Luminescent Photofragments of (1,1,1,5,5,5-Hexafluoro-2,4-pentanedionato) Metal Complexes in the Gas Phase

Article abstract of DOI:10.1021/ic971340x

The luminescence that is observed under gas phase photolytic deposition conditions is studied for Cr(hfac)₃, Ni(hfac)₂, and Pt(hfac)₂. This luminescence is analyzed under a variety of conditions, including the relatively high pressures of an evacuated gas cell and the collision-free conditions of a molecular beam. The effects of inert buffer gas are also studied. Features in these spectra indicate that, in general, multiple photolysis processes occur. Some simple fragments that are produced from these compounds are identified, including bare metal atoms (Ni, Cr), metal monofluorides (NiF, CrF), CH (in the case of Ni(hfac)₂), and metal carbide from Pt(hfac)₂. It is postulated that the difference in the observed photofragmentation pathway in the case of platinum is due to σ bonding to the β carbon of the hfac moiety as opposed to the bidentate bonding of the other two metals. Possible mechanisms are presented. Detailed analysis of the spectra allows characterization of the internal energy of the platinum carbide photofragment.

Full text of DOI:10.1021/ic971340x

2880
Inorg. Chem. 1998, 37, 2880-2887
Luminescent Photofragments of (1,1,1,5,5,5-Hexafluoro-2,4-pentanedionato) Metal
Complexes in the Gas Phase
David S. Talaga, Stephen D. Hanna, and Jeffrey I. Zink*
Department of Chemisty and Biochemistry, University of California, Los Angeles,
Los Angeles, California 90095-1569
ReceiVed October 22, 1997
The luminescence that is observed under gas phase photolytic deposition conditions is studied for Cr(hfac)3,
Ni(hfac)2, and Pt(hfac)2. This luminescence is analyzed under a variety of conditions, including the relatively
high pressures of an evacuated gas cell and the collision-free conditions of a molecular beam. The effects of
inert buffer gas are also studied. Features in these spectra indicate that, in general, multiple photolysis processes
occur. Some simple fragments that are produced from these compounds are identified, including bare metal
atoms (Ni, Cr), metal monofluorides (NiF, CrF), CH (in the case of Ni(hfac)2), and metal carbide from Pt(hfac)2.
It is postulated that the difference in the observed photofragmentation pathway in the case of platinum is due to
σ bonding to the â carbon of the hfac moiety as opposed to the bidentate bonding of the other two metals.
Possible mechanisms are presented. Detailed analysis of the spectra allows characterization of the internal energy
of the platinum carbide photofragment.
Introduction
structures with well-defined physical properties and abrupt
interfaces benefit from the low growth temperatures.22-24
Metal acetylacetonate (2,4-pentanedionate ) acac) complexes
have found utility as CVD precursors25-27 including applications
in metallization for semiconductor interconnects17 and prepara-
tion of metal oxides for high Tc superconducting films.28
Fluorinated acetylacetonate Iigands (1,1,1,5,5,5-hexafluoro-2,4-
pentanedionate ) hfac) increase sample volatility and allow for
more facile transport in the gas phase but may result in fluorine
contamination of the final deposit.29 For example, photolytic
CVD of copper from Cu(hfac)2 shows measurable fluorine
incorporation.30,31 Luminescence spectra were used to identify
the photofragments resulting from photolysis of Cu(hfac)2 in
the gas phase.1 The luminescence from 308 nm excitation of
Cu(hfac)2 under “high pressure” conditions (∼1 bar), “low
pressure” conditions (∼0.1-10 mbar), and collision-free condi-
tions (molecular beam, 10-4 mbar) was studied to provide
Recent studies have shown that luminescence is observed
when metal-containing molecules are irradiated under photo-
chemically driven chemical vapor deposition (CVD) conditions
and that luminescence spectroscopy can be used to identify
photofragments and assist in the elucidation of the photolytic
deposition pathways.1-3 Laser-assisted metal organic CVD
relies on the use of volatile organometallic compounds to deliver
material to a substrate where the compound is photodecomposed
leaving behind the material to be deposited (and possibly
undesired contaminants).2-14 Laser-assisted CVD has the
advantages of spatially selective deposition on the substrate,
selective energy transfer to the deposition precursor, and low
processing temperature.15-21 Efforts to fabricate multilayer
(1) Talaga, D. S.; Zink, J. I. Inorg. Chem. 1996, 35, 5050.
(2) Wexler, D.; Zink, J. I.; Tutt, L. W.; Lunt, S. R. J. Phys. Chem. 1993,
97, 13563.
