic metal
oxide series.7 In more recent experiments, the Al2O molecule
has been prepared by the oxidation of laser vaporized alumi-
num and studied under supersonic beam conditions.8 Ther-
mochemical studies of AlO and AlOϩ with various reactants
have also been made in the gas phase,9–12 as well as reactive
quenching of electronically excited AlO.13 Properties of AlO
are especially interesting given its presence in a wide range
of high temperature environments from rocket exhausts and
explosives to upper atmospheric monitoring and astrophysi-
cal and geophysical applications. ͑See Refs. 8–10 and 14–
17.͒ In earlier experiments, laser ablated Al atoms and AlO
from Al2O3 surfaces were used to study the laser ablation
mechanism.18
It is interesting that recent vibrational studies of reac-
tions between laser ablated gallium and indium metals with
O2 in argon matrices did not detect the direct presence of the
monoxide radicals, however GaO2 and InO2 were clearly
observed.19 The vibrational matrix studies were not con-
ducted in neon matrices. These ESR matrix experiments are
consistent with the earlier vibrational studies in that we ob-
served intense GaO and InO absorptions in neon but ex-
tremely weak signals for InO and none for GaO in argon
matrices. Aluminum and boron reactions under such argon
conditions did produce AlO, AlO2, BO, and BO2. Vibra-
tional matrix studies of a wide variety of MxOy species have
recently been summarized by Andrews and co-workers, in-
cluding the earlier matrix work which employed thermal
sources of the metal atom in codeposition reactions with
O2.19,20,21–24
A wide variety of neutral and charged radicals can be
studied in rare gas matrices by ESR. Weltner’s review of the
ESR matrix method includes a listing of the diatomic metal
oxides studied in earlier years—RbO, CsO, BO, AlO, ScO,
YO, LaO, VO, NbO, MnO, and GdO.25 Dioxides studied by
this method include MnO2,25 PO2, AsO2,26 and VO2.27 Re-
cently studied small radicals relevant to this AlO, GaO, and
InO series include AlHϩ,28 AlFϩ,29 AlC,30 AlH2,31 Al3,32
HAlCH3,33 Al͑CO͒2,34 GaAsϩ,35 Ga3,36 GaH2,37 Ga2As3,38
and Ga͑C2H4͒.39 Only a small number of gallium centered
radicals have been investigated37 and apparently no indium
radical ESR studies have been reported, although GaH2 and
InH2 have been investigated by vibrational spectroscopy in
rare gas hosts.40
In the direct reaction mode described above ͑O2 passed
directly over the metal undergoing laser ablation͒, the
amount of O2 introduced was varied from approximately 1%
to 0.01% of the rare gas amount in a series of different depo-
sitions, each lasting for 30 min. This same concentration
range was also employed in a series of indirect reaction
deposition experiments. In the indirect scheme, the O2 was
premixed with the rare gas—hence, no O2 gas was passed
directly over the metal vaporization surface. While there is
not complete separation, most metal atom–O2 encounters oc-
curred during the matrix condensation process in this indirect
experimental reaction arrangement.
For both reaction methods, background pressures were
typically 5ϫ10Ϫ7 Torr with water being the predominant
contaminant. Background gases and isotopically enriched
17O2 samples, obtained from Isotec, were monitored with a
quadrupole mass spectrometer ͑Stanford Research System
RGA200͒ which is built into the vacuum system of the ma-
trix isolation cryostat.
III. RESULTS: ESR ANALYSIS
The analysis of the observed ESR spectra for AlO, GaO,
and InO was straightforward. However, the large magnitude
of the metal hyperfine interactions required an exact diago-
nalization of the determinants resulting from the following
spin Hamiltonian:
ˆ
ជ
ជ
ជ ˆ ជ
ˆ
Hϭ␤H•g•SϩS•A•I,
where all symbols have their standard meanings.25 Quadru-
pole interaction was negligible since this effect decreases as
A, the hyperfine tensor, increases. For the allowed ⌬MS
ϭϩ1 and ⌬MIϭO transitions, the nuclear Zeeman interac-
tion can also be neglected. The magnetic energy levels re-
sulting from this exact diagonalization treatment at all angles
are coupled with a transition probability routine and a
Lorentzian lineshape program for simulating the first deriva-
tive ESR spectra. These combined programs, written in our
II. EXPERIMENT
The rare gas matrix apparatus used in our laboratory
͑Furman University͒ for X-band ESR studies of small radi-
cals has been described previously.26–28,41 Both liquid helium
and closed cycle helium refrigerators are used in various ex-
perimental arrangements to cool the copper matrix deposi-
tion surface to 4 to 5 K. Matrix depositions under reactive
laser vaporization conditions were conducted by two differ-
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
7013
TABLE I. aObserved ESR line positions ͑Gauss͒ for 27Al16O in neon matrix
at 4 K.
27Al:Iϭ5/2
Perpendicular (␪ϭ90°)
Parallel (␪ϭ0°)
MIϭ5/2
3/2
2750
2956
3188
3443
3722
4026
2614
2886
3178
3490
3820
4171
1/2
Ϫ1/2
Ϫ3/2
Ϫ5/2
aMicrowave frequency is 9570.9͑4͒ MHz. Calculated line positions agree
with these observed lines within the experimental uncertainty of Ϯ1 G.
Additional 17O(Iϭ5/2) hfs for AlO was also observed. See magnetic pa-
rameters listed in Table III and spectra in Fig. 1.