(3) Cheon, J.; Talaga, D. S.; Zink, J. I. J. Am. Chem. Soc. 1997, 119,
163.
(4) Cheon, J.; Talaga, D. S.; Zink, J. I. Chem. Mater. 1997, 9, 1208.
(5) Cheon, J.; Zink, J. I. J. Am. Chem. Soc. 1997, 119, 3838.
(6) Mogyorosi, P.; Carlsson, J. O.; Moradi, M. Appl. Surf. Sci. 1992, 54,
46.
(18) Lee, E. J. B.; Bitner, T. W.; Ha, J. S.; Shane, M. J.; Sailor, H. J. J.
Am. Chem. Soc. 1996, 118, 5375.
(19) Bhatia, S. K.; Hickman, J. J.; Ligler, F. S. J. Am. Chem. Soc. 1992,
114, 4432.
(20) Dressick, W. J.; Dulcey, C. S.; Georger, J. H.; Calvert, J. M. Chem.
Mater. 1993, 5, 148.
(7) Messelhauser, J.; Flint, E. B.; Suhr, H. Appl. Surf. Sci. 1992, 54,64.
(8) Vogel Koplitz, L.; Shuh, D. K.; Chen, Y. J.; Williams, R. S.; Zink, J.
I. Appl. Phys. Lett. 1988, 53, 1705.
(9) Rand, M. J. J. Electrochem. Soc. 1973, 120, 686.
(10) Cohan, J. S.; Yuan, H.; Williams, R. S.; Zink, J. I. Appl. Phys. Lett.
1992, 60, 1402.
(11) Flint, E. B.; Messelhauser, J.; Suhr, H. Appl. Phys. A 1991, 53, 430.
(12) Kargl, P. B.; Kullmer, R.; Bauerle, D. Appl. Phys. A 1993, 57, 175.
(13) Mogyorosi, P.; Carlsson, J. O. Mater. Res. Soc. Symp. Proc. 1992,
236,123.
(14) (a) Heszler, P.; Carlsson, J. O.; Mogyorosi, P. J. Vac. Sci. Technol. A
1993, 11, 2924. (b) Hester, P.; Mogyorosi, P.; Carlsson, J. O. Appl.
Surf. Sci. 1993, 69, 272. (c) Heszler, P.; Mogyorosi, P.; Carlsson, J.
O. Appl. Surf. Sci. 1993, 69, 376. (d) Heszler, P.; Mogyorosi, P.;
Carlsson, J. O. J. Appl. Phys. 1995, 78, 5277.
(21) Moreau, W. M. Semiconductor Lithography; Principles, Practices,
and Materials; Plenum: New York, 1988.
(22) Kompa, K. L. Angew. Chem., Int. Ed. Engl. 1988, 27, 1314.
(23) Herman, I. P. Chem. ReV. 1989, 89, 1323.
(24) Eden, J. G. Photochemical Vapor Deposition; Wiley: New York, 1992.
(25) Hitchman, M. L.; Jensen, K F. Chemical Vapor Deposition: Principles
and Applications; Academic Press: San Diego, 1993.
(26) Kodas, T. T.; Hampden-Smith, M. J. The Chemistry of Metal CVD;
VCH: Weinheim, 1994.
(27) Kaesz, H. D.; Williams, R. S.; Hicks, R. F.; Zink, J. I.; Chen, Y. J.;
Muller, H. J.; Xue, Z. L.; Xu, D. Q.; Shuh, D. K.; Kim, Y. K. New J.
Chem. 1990, 14, 527.
(28) Schulz, D. L.; Marks, T. J. AdV. Mater. 1994, 6, 719.
(29) Nemoto, M.; Yamanaka, M. J. Mater. Res. 1990, 5, 1.
(30) Jones, C. R.; Houle, F. A.; Kovac, C. A.; Baum, T. H. Appl. Phys.
Lett. 1985, 46, 97.
(31) Houle, F. A.; Wilson, R. J.; Baum, T. H. J. Vac. Sci. Technol. A 1986,
4, 2452.
(15) Avey, A. A.; Hill, R. H. J. Am. Chem. Soc. 1996, 118, 237.
(16) Larciprete, R. Appl. Surf. Sci. 1990, 46, 19.
(17) Doppelt, P.; Baum, T. H. Mater. Res. Soc. Bull. 1994, 19, 41.
S0020-1669(97)01340-2 CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/15/1998

Luminescent Photofragments of hfac Complexes
Inorganic Chemistry, Vol. 37, No. 12, 1998 2881
Luminescence spectra under deposition conditions are measured in
a continuously evacuated glass spectroscopy cell fitted with quartz
windows. A sample of the CVD precursor is placed in the sample
chamber and is leaked into the photolysis chamber with a Teflon valve.
Buffer gases are introduced via a separate inlet. Alternatively, an
evacuated stainless steel six-way cross with synthetic fused silica
windows is used. A sample of the CVD precursor is placed in the
sample chamber and is leaked into the photolysis chamber with a needle
valve. Buffer gases can be introduced in conjunction with the sample
or via a separate inlet allowing luminescence spectra to be obtained
under a variety of pressure conditions [e.g. dynamic vacuum (10-2 Torr),
static vacuum, Ar buffer gas (∼1-1000 Torr)]. For both chambers,
the entire cell is heated to the sublimation temperature of the compound
studied (typically 70-140 °C) using thermostated heating tape. Both
of these chambers are referred to as the “gas cell” in the text.
The laser is focused with a L ) 1 in., f ) 50 mm, quartz lens. The
resulting fluence is typically ∼3 MW/cm2 for the excimer and ∼12
MW/cm2 for the dye laser. The focused output of a XeCl excimer
laser (308 nm), or an excimer pumped dye laser (343 or 440 nm) excites
the gaseous sample and the emitted light is collected by f/4 optics at
right angles and directed into a 0.32 m single monochromator (JY
HR320) where it is dispersed by a 300 or 600 gr/mm holographic
grating and detected by an UV-intensified diode array detector (EG&G
Princeton Applied Research OMA3, 1024 × 1) or a liquid nitrogen
cooled CCD detector (Princeton Instruments, 1024 × 256). The spectral
width of a pixel is ∼0.24 and ∼0.12 nm for the 300 and 600 gr/mm
gratings, respectively. The UV intensifier allows time-resolved gating
of the detector with a resolution of 10 ns.
The molecular beam experiments are carried out in a cubic stainless
steel vacuum chamber (30 cm edge length). A mechanical pump and
a directly connected 12 in. diffusion pump are used to evacuate the
chamber. Background pressures of 6 × 10-5 and 5 × 10-6 torr are
typical of operation with and without the molecular beam, respectively.
The seeded carrier gas (Ar or He at 50-1000 mbar stagnation pressure)
expands continuously through a ∼100 µm pinhole into the chamber.
The laser is focused with a L ) 1 in., f ) 1000 mm, quartz lens. The
resulting fluence is typically ∼8 MW/cm2 for the excimer and ∼2 MW/
cm2 for the dye laser. The molecular beam is intersected by the focused
output of the laser 1.5 cm downstream. The light is collected by f/5
optics and directed into a 0.67 m single monochromator, where it is
dispersed by a 600 gr/mm or a 1800 gr/mm holographic grating and
detected by a liquid nitrogen cooled CCD detector (Princeton Instru-
ments, 1024 × 256). The spectral width of a pixel is ∼0.066 and
∼0.019 nm for the 600 and 1800 gr/mm gratings, respectively. Spectra
were calibrated externally by taking standard neon and mercury spectra
assuming linear dispersion of wavelength over the spectral region.
Spectra were more accurately calibrated in cases where known atomic
lines appeared.
information about the photofragmentation process and the role
of molecular collisions in photofragmentation to form metal
deposits. Surprisingly, under collision-free conditions only CuF
luminescence was observed. One goal of the present study is
to determine if metal fluorides are a common photoproduct
resulting from photolysis of metal hexafIuoroacetylacetonate
complexes. An obstacle in determining the generality of
photolytic diatomic metal fluoride production from metal
hexafluoroacetylacetonates is the general paucity of spectro-
scopic data available on the diatomic metal fluorides. For a
definitive assignment of an observed luminescence band to a
metal fluoride (or any other species for that matter), it is helpful
to use known spectroscopic band assignments. In our original
study of Cu(hfac)2 we used the detailed spectroscopic analyses
of the various band systems of CuF. Unfortunately there are
very few metal fluorides for which this level of detail is known.
Many electronic states of nickel fluoride have been assigned in
the literature.32 Because of this and because of nickel’s
proximity to copper in the periodic table, we chose to investigate
the photofragmentation of Ni(hfac)2. Similarly, CrF has been
spectroscopically characterized in the literature33-35 allowing
the possibility of definitive identification of this species if it is
a photoproduct of Cr(hfac)3. Cr(hfac)3 is also studied to
investigate the differences, if any, in the gas phase photode-
composition under deposition conditions between metal bis- and
trishexafluoroacetylacetonate species. Cr(hfac)3 is six-coordi-
nate based on an octahedral coordination structure, whereas Cu-
(hfac)2 and Ni(hfac)2 are four-coordinate planar complexes.