69Ga16O and 71Ga16O in Fig. 2 reflects these isotopic charac-
teristics, with each isotopomer exhibiting a quartet of parallel
and perpendicular components. As shown in the spectrum,
the weaker 71Ga16O lines occur at higher magnetic fields for
the high field MIϭϪ3/2 transitions and at lower fields for
the MIϭϩ3/2 transitions relative to 69Ga16O. All of these
transitions exhibited the same, well resolved 17O(Iϭ5/2)
sextet hyperfine structure as shown in the highly expanded
spectra of Fig. 3 for the highest field Ga transitions (MI
ϭϪ3/2). The lower spectrum in Fig. 3 shows the perpen-
dicular and much weaker parallel lines for 69Ga16O and
71Ga16O. Centered about each of these components are the
17O sextets shown in the top spectrum of this figure for
69Ga17O and 71Ga17O. Since the 17O enrichment is approxi-
mately 50%, the central Ga16O lines also appear in this top
spectrum of Fig. 3. Note the amplification factor (X10) in-
creases for the highest field, weakest features, namely, the
parallel lines of 71Ga17O. The weak parallel lines of 69Ga17O
were detected but several of these were partially obscured by
the more intense 71Ga17O perpendicular components.
FIG. 1. Sextet 17O(Iϭ5/2) hyperfine structure is shown in the neon matrix
ESR spectra of the lowest (MIϭ5/2) and highest (MIϭϪ5/2) field parallel
(⌰ϭ0°) and perpendicular (⌰ϭ90°) transitions of Al(Iϭ5/2) for the
27Al17O radical in its X 2⌺ ground electronic state. The central region of
each 17O sextet pattern exhibits a single line resulting from 27Al16O which
has no oxygen splitting. Line positions for AlO are listed in Table I.
laboratory and described previously,41 were used to obtain
all the calculated line positions and simulated ESR spectra
presented in this article.
All spectra exhibited random orientation in the rare gas
lattice and axial symmetry. Only two molecular g compo-
nents are involved and for each magnetic nucleus there are
only two A tensor components, referred to as Aʈ and AЌ .
These molecular magnetic parameters were obtained by fit-
ting the observed and calculated line positions. As discussed
below, the ESR spectra for 115InO was especially challenging
since ‘‘extra’’ or ‘‘off-angle’’ absorptions were observed
whose intensities for some transitions were comparable to
the ⌰ϭ0° and ⌰ϭ90° limits of the various hyperfine tran-
sitions. Theta ‘‘⌰’’ is defined as the angle between the mo-
lecular axis of these diatomic radicals and the externally ap-
plied magnetic field.
A. 27Al16,17
O
The neon matrix ESR spectra for 27Al16O consisted of
aluminum (Iϭ5/2) sextets of parallel (⌰ϭ0°) and perpen-
dicular (⌰ϭ90°) absorptions with linewidths of approxi-
mately 1 to 2 Gauss. The lowest (MIϭ5/2) and highest
(MIϭϪ5/2) field absorptions for a mixture of 27Al16O and
27Al17O are shown in Fig. 1. Note the consistent phase rela-
tionships and the 17O(Iϭ5/2) sextet hyperfine structure for
both the perpendicular and parallel components. The ob-
served ESR line positions for AlO in neon at 4 K are listed in
Table I and the extracted magnetic parameters are presented
in Table III.
FIG. 2. The overall ESR spectrum of 69Ga16O and 71Ga16O isolated in a
neon matrix at 4 K is shown. A quartet of parallel (␪ϭ0°) and perpendicu-
lar (␪ϭ90°) hyperfine lines is observed for each isotopomer since each
gallium isotope has Iϭ3/2. Background radical absorptions of H and CH3
are marked accordingly. This GaO sample was produced by the pulsed laser
vaporization of elemental gallium in the presence of O2.
B. 69,71Ga16O and 69,71Ga17O
Gallium has two magnetic isotopes, 69Ga ͑60% nat’l
abund., Iϭ3/2, ␮Iϭ2.0145 nm͒ and 71Ga ͑40%, Iϭ3/2, ␮
I
ϭ2.5597 nm͒. The neon matrix ESR spectrum assigned to
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

7014
Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
FIG. 3. The lower neon matrix ESR spectrum shows the intense perpen-
dicular (␪ϭ90°) and weaker parallel (␪ϭ0°) lines for the highest field
transitions (MIϭϪ3/2) of 69Ga16O and 71Ga16O. ͑See overall 69,71Ga16O
spectrum in Fig. 2.͒ The top spectrum covers this same magnetic field region
but for an experiment involving 17O enriched oxygen. The 17O(Iϭ5/2) sex-
tet hyperfine structure is clearly observed on the perpendicular lines and is
detectable on the weaker parallel features when the signal amplification is
increased tenfold as indicated for the highest field region.
FIG. 4. The overall simulated ESR spectrum of 115In16O is shown. This
complex spectrum can be interpreted by reference to its associated HRES vs
␪ plots presented directly beneath the spectrum, where ␪ is the angle be-
tween the molecular axis and the externally applied magnetic field. In addi-
tion to the expected ten parallel (␪ϭ0°) and ten perpendicular (␪ϭ90°)
115In(Iϭ9/2) hyperfine lines, several extra absorption features are seen to
occur at off-angle positions. Only HRES vs ␪ features having significant
transition probabilities are shown. Note that the 115In16O absorptions extend
over the extremely large range of approximately 500–6200 Gauss.
Gallium monoxide ESR spectra of this clarity required
numerous depositions ͑approximately 45 separate experi-
ments͒ in order to optimize the various experimental deposi-
tion conditions such as laser power, matrix flow and oxygen
concentrations, etc. The spectra unambiguously reveal the
presence of one gallium and one oxygen atom in the radical
transitions of 115In(Iϭ9/2) are presented in Fig. 4. This InO
ESR spectrum is a most interesting diatomic radical example
which could not be thoroughly analyzed without the assis-
tance of simulated comparisons. The spectrum is especially
2
species, hence the assignment to GaO in its X ⌺ state seems
unequivocal. The observed GaO line positions are listed in
Table II and the magnetic parameters in Table III.