In our prior work on Cu(hfac)2 we proposed two mechanisms
by which metal fluoride production could occur; one was based
on metal binding to the hfac through the â carbon. Because
this type of bonding has been shown to exist in Pt(hfac)2 this
molecule is studied to investigate the possible contribution of
Pt-C bonding to the fragmentation mechanism and its possible
importance in platinum CVD. Platinum films grown from the
unfluorinated Pt(acac)2 precursor exhibit large amounts of
carbon contamination.8,9
In this paper, the luminescence that is observed under gas
phase photolytic deposition conditions is studied for Cr, Ni, and
Pt hfac compounds. This luminescence is analyzed under a
variety of conditions, including the relatively wide range of
pressures available in an evacuated gas cell and the collision-
free conditions of a molecular beam. The effects of inert buffer
gas are also studied. Features in these spectra indicate that, in
general, multiple photolysis processes occur. We identify some
simple fragments that are produced from these compounds and
discuss the differences and similarities between the compounds
studied. Detailed analysis of the spectra allows characterization
of the internal energies of the fragments to be determined. We
provide this level of analysis for the platinum photofragments.
Materials. Chromium hexafluoroacetylacetonate, nickel hexafluo-
roacetylacetonate, and platinum hexafluoroacetylacetonate were ob-
tained from Strem Chemical Company and used either without further
purification or following sublimation. Sublimation of the compounds
did not change the spectra observed.
Results
Experimental Procedures
1. Ni(hfac)2. Low-Pressure Gas Phase Luminescence.
Under 308 nm excitation and low-pressure conditions (pressure
less than or equal to the vapor pressure of the compound), the
spectrum in Figure 1a is observed. There are a multitude of
sharp peaks between 340 and 400 nm with additional sharp
peaks at 538.7, 542.1, and 547.4 nm. These peaks have a short
lifetime as illustrated in the short-time gated spectrum in Figure
1b. This spectrum is acquired with the high-voltage gate on
for 10 ns during the laser shot. Longer delay times result in
the spectrum that appears in Figure 1c. This spectrum is
acquired by delaying the gate for 100 ns and turning it on for
100 µs thereafter, thereby reducing the contributions from short-
lived species. These longer lived peaks, which are broader, are
Spectroscopic Techniques. The light source for the spectroscopic
investigations is a Lambda Physik EMG 201 MSC XeCl excimer laser
coupled with a Lambda Physik FL2001 excimer-pumped dye laser
operating with coumarin 440 or p-terphenyl as the dye. The pulse width
is 12 ns. The 308 nm pulse energy used for excitation of the CVD
precursors is varied from approximately 0.3 to 30 mJ. The dye laser
pulse energy was typically 2 mJ for 343 nm and 4 mJ for 440 nm.
(32) Dufour, C.; Hikmet, I.; Pinchemel, B. J. Mol. Spectrosc. 1994, 165,
398.
(33) Dubov, V. M.; Shenyavskaya, E. A. Opt. Spektrosk. 1987, 62, 326.
(34) Launila, O. J. Mol. Spectrosc. 1995, 169, 373.
(35) Devore, T. C.; McQuaid, M.; Gole, J. L. High Temp. Sci. 1990, 29, 1.

2882 Inorganic Chemistry, Vol. 37, No. 12, 1998
Talaga et al.
Figure 1. Emission spectra observed from 308 nm excitation of Ni-
(hfac)2 in the gas phase. (a) Spectrum observed at a pressure of no
greater than the vapor pressure of Ni(hfac)2. The triangles mark the
wavelengths of known NiF transitions as summarized in Table 2. (b)
Low-pressure gas cell spectrum of Ni(hfac)2 showing the components
which have a short (∼10 ns) lifetime. The diamonds mark luminescence
due to CH. (c) Low-pressure gas cell spectrum of Ni(hfac)2 showing
the components which have a long (>100 ns) lifetime. Note the relative
enhancement of the NiF bands. (d) Emission spectra arising from 308
nm excitation of Ni(hfac)2 in a molecular beam of Ni(hfac)2 seeded in
argon. The short verticle lines mark the wavelengths of Ni atomic
emission. Prominent Ni atomic lines are observed at 346.14, 352.41,
362.43, 542.38, and 546.07 nm.
observable at 425.0, 429.2, 435.2, 439.9, 451.6, 455.1, 467.1,
499.8, 506.0, 513.2, 521.1, 551.0, 563.2, and 571.1 nm.
Additional peaks are present at 371.0, 371.7, 387.1, 389.9, 399.1,
411.5, 422.2, 445.7, 448.1, 452.8, 461.2, 469.5, 475.0, 483.8,
486.1, 516.0, 532.8, 544.6, and 545.1 nm.
Collision-Free Luminescence from a Molecular Beam.
Under the collision-free conditions of the molecular beam and
308 nm excitation the spectral features present under the low-
pressure conditions resolve into distinct structured bands (Figure
1d). These spectra were acquired using the 600 gr/mm grating.
The sharp bands between 340 and 400 nm resolve further into
a multitude of bands each with a fwhm of approximately 10
cm-1 with some weaker broad bands underneath them at 371.0,
371.7, 387.1, 389.9, and 399.1 nm. The band system at 431
nm resolves into a series of rotational lines with a spacing of
∼28 cm-1 and band heads. The bands at 425.0, 429.2, 435.2,
439.9, 451.6, 455.1, 467.1, 499.8, 506.0, 513.2, 521.1, 551.0,
563.2, and 571.1 nm are present but remain unstructured. The
additional resolution and reduction in background allows
observation of additional bands at 365.71, 366.14, 366.90,
373.02, 377.23, 377.76, 379.19, 385.80, 388.15, 391.25, 397.12,
400.55, 447.03, 460.44, 471.52, 501.12, 503.36, 506.49, 507.97,
509.54, 511.40, 514.52, 546.07, 558.75, and 564.19 nm.
2. Cr(hfac)3. Low-Pressure Gas Phase Luminescence.
Figure 2a and b displays the luminescence spectra obtained from
the low-pressure gas cell. The spectra from the chromium
compound are substantially richer in features than those from
the nickel and platinum compounds. All features are observed
under both 308 nm excitation and 343 nm excitation. Under
Figure 2. (a) Low-pressure gas cell spectrum of Cr(hfac)3 showing
spectral features in the red and near-infrared. Assignments of the
vibronic transitions observed for the CrF A6Σ f X6Σ system are shown.
The ∆V ) -5 cluster is assigned in detail. Further details of the
assignments are in the text and in Table 1. (b) Low-pressure gas cell
spectrum of Cr(hfac)3 showing spectral features in the visible. The top
trace is 10× expansion of the lower trace to show details in the baseline.
(c) Dispersed luminescence from Cr(hfac)3 excited at 308 nm in a
molecular beam. Cr atomic lines are marked with short vertical lines.
Prominent Cr atomic lines are observed at 425.43, 427.48, 428.97,
520.45, 520.60, 520.84, 740.02, 746.24, and 894.73 nm. Various band
systems are indicated and are described in more detail in the text.
343 nm excitation, some features are more pronounced and
better signal to noise is achieved (see below).

Luminescent Photofragments of hfac Complexes
Inorganic Chemistry, Vol. 37, No. 12, 1998 2883
Table 1. Observed Peaks from CrF A6Σ f X6Σ Band Emission
and Their Assignmentsa
assignment (V,V′)
observed (cm-1
)
reported34 (cm-1
)
Figure 3. Emission spectra observed from 440 nm excitation of Pt-
(hfac)2. (a) Spectrum observed at a pressure of no greater than the vapor
pressure of Pt(hfac)2. (b) Spectrum observed using a static chamber
with Pt(hfac)2 at its vapor pressure with ∼1 bar Ar buffer.
(hfac)3 in a molecular beam under 308 nm excitation. All of
the features observed under deposition conditions (discussed
above) are also observed in the molecular beam, but many new
features appear.
Most of the numerous features between 14 000 and 10 000
cm-1 correlate peak for peak with those observed in the gas cell
experiment discussed above. Both the overall signal and the
signal to noise ratios observed in the molecular beam are lower
than those which were observed in the static gas cell. By 13 500
cm-1 the appearance of new bands and spectral congestion
obscure the features previously observed in the cell spectrum.
At 13 950 cm-1 a new narrow band appears. This feature
has a fwhm of approximately 250 cm-1. When the band is
viewed at the full resolution of the spectrum, what appears to
be noise at first glance in Figure 2c is seen to actually be a
multitude of sharp lines.36 The spacing between these lines
a The literature values for the band edges of identically assigned
peaks are also given.
In the red and near-infrared regions of the emission spectrum,
a broad vibronic feature is observed (see Figure 2a). This
feature is observed under both 343 and 308 nm excitation, with
the best signal to noise ratio obtained under 343 nm excitation.