C. 115In16O and 115In17O
The overall simulated ESR spectrum of 115In16O and its
associated HRES vs ⌰ plots for each of the ten hyperfine
TABLE II. aObserved ESR line positions ͑Gauss͒ for 69Ga16O and 71Ga16O
in neon matrix at 4 K.
69Ga16O
Perpendicular (␪ϭ90°)
Parallel (␪ϭ0°)
Iϭ3/2
MIϭ3/2:
1/2:
Ϫ1/2:
Ϫ3/2:
71Ga16O
Iϭ3/2
MIϭ3/2:
1/2:
2654
3054
3549
4143
2428
2978
3602
4298
2406
2878
3511
4305
2137
2814
3609
4524
Ϫ1/2:
Ϫ3/2
FIG. 5. The three highest field 115In(Iϭ9/2) hyperfine transitions observed
aMicrowave frequency is 9551.8͑4͒ MHz. Calculated line positions, from an
exact diagonalization analysis, agree with these observed lines within the
experimental uncertainty of Ϯ1 Gauss. The magnetic parameters for GaO,
extracted from the line positiions, are listed in Table III. For Ga17O, sextet
hfs for 17O(Iϭ5/2) was resolved for both the parallel and perpendicular
transitions. See Fig. 3.
in a neon matrix ESR spectrum of 115In16O are shown for both the parallel
ʈ
͑ ͒ and perpendicular ͑Ќ͒ components. The simulated ESR spectrum is
shown below the observed. See Fig. 4 for the overall ESR spectrum of
115In16O. The applied magnetic field is given in Gauss. The absorption fea-
ture labeled ‘‘H’’ is the high field transition of the hydrogen atom doublet
hyperfine pattern.
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
7015
TABLE III. aObserved magnetic parameters for Al17O, 69,71Ga17O, and 115In17O in their X 2⌺ ground states in
neon matrices at 4 K.
gʈ
gЌ
bAʈ
871
Ϫ98
1736
bAЌ
714
52
1356
1722
85
1188 ͑2͒
166
cAdip
52 ͑61͒
Ϫ50 (Ϫ65)
127 ͑123͒
162 ͑156͒
Ϫ77 (Ϫ89)
180 ͑173͒
cAiso
Al16O:
17O:
2.0020
2.0011
766 ͑540͒
2 ͑9͒
1483 ͑648͒
1884 ͑823͒
8 ͑2͒
69Ga16O:
71Ga16O:
17O:
1.9990
1.9990
1.9680
1.9680
2207
Ϫ145
115In16O:
17O:
1.975 ͑3͒
1.810 ͑2͒
1728 ͑3͒
Ϫ228
1368 ͑523͒
35 (Ϫ4)
Ϫ131 (Ϫ90)
aAll A values are in MHz. Experimental uncertainties are approximately Ϯ0.0002 in the g values and Ϯ1 MHz
in the A values unless otherwise noted. Aiso and Adip are defined in the text.
bThe signs of the metal A values are clearly positive based upon their large magnitudes. The indicated signs for
the 17O A values were selected to yield the closest agreement with the theoretical calculations for Aiso and Adip
.
cThe Adip and Aiso values shown in parenthesis were calculated using UB3LYP, a density functional approach.
See Table IV. The negative sign of the nuclear magnetic moment for 17O is reflected in the 17O A values listed
in this table.
115In17O. The magnetic parameter for 115In17O are listed in
Table III.
complex at low magnetic fields since the inherent g and A
tensor anisotropies, coupled with the large Iϭ9/2 value, pro-
duce ‘‘extra’’ or off-angle absorptions on the MIϭ7/2, 5/2,
3/2, and 1/2 hyperfine transitions ͑see Table V͒. Also the two
lowest field perpendicular lines are extremely weak and oc-
cur very close together. These spectral properties are consis-
tent with the nature of the HRES vs ⌰ plots shown in Fig. 4.
The high field region of the 115In16O spectrum is straight
forward to interpret and is shown in Fig. 5 where the experi-
mental spectrum is compared directly with a simulated spec-
trum. The magnetic parameters employed to produce this
close fit between simulated and observed spectra are listed in
Table III. The intensity variations of the individual hyperfine
lines are strongly influenced by varying degrees of overlap of
115InO lines in different matrix trapping sites which have
slightly different A and g values. As usual, these matrix site
effects varied greatly among the different depositions and
were partially removed by annealing the neon matrix ͑warm-
ing briefly to 9 K and recooling to 4 K͒.
As discussed below, only a tentative assignment for
115In16O in argon matrices was possible given the difficulties
of obtaining sufficiently intense ESR signals in this host. A
tentative assignment of the 115In16O argon magnetic param-
eters are: gʈϭ1.982(4); gЌϭ1.811(3); Aʈϭ1765(10); and
AЌϭ1205(7) MHz. These show close agreement with the
neon results, and like AlO, exhibit larger A values in argon
relative to neon.
IV. DISCUSSION
The neon magnetic parameters for 27Al16O, produced in
these experiments by reactive pulsed-laser vaporization,
agree within experimental uncertainty with the earlier results
where the AlO radical was generated by conventional high
temperature vaporization of Al2O3͑s͒ at 2500 K from a stan-
dard effusion oven.1 One other example where a detailed
comparison between these two quite difference generation
methods for the study of high temperature species has been
Extremely clear and well-resolved 17O(Iϭ5/2) sextet
hyperfine structure was observed for 115In17O as shown in
Fig. 6 for the parallel and perpendicular 115In: MIϭϪ7/2
high field transition. The same 17O perpendicular and parallel
splittings were observed on the other intense features of
42
2
made is the BFϩ(X ⌺) radical. Since the initial matrix
ESR measurements of the aluminum hyperfine interactions
were reported, gas phase results have become available.2 As
TABLE IV. aCalculated nuclear hyperfine properties for AlO, GaO, and InO in their X 2⌺ states.
c
Radical
27Al17O
Method
E
A
iso(M)
bBasis
gN
R
cq(M)
cqs(M)
S2
Adip(M)
Aiso͑17O͒
Adip͑17O͒
͗ ͘
UHF
UB3LYP
UHF
UB3LYP
UHF
UB3LYP
Ϫ316.7785
Ϫ317.6817
Ϫ1998.1415
Ϫ2000.1706
Ϫ5814.9759
Ϫ5817.8483
Ϫ466
540
61
61
118
123
163
173
Ϫ41
9
Ϫ33
2
Ϫ55
Ϫ4
Ϫ106
Ϫ65
Ϫ106
Ϫ89
Ϫ112
Ϫ90
22s,16p,4d
1.4565
1.6179
0.612
0.732
0.589
0.598
0.878
0.878
0.014
0.489
0.059
0.313
0.079
0.306
0.812
0.758
0.805
0.761
0.804
0.763
69Ga17O
115In17O
Ϫ1048
648
Ϫ1440
523
22s,17p,12d
23s,18p,13d
1.3430
1.2300
1.7436
1.8768
aA values reported in MHz, R in Å, and energy (E) in hartrees. The nuclear g factor is denoted by gN and has a value of Ϫ0.7575 for 17O. The Aiso and Adip
parameters are defined in the text. The density functional theory method is designated UB3LYP.
bThe oxygen basis set was 21s, 15p, 4d. All basis functions were uncontracted, spherical harmonic Gaussians.
cq(M) is total charge on metal; g (M) in total spin density on metal; S2 reflects the degree of spin contamination of the wave function. For pure doublet
͗
͘
s
states, the expected value is 0.75.
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

7016
Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
TABLE V. aObserved ESR line positions ͑Gaussian͒ for 115In16O in neon
matrix at 4 K.
what less dependent on the particular matrix host. Since the
nuclear hyperfine interaction for only a very small number of
high temperature radicals has been resolved in the gas phase,
this AlO comparison is especially valuable. It also implies
that these new neon results for GaO and InO should be close,
within a few percent, of the gas phase values. This assump-
tion will be an important consideration when detailed
experimental-theoretical comparisons are made in the fol-
lowing section of this article.
Parallel (␪ϭ0°)
Off angle
Perpendicular (␪ϭ90°)
115In:Iϭ9/2
MIϭ9/2
7/2
523
920
•••
1482 ͑81°͒
1667 ͑72°͒
1980 ͑57°͒
2440 ͑36°͒
•••
1358
1390
1589
1834
2199
2692
b3309
4033
4840
5708
5/2
3/2
1/2
Ϫ1/2
Ϫ3/2
Ϫ5/2
Ϫ7/2
Ϫ9/2
1362
1859
2419
b3046
3741
4499
5312
6171
The matrix isolation Fourier transform infrared ͑FTIR͒
vibrational studies using the reactive laser vaporization
method have detected BO and AlO in argon but not GaO and
InO.19,20 However, reaction products presumably resulting
from the heavier MO molecules under these conditions were
detected, for example M2O2. Neon matrix vibrational experi-
ments were apparently not conducted. Consistent with the
argon vibrational results, we were able to obtain only weak
InO and no GaO argon matrix ESR spectra despite numerous
experimental attempts under a variety of experimental con-
ditions. Both direct and indirect encounters between O2(g)
and laser-ablated metal were tried over wide concentration
ranges. In the direct scheme, pure O2(g) is passed directly
over the metal undergoing laser vaporization. The indirect
method involves the introduction of O2(g) as a dopant in the
rare gas flow, hence encounters between metal atoms and
O2(g) occur predominantly during the matrix condensation
process. The difficulties of detecting GaO and InO in argon
matrices by both ESR and IR spectroscopy is puzzling. Re-
action between O2 and these heavier metals might depend
upon excited state encounters. Hence, more efficient excited
state quenching by the heavier rare gases might account for
these observations. The smaller atomic and molecular diffu-
sion rates in argon relative to neon might also be an impor-
tant difference.
•••
•••
•••
•••
aThe microwave frequency is 9587.0͑4͒ MHz. Given the larger linewidths of
these 115InO absorptions, the experimental uncertainty is approximately
Ϯ3 Gauss. All lines, including the ‘‘extra’’ or off-angle features were fitted
with our exact diagonalization program using the magnetic parameters
listed in Table III.
bThese lines were obscured by intense background absorptions.
is usually the case, the neon results show the closest agree-
ment with the gas phase values. The gas phase, neon, argon,
and krypton Aiso values for 27Al in 27Al16O are 738.0͑14͒,
766͑2͒, 899͑3͒, and 920͑3͒ MHz, respectively. The 27Al di-
polar parameters (Adip), listed in the same order are,
56.392͑82͒, 53.0͑7͒, 54͑1͒, and 51͑1͒ MHz.
For Aiso , these comparisons show a neon-gas shift of
3.8% and an argon shift of 22%. The Adip parameter is some-
The conversions from the observed ESR hyperfine pa-
rameters ͑Aʈ and AЌ͒ to the properties ͑Aiso and Adip͒ that are
required to establish electronic structure trends among these
diatomic oxide radicals are conducted with the following re-
lationships:
A
A
isoϭ8/3␲gegn␤e␤n ␦ r͒ ϭ 2A ϩA ͒/3
͑ ͑
͗ ͘
Ќ
ʈ
dipϭ1/2gegn␤e␤n 3 cos2 ␪Ϫ1͒/r3 ϭ A ϪA ͒/3,
͑
͑
ʈ
͗
͘
Ќ
where all symbols have their standard definitions and the
averages are taken over the spin densities.25 The small L•I
ជ ជ
terms are neglected since the g values do not differ substan-
tially from ge .