This band system is centered at approximately 13 000 cm-1 and
is highly structured. The peaks are clustered into well-defined
groups. The spacing between each group of peaks is ap-
proximately 585 cm-1, while the spacing between individual
adjacent peaks ranges from 85 to 125 cm-1. There are at least
five well-defined groupings of peaks. This band system extends
from 15 500 to 10 000 cm-1. Toward higher energy the bands
become obscured due to spectral congestion. The dominant
features of this band system are listed in Table 1.
varies from 5 to 15 cm-1
.
In the visible region of the spectrum, a variety of sharp peaks
similar to those observed in the cell experiment are observed
with identical peak locations. In addition a variety of broader
peaks are also observed. Numerous poorly resolved bands are
observed between 17 500 and 15 500 cm-1. Identification of
clear progressions is difficult due to spectral congestion in this
region. However a spacing of 600 cm-1 between several of
the bands can be measured from the molecular beam spectrum.
There is a new band system consisting of at least six peaks
appearing at 19 000 cm-1. The band system at 23 000 cm-1
that was identified in the gas cell spectra is clearly observable
in the molecular beam spectra.
In the region around the laser excitation source at 308 nm a
number of features are present. A sharp peak at 31 830 cm-1
is observed at slightly lower energy than the laser line. In
addition, a shoulder is observed on the Rayleigh edge of the
laser line at 31 940 cm-1. At higher energy than the laser, a
vibronic progression similar to that observed in the cell spectrum
is observed. The features are separated by a spacing of
approximately 650 cm-1 just as in the cell spectrum.
3. Pt(hfac)2. Gas Cell Luminescence. Figure 3 displays
the luminescence spectra obtained from Pt(hfac)2 in the low-
pressure gas cell. The spectrum consists of many features, all
of which are observed under 308 nm excitation and 440 nm
excitation.
In the visible region of the spectrum (25 000-16 000 cm-1
)
many intense sharp (fwhm ) 10 cm-1) line features are observed
as a result of 308 nm photolysis (see Figure 2b). In addition
there are many broader features observed at 23 154, 22 498,
and 21 814 cm-1. These broader features are obscured by the
sharp line features.
At energies higher than the 308 nm laser excitation source,
a long regular vibronic progression centered at 37 000 cm-1, is
observed. The band consists of eight vibronic features but is
obscured to lower energy by the presence of the laser line. The
spacing between features is approximately 650 cm-1. The bands
have an average fwhm of approximately 300 cm-1. Features
are observed between 35 000 and 40 000 cm-1. This band
system was observed with the monochromator in second order
using the appropriate cutoff filters.
Collision-Free Luminescence from a Molecular Beam.
Figure 2c shows the luminescence spectrum observed from Cr-
(36) Hanna, S. D. Ph.D. Dissertation, UCLA, 1995; Chapter 4.

2884 Inorganic Chemistry, Vol. 37, No. 12, 1998
Talaga et al.
Table 2. Assignments of the Emission Bands of NiF Resulting
from the Photolysis of Ni(hfac)2
Because of this a rotational spectral simulation was not
attempted. Similarly, the lack of observable vibrational bands
and the corresponding lack of vibrational characterization of
NiF prevents the analysis of the internal energies of the NiF
photoproducts. Nevertheless, the observation of these states of
NiF clearly demonstrates the presence of NiF and shows that
metal fluoride production is not unique to copper in the
photofragmentation of metal hfac compounds.
CH Luminescence. Emission from the CH diatomic mol-
ecule is labeled by diamonds in Figure 1b. The unambiguous
assignment of CH comes not only from the band head positions
but also from the rotational line spacing of ∼28 cm-1 observable
in the molecular beam spectra.39 The presence of CH under
both gas cell and collision-free molecular beam conditions
implies that this product is generated directly from Ni(hfac)2.
The origin of this CH is therefore probably from the â carbon
CH on the hfac moiety.