V. ELECTRONIC STRUCTURE TRENDS
FIG. 6. The 17O(Iϭ5/2) hyperfine structure associated with the parallel and
perpendicular components of the 115In:MIϭϪ7/2 transition is shown for the
115In17O radical. The unsplit central features result from 115In16O radicals
which are also present in this neon matrix sample. The up-phase absorption
feature labelled ‘‘NG’’ ͑lower right͒ results from the beginning of the ‘‘next
group’’ of perpendicular lines associated with the 115In:MIϭϪ9/2 transition
for the 115In17O radical. The signal amplification factor for the weaker par-
allel spectrum ͑lower trace͒ was four times that employed for the more
intense perpendicular components ͑upper trace͒.
These experimental Aiso and Adip values for a specific
nucleus in a molecular radical can be compared to a set of
calculated free atom parameters to yield estimates of valence
‘‘s’’ and ‘‘pz’’ orbital characters, respectively. The free
atom comparison method ͑FACM͒ results using a commonly
applied set of atomic values43 for BO, AlO, GaO, and InO
are listed in Table VI where both metal and oxygen valence
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
7017
TABLE VI. aFree atom comparison method ͑FACM͒ results for the MO diatomic radicals: valence electronic
structure trends.
Metal
Adip
Oxygen (17O)
X(2pz) f⌺ TOTAL
X(2s)
Aiso
X(npz)
Aiso
X(ns)
Adip
b11B17O
1033
1027.4
766
738.0
1483
25
27.1
52
0.41
0.39 Ϫ19 (3) Ϫ12 (3)
0.004
0.07
0.87
1.13
cGas phase
d27Al17O
0.20
0.63
2
Ϫ50 Ϫ0.0004 0.30
eGas phase
d69Ga17O
d115In17O
56.4
127
180 ͑3͒
0.12
0.07
0.62
0.63
8
35
Ϫ77 Ϫ0.002
Ϫ131 Ϫ0.007
0.46
0.78
1.20
1.48
1368 ͑3͒
aA values reported in MHz. See text for Aiso and Adip definitions and the sign choices for the A values. The
valence orbital characters (X) are obtained by a simple comparison with a standard set of free atom Aiso and
Adip values. ͑Ref. 43͒ Unless otherwise indicated all A values are from neon matrix ESR measurements and
have an experimental uncertainty of Ϯ1 MHz.
bRef. 3.
cRef. 4.
dThis work.
eRef. 2.
fA discussion as to why some of these totals exceed unity is given in the text.
orbital characters are shown. A detailed analysis of inaccu-
racies that can arise with this simple FACM treatment have
been thoroughly discussed in an earlier study of several di-
atomic radicals, including AlO and its isoelectronic cation,
SiOϩ.44 A major source of error is introduced into FACM
when ‘‘core-other valence overlap’’ is significant. This ef-
fect, which is not spin polarization, was shown to be large
for oxygen 2pz overlap with the inner shells of aluminum.
The FACM values for Al in AlO, 3sϭ0.20 and 3pzϭ0.63,
needed to be adjusted to 0.41 and 0.26, respectively.44
Hence, the FACM s/pz ratio changed from 0.32 to 1.58 after
this effect and others were included in the theoretical analy-
sis.
The origin of the core-other valence overlap effect is
briefly described for the A–B diatomic radical. Orthogonal-
izing the valence orbitals of A to the core of B introduces
nonzero coefficients multiplying the core orbitals of B into
the valence MOs. Because of the definition of the Mulliken
population, these core orbitals make no contribution to the
Mulliken population. When the valence MO is evaluated at
the B nucleus, however, the core orbital part of the valence
MO makes a large contribution. This makes the FACM re-
sult of dividing the spin density by the valence spin density
of the atom an unreliable method for determining spin popu-
lations.
It is highly significant that the oxygen A values, which
were not available at the time of the earlier Al16O/Si16Oϩ
study, yield FACM 2s and 2pz characters that show nearly
exact agreement with the earlier calculated ͑CI͒ results of
2sϷ0 and 2pzϭ0.30. Since the oxygen 2pz character is not
affected by core-other valence overlap effects, the FACM
values for this parameter should provide a more accurate
description of the bonding trend for this series of metal oxide
radicals. Going from B17O to In17O, the oxygen 2pz charac-
ter increases in the order 0.07, 0.30, 0.46, and 0.78. As ex-
pected, the oxygen 2s character is nearly zero for all four
MO radicals.
down the group would seem to reflect an overall electronic
shift from MϩϩOϪϪ to MϩOϪ. Certainly, the experimental
trend in the 17O Adip parameters does indicate that consider-
ably more spin density does reside on oxygen going down
the group but this alone does not necessarily require large
shifts in the overall charge—a topic that is discussed more
fully in the theoretical section.
Although we have not conducted as high a level of or-
bital population analysis and calculation for GaO and InO as
was done earlier for AlO, it would not be surprising to also
find a large core-other valence overlap effect operative in the
heavier oxides. Inclusion of the effect would most probably
reduce the large npz metal characters predicted by FACM by
a significant amount. This seems reasonable since the sum of
the valence characters for GaO ͑1.20͒ and InO ͑1.48͒ are
considerably greater than unity and large corrections to the
oxygen 2pz characters are not warranted, given the absence
of an inner ‘‘p’’ shell. A Mulliken population analysis of the
DFT Slater determinant with the extended basis set is subject
to many criticisms. For what it is worth, it shows that the
spin density for GaO has 0.1s, 0.24p on Ga and 0.1s and
0.55p on O. This agrees with the idea that the neutral p
character on the metal is overestimated by FACM, but the
oxygen is reasonable. Based upon FACM predictions alone,
the metal ns valence character decreases in the order 0.41,
0.20, 0.12, and 0.07 going down the group, BO to InO. With-
out extensive calculations on these heavier oxides with ex-
tremely complicated inner-shell effects, it is difficult to judge
how these FACM estimates of the ns character should be
adjusted.
The theoretical calculations of Aiso and Adip for the metal
atoms and oxygen are given in Table III in parenthesis
alongside the experimental results. A detailed description of
these calculations is provided in the next section. Agreement
between theory ͑UB3LYP͒ and experiment is quite good for
the Adip values in all cases except for oxygen in the AlO and
InO cases, and even there, the agreement is within 33%.
Deviation between observed and calculated metal Aiso values
The large increases in the oxygen 2pz character going
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

7018
Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
increases from Ϫ30% in AlO to Ϫ62% in InO.
ing the DFT ͑UB3LYP͒ method show reasonable agreement
with experiment as shown by the comparison data of Table
III. For example, Adip for Al in AlO is calculated to be 61
MHz compared to the observed value in neon of 52 MHz.
Even for InO, the metal’s observed Adip value of 180 MHz
shows good agreement with the calculated value of 173
MHz.
The increasing negative deviation of the gʈ and gЌ val-
ues from ge going down this MO radical series ͑see Table
III͒ reflects the large increases in the spin-orbit coupling pa-
rameters of the metal atoms and possible changes in the na-
ture of the excited 2⌸ states. Spin-orbit coupling between the
2
2
ground X ⌺ electronic states and excited ⌸ states can be-
2
2
come quite complicated since both ⌸i and ⌸r types are
involved, and these two types shift the g values in opposite
directions.7 Whether the excited 2⌸ states are similar in GaO
and InO, for example, can be estimated by determining if
⌬gЌ(gЌϪge) is merely reflecting changes in the metal atom
spin-orbit parameters ͑␭͒. As shown by the g tensor data in
Table III, the ⌬gЌ ratio for InO/GaO is 5.6 compared to a
spin-orbit ratio of 2.7 for these two metals. Hence, the
greater than expected gЌ shift for InO, based on ␭ alone,
Past results for AlO and SiOϩ have already
demonstrated44,49 that these were very difficult molecules for
spin density calculations. The MO molecule is actually best
described as somewhere between MϩϩOϪϪ with the un-
paired electron in the metal s orbital and MϩOϪ with the
unpaired electron in the oxygen sp hybrid orbital polarized
away from the metal. Without a proper description of elec-
tron correlation, the MϩϩOϪϪ configuration lies too high in
energy relative to MϩOϪ so the wave function is biased to-
wards MϩOϪ. Obviously the A values measured in the ESR
experiment are extremely sensitive to this balance. Since
these two configurations differ nominally by a single excita-
tion, ␴(M)2␣(O)1→␴(M)1␴(O)2, they have no connecting
matrix element in the CI. Mixing between them, in the CI, is
described by coupling through a large number of double ex-
citations. Consequently, MRSDCI is not very effective since
no small reference space brings in the single excitation to the
correct extent. MRSDCI works best when all of the coeffi-
cients in the reference space are well represented by first
order perturbation theory.
2
implies that the ⌸r state is more strongly coupled to the
2
ground ⌺ state relative to GaO.
VI. THEORETICAL RESULTS
Calculations for AlO, GaO, and InO were performed
46
with an extended basis set45 using the MELD
and
47
GAUSSIAN94 programs in order to compare theoretical esti-
mates of the nuclear hyperfine properties with these neon
matrix experimental results ͑see Table IV͒. For AlO, the re-
sults were generally unstable with respect to the quality of
the CI calculation. The spin density at the metal nucleus,
especially, varied greatly with method.
Because these two configurations can always be rewrit-
ten as a single Slater determinant, methods like DFT that
build the effect of electron correlation into an effective po-
tential can obtain an improved spin density from a single
determinant model. Certainly in this case, the spin density is
greatly improved. On the other hand, agreement with experi-
ment is still far from perfect. Examination of the Mulliken
charges shows, however, that the charge on the metal is not
greatly affected by going from UHF to UB3LYP. Rather, the
spin density change is associated with a redistribution of the
singly occupied orbital compensated by a shift in the other
direction of the doubly occupied orbitals so that the charge
distribution remains nearly unchanged. For example, the ini-
For AlO, Yoshimine et al.48 have pointed out that there
are two solutions to the ͑ROHF͒ equations with nearly equal
energy, and that the one obtained depends on the initial
guess. The spin density for AlO has also been calculated in
two of our previous articles44,49 using CI methods and itera-
tive natural orbitals, but with smaller basis sets. These pre-
vious calculations gave reasonable (Ϯ50%) agreement with
the aluminum Aiso value. With the present large uncontracted
basis, the value of Aiso at Al was ϩ3 MHz but with a con-
traction of the tightest s and p functions, the value jumped to
ϩ1110 MHz. The UHF value with the uncontracted basis is
Ϫ416 MHz and single double configuration interaction
͑SDCI͒, which includes most of the spin polarization effects,
likewise gave a negative value with both basis sets. Very
large multireference single double configuration interaction
͑MRSDCI͒ with either basis gave positive values for Aiso in
the 100–300 MHz range with the uncontracted basis and the
700–900 MHz range with the contracted basis.