lower state
energy
upper state
energy
transition
symmetry energy observed
symmetry
0
251
0
251
830
1574
1574
0
2Π3/2
2Π1/2
2Π3/2
23 504
23 504
22 954
22 954
22 954
23 504
22 954
19 983
19 719
19 719
19 719
19 719
19 983
19 719
2Π1/2
23 504
23 253
22 954
22 703
22 124
21 930
21 380
19 983
19 719
19 468
19 154
18 145
17 759
17 495
23 532
23 301
22 976
22 733
22 145
21 975
21 410
20 008
19 764
19 484
19 188
18 147
17 756
17 510
2Π1/2
2Π3/2
2Π1/2
2Π3/2
2
∆
2Π3/2
5/2
2Σ
2Π1/2
2Σ
2Π3/2
2Π3/2
2Π3/2
2Π1/2
Ω ) 3/2
2Σ
Ω ) 1/2
Ω ) 3/2
Ω ) 3/2
Ω ) 3/2
Ω ) 3/2
Ω ) 1/2
Ω ) 3/2
0
251
565
1574
2224
2224
Ω ) 3/2
Ω ) 3/2
Unassigned Luminescence Bands. There are several bands
which we are unable to assign as either a ligand fragment,
atomic emission, or metal-containing fragment emission. In the
low-pressure chamber, the bands at 371.0, 371.7, 387.1, 389.9,
399.1, 411.5, 422.2, 445.7, 448.1, 452.8, 461.2, 469.5, 475.0,
483.8, 486.1, 516.0, 532.8, 544.6, and 545.1 nm cannot be
assigned.
In the molecular beam, of the new peaks that appear, only
the ones at 397.1, 400.5, 447.0, 506.5, 508.0, 509.5, 511.4, and
514.5 nm cannot be assigned to Ni atomic emission. Because
photolysis produces luminescent metal fluorides, and because
NiF luminescence is prominent in the spectra, we speculate that
these unassigned bands could be due to previously unobserved
states of NiF. This gas-phase synthesis of metal fluorides may
provide a unique opportunity for further study of metal fluorides
in the gas phase.
2. Assignments from the Cr(hfac)3 Spectra. Lumines-
cence from Chromium Atoms. The intense sharp lines
observed in both the gas cell spectrum and the molecular beam
are assigned to atomic chromium emission. Chromium atomic
lines are observed under all conditions studied. Their observa-
tion confirms that chromium atoms are produced in the gas
phase due to laser photolysis.
Under 440 nm excitation and low-pressure conditions (cell
pressure less than or equal to the vapor pressure of the
compound), the spectrum in Figure 3a is observed. In the visible
region of the emission spectrum, a well-defined vibronic
progression is observed. This feature is observed under both
440 and 308 nm excitation, and under the lower energy
excitation, a higher signal to noise ratio is achieved. This band
system has its origin at approximately 18 500 cm-1. The peaks
are clustered into well-defined groups proceeding to lower
energy with the most intense members at 540, 572, 608, 649,
and 694 nm. The spacing between each group of peaks is
approximately 1050 cm-1, while the spacing between individual
adjacent peaks in a group is approximately 230 cm-1. To higher
energy from the origin there are many peaks in the region from
445 to 530 nm with prominent peaks at 517, 497, and 477 nm.
Effect of Inert Buffer Gas. Under “high”-pressure condi-
tions with added Ar buffer gas to a total pressure of ∼1 bar,
similar spectra are observed as illustrated in Figure 3b.
However, the peaks to high energy show reduced intensity as
is apparent by comparing Figures 3a and 3b. In particular the
bands between 445 and 530 nm show different relative
intensities. Also, the smaller peaks in the main progression to
lower energy show reduced intensity.
Luminescence from Chromium Fluoride. The highly
structured luminescence observed in the high-pressure gas cell
centered at 850 nm under 343 nm excitations is assigned as the
transition from the A6Σ excited state of CrF to the X6Σ ground
state. Previous studies on chromium flux heated in the presence
of F2 or CaF revealed a similar spectrum.34 The E0 of this
emission is reported to occur at 9624 cm-1. The stretching
frequency in the excited state is 581 cm-1, while that of the
ground state is 661 cm-1. Hot bands through ∆n ) -9 are
readily identifiable in the cell spectrum shown in Figure 2a. A
summary of the observed features with vibronic assignments
for the A6Σ f X6Σ transition of CrF is given in Table 1. To
higher energy than the ∆n ) -6 band the structure begins to
be washed out due to spectral congestion. The B6Σ state of
CrF is at 16 864 cm-l and may be contributing to the spectral
congestion in this region. In the molecular beam under 308
nm excitation, the A6Σ f X6Σ emission is observed but not as
strongly as in the cell under 343 nm excitation. This is
explained by the fact that the D6Σ state of CrF absorbs strongly
at 308 nm. Thus CrF produced in the molecular beam may be
further fragmenting due to absorption at the excitation wave-
length. The molecular beam produces molecules which are
Discussion
1. Assignments from the Ni(hfac)2 Spectra. Lumines-
cence from Nickel Atoms. The intense sharp lines observed
in both the gas cell spectrum and the molecular beam are
assignable as nickel atomic lines. The bands marked with short
vertical lines in Figure 1d are assigned to nickel atomic
emission.37,38 The majority of the new bands appearing in the
molecular beam spectra are attributable to Ni atomic lines.