Density functional theory with spin-dependent general-
ized gradient functionals has been found empirically50 to
give reasonable spin densities for many systems including
those with heavy metals. The uncontracted basis and the
B3LYP51,52 method applied to AlO gave 540 MHz for Aiso
which shows better agreement with experiment than any of
the CI calculations with the same basis. For GaO, the results
were similar. The ROHF Aiso(69Ga) was ϩ35 MHz, while
the largest MRSDCI yielded Ϫ47 MHz. By contrast, the
UHF result was Ϫ1048 MHz and B3LYP gave ϩ648 MHz.
The calculated Adip values for the metal and oxygen us-
2
tial p electron on the free metal atom in its P ground state
goes into forming a ␲ bond with the oxygen. The polarity of
this ␲ bond is not directly measured by ESR, but is indirectly
connected to the distribution of ␴ electrons.
It is dangerous to compare Mulliken charge between dif-
ferent molecules because ͑a͒ the extended basis sets with
diffuse functions are only nominally associated with each
atom and ͑b͒ density functional theory yields a density but
does not give a density matrix. Like S2 , the Mulliken
͗
͘
population relies on treating the Slater determinant formed
from DFT orbitals as if it were an actual wave function.
ACKNOWLEDGMENTS
Support of the National Science Foundation ͑LBK:
CHE-9319291͒ and ͑ERD: CHE-9526161͒ is gratefully ac-
knowledged. A Duke Endowment grant to Furman Univer-
sity and a Research Corporation Cottrell College Science
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07

Knight, Jr. et al.: Matrix isolation studies of 27Al16,17
O
7019
Award ͑L.B.K.͒ also provided valuable support for this
project. The authors would like to express their appreciation
to Robert Babb and Edward Earl for providing important
computer assistance in various stages of this work.
30 L. B. Knight, Jr., S. T. Cobranchi, J. O. Herlong, and C. A. Arrington, J.
Chem. Phys. 92, 5865 ͑1990͒.
31 L. B. Knight, Jr., J. R. Woodward, T. J. Kirk, and C. A. Arrington, J.
Phys. Chem. 97, 1304 ͑1993͒.
32 Y. M. Hamrick, R. J. Van Zee, and W. Weltner, Jr., J. Chem. Phys. 96,
1767 ͑1992͒.
33 J. M. Parnis and G. A. Ozin, J. Phys. Chem. 93, 1215 ͑1989͒; 93, 1204
͑1989͒; G. Jeong and K. J. Klabunde, J. Am. Chem. Soc. 108, 7103
͑1986͒.
1 L. B. Knight, Jr. and W. Weltner, Jr., J. Chem. Phys. 55, 5066 ͑1971͒.
2 C. Yamada, E. A. Cohen, M. Fujitake, and E. Hirota, J. Chem. Phys. 92,
2146 ͑1990͒.
34 P. H. Kasai and P. M. Jones, J. Am. Chem. Soc. 106, 8018 ͑1984͒.
35 L. B. Knight, Jr. and J. T. Petty, J. Chem. Phys. 88, 481 ͑1988͒.
36 J. A. Howard and B. Mile, Acc. Chem. Res. 20, 173 ͑1987͒.
37 L. B. Knight, Jr., J. J. Banisaukas III, R. Babb, and E. R. Davidson, J.
Chem. Phys. 105, 6607 ͑1996͒.
3 L. B. Knight, Jr., J. O. Herlong, T. J. Kirk, and C. A. Arrington, J. Chem.
Phys. 96, 5604 ͑1992͒.
4 M. Tanimoto, S. Saita, and E. Hirota, J. Chem. Phys. 84, 1210 ͑1986͒.
5 B. H. Lengsfield III and B. Liu, J. Chem. Phys. 77, 6083 ͑1982͒.
6 W. Weltner, Jr., D. McLeod, and P. H. Kasai, J. Chem. Phys. 46, 3172
͑1967͒.
38 R. J. Van Zee, S. Li, and W. Weltner, Jr., J. Chem. Phys. 98, 4335 ͑1993͒.
39 P. M. Jones and P. J. Kasai, J. Phys. Chem. 92, 1060 ͑1988͒.
40 P. Pullumbi, C. Mijoule, L. Manceron, and Y. Bouteiller, Chem. Phys.
185, 13 ͑1994͒.
7 K. P. Huber and G. Herzberg, Constants of Diatomic Molecules (IV) ͑Van
Nostrand Reinhold, New York, 1979͒; P. J. Dagdigian, H. W. Cruse, and
R. N. Zare, J. Chem. Phys. 62, 1824 ͑1975͒.
41 L. B. Knight, Jr., Radical Ionic Systems, edited by A. Lund and M. Shio-
tani ͑Kluwer Academic, New York, 1991͒, pp. 73–97.
8 M. Cai, C. C. Carter, T. A. Miller, and V. E. Bondybey, J. Chem. Phys.
95, 73 ͑1991͒.
42 L. B. Knight, Jr., A. Ligon, S. T. Cobranchi, D. P. Cobranchi, E. Earl, D.
Feller, and E. R. Davidson, J. Chem. Phys. 85, 5437 ͑1986͒.
43 J. R. Morton and K. R. Preston, J. Magn. Reson. 30, 577 ͑1978͒.
44 L. B. Knight, Jr., A. Ligon, R. W. Woodward, D. Feller, and E. R. David-
son, J. Am. Chem. Soc. 107, 2857 ͑1985͒.
9 D. E. Clemmer, M. E. Weber, and P. B. Armentrout, J. Chem. Phys. 96,
10 888 ͑1992͒.
10 D. P. Belyung, A. Fontijn, and P. Marshall, J. Phys. Chem. 97, 3456
͑1993͒.
11 G. A. Slavejkov, C. T. Clyde, and A. Fontijn, J. Phys. Chem. 94, 3347
͑1990͒.
45 H. Partridge, NASA Technical Memorandum 101 044, 1989. The largest
basis sets from this source were used for gallium (21s16p10d), supple-
mented by one additional diffuse s, p, and two additional diffuse d func-
tions. For aluminum, the Partridge 20s15p basis was supplemented by
two additional diffuse s, one additional p, and four d polarization func-
tions. For oxygen, three additional s, two additional p, and four d polar-
ization functions were added to his 18s13p set. For indium, no optimized
basis was available, so the s, p basis of gallium was scaled by 1.6 ͑the
ratio of nuclear charges͒ and the d basis was scaled by 2.0. Three addi-
tional s, 2 additional p, and one additional d function were added to cover
the additional valence shell.
12 A. P. Salzberg, D. I. Santiago, F. Asmar, D. N. Sandoval, and B. R.
Weiner, Chem. Phys. Lett. 180, 161 ͑1991͒.
13 M. L. Campbell, R. E. McClean, N. L. Garland, and H. H. Nelson, Chem.
Phys. Lett. 194, 187 ͑1992͒.
14 M. Singh, Astrophys. Space Sci. 140, 421 ͑1988͒.
15 H. Partridge, S. R. Langhoff, B. H. Lengsfield III, and B. Liu, J. Quant.
Spectrosc. Radiat. Transf. 30, 449 ͑1983͒.
16 H. R. Johnson and A. J. Sauval, Astron. Astrophys. Suppl. Ser. 49, 77
͑1982͒.
17 J. A. Coxon and S. Naxakis, J. Mol. Spectrosc. 111, 102 ͑1985͒.
18 R. W. Dreyfus, R. Kelly, and R. E. Walkup, Appl. Phys. Lett. 49, 1478
͑1986͒.
46
MELD is a set of electronic structure programs written by L. E. McMurchie,
19 L. Andrews, T. R. Burkholder, and J. T. Yustein, J. Phys. Chem. 96,
10 182 ͑1992͒.
S. T. Elbert, S. R. Langhoff, and E. R. Davidson, with extensive modifi-
cations by D. Feller and D. C. Rawlings.
GAUSSIAN94, M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B.
47
20 T. R. Burkholder, J. T. Yustein, and L. Andrews, J. Phys. Chem. 96,
10 189 ͑1992͒.
G. Johnson, M. A. Robb, J. R. Cheeseman, T. A. Keith, G. A. Peterson, J.
A. Montgomery, K. Raghavachari, M. A. Al-Laham, V. G. Zakrzewski, J.
V. Ortiz, J. B. Foresman, J. Cioslowski, B. B. Stefanov, A. Nanayakkara,
M. Challacombe, C. Y. Peng, P. Y. Ayala, W. Chen, M. W. Wong, J. L.
Andres, E. S. Replogle, R. Gomperts, R. L. Martin, D. J. Fox, J. S. Bin-
kley, D. J. Defrees, J. Baker, J. P. Stewart, M. Head-Gordon, C. Gonzalez,
and J. A. Pople ͑Gaussian, Inc., Pittsburgh, PA, 1995͒.
21 S. J. Bares, M. Haak, and J. W. Nibler, J. Chem. Phys. 82, 670 ͑1985͒.
22 S. J. Bares and J. W. Nibler, J. Mol. Struct., 157, 215 ͑1987͒.
23 A. J. Hinchcliffe and J. S. Ogden, J. Phys. Chem. 77, 2537 ͑1973͒.
24 M. A. Douglas, R. H. Hauge, and J. L. Margrave, High Temp. Sci. 16, 35
͑1983͒; M. A. Douglas, R. H. Hauge, and J. L. Margrave, J. Am. Chem.
Soc. 102, 6005 ͑1980͒.
25 W. Weltner, Jr., Magnetic Atoms and Molecules ͑Dover, Mineola, New
York, 1989͒.
48 M. Yoshimine, A. D. McLean, and B. Liu, J. Chem. Phys. 58, 4412
͑1973͒.
26 L. B. Knight, Jr., G. C. Jones, G. M. King, R. M. Babb, and A. J. McK-
inley, J. Chem. Phys. 103, 497 ͑1995͒.
49 L. B. Knight, Jr., M. B. Wise, E. R. Davidson, and L. E. McMurchie, J.
Chem. Phys. 76, 126 ͑1982͒; L. B. Knight, Jr., J. O. Herlong, T. J. Kirk,
and C. A. Arrington, J. Chem. Phys. 96, 5604 ͑1992͒.
27 L. B. Knight, Jr., R. Babb, M. Ray, T. J. Banisaukas III, L. Russon, R. S.
Dailey, and E. R. Davidson, J. Chem. Phys. 105, 10 237 ͑1996͒.
28 L. B. Knight, Jr., S. T. Cobranchi, B. W. Gregory, and E. Earl, J. Chem.
Phys. 86, 3143 ͑1987͒.
50 E. R. Davidson and R. F. Frey, J. Chem. Phys. 106, 2331 ͑1997͒.
51 A. D. Becke, J. Chem. Phys. 98, 5648 ͑1993͒; A. D. Becke, Phys. Rev. A.
38, 3098 ͑1988͒.
29 L. B. Knight, Jr., E. Earl, A. R. Ligon, D. P. Cobranchi, J. R. Woodward,
J. M. Bostick, D. R. Davidson, and D. Feller, J. Am. Chem. Soc. 108,
5065 ͑1986͒.
52 C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B 37, 785 ͑1988͒; B. Mie-
hlich, A. Savin, H. Stoll, and H. Preuss, Chem. Phys. Lett. 157, 200
͑1989͒.
J. Chem. Phys., Vol. 107, No. 18, 8 November 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
132.174.255.116 On: Thu, 18 Dec 2014 10:12:07