Nickel atomic lines are observed under all conditions studied.
The observation of nickel atomic lines under CVD conditions
is not unexpected for Ni(hfac)2. Their observation, however,
does confirm that nickel atoms are produced in the gas phase
due to laser photolysis.
Luminescence from Nickel Fluoride. The bands marked
with triangles in Figure 1a are assigned as luminescence from
NiF. The state assignments are summarized in Table 2. The
bands due to NiF do not have well-resolved rotational band
profiles beyond one case of a doubled band head at 451.6 nm.
(37) Wiese, W. L.; Martin, G. A. WaVelength and Transition Probabilities
for Atoms and Atomic Ions; U.S. Government Printing Office:
Washington, DC, 1980.
(38) Striganov, A. R.; Sventitskii, N. S. Tables of Spectral Lines of Neutral
and Ionized Atoms; IFI/Plenum: New York, 1968.
(39) Pearse, R. W. B., Gaydon, A. G. The Identification of Molecular
Spectra, 3rd ed.; Chapman and Hall: London, 1976.

Luminescent Photofragments of hfac Complexes
Inorganic Chemistry, Vol. 37, No. 12, 1998 2885
Table 3. Observed Electronic States of CrF Resulting from the
33-35
Photolysis of Cr(hfac)3
state
Te
ωe
ωe xe
characteristic bands (cm-1
)
X6Σ
A6Σ
B6Σ
?
D
?
0
664
581
605
9 624
16 864
23 154
2
9 916, 10 490, 11 064
16 864, 16 203, 16 137
23 154, 22 498, 21 814
31 777, 31 180, 30 051
34 123, 35 023, 35 700
∼600
650
colder initially and which would be expected to have slower
rates of internal conversion. In addition the lack of collisions
prevents relaxation to the lowest excited state (A6Σ) of CrF.
These factors all contribute to the molecular beam spectra being
far richer in features and correspondingly more congested.
The B6Σ f X6Σ luminescence from CrF is clearly observed
in the molecular beam spectrum. Identification of many of the
hot bands is difficult due to spectral congestion in this region.
The B6Σ f X6Σ state has an E0 of 16 864 cm-1 and was
reported previously. A vibrational frequency of approximately
605 cm-1 was also reported which matches well with the
Figure 4. Calculated low-pressure emission spectrum of PtC. The insert
shows the vibrational populations used in the calculation and the
Boltzmann populations for T ) 868 K.
tion implies that the formation of this photoproduct is most likely
a unimolecular process. It is possible that clustering occurs in
the molecular beam due to the cold temperature and that such
clusters could provide direct interactions between otherwise
nonbonded metal complexes. However, we also observe the
metal fluorides in high-temperature gas phase experiments.
Therefore we conclude that the appearance MF in collision free
conditions is due to an intramolecular process and not a
polymolecular complex.
The metal-containing diatomic fragments are not formed by
generation of a plasma followed by recombination of atoms or
ions. A plasma would produce metal ions; we only observe
emission from neutral metal atoms in both the evacuated gas
cell experiments and the molecular beam experiments. A re-
combination mechanism cannot explain the specificity of our
observations of MF formation from Cu, Ni, and Cr hexafluo-
roacetonate complexes and PtC from the Pt hexafluoroacetonate
complex. A nonspecific physical (as opposed to chemical)
mechanism such as plasma formation would be expected to
produce many different metal-containing diatomics including
metal oxides; we do not see species such as CrO or NiO.
Finally, the results of the experiments in the supersonic
expansion molecular beam, where there are no collisions, show
that metal fluorides do not arise from a plasma. Therefore we
conclude that the appearance of MF under the beam conditions
does not result from a plasma, and that the appearance of MF
and MC in the gas cell is not dominated by the cooling phase
of a plasma. Rather, an intramolecular chemical mechanism
must be dominant.
The production of metal monofluorides requires that a large
amount of internal motion occur in the molecule such that the
fluorine may come in contact with the metal. We propose that
intramolecular fluorine atom abstraction by the metal might
occur by either or both of two simple mechanisms. The first
mechanism involves metal migration as illustrated in the bottom
pathway in Figure 5. Alternate binding modes of metals to
â-diketonates are known,40-42 particularly for platinum,40,41
including binding through the π system (η3) of the acetylace-
tonyl moiety or binding in a sigma manner (η1) to the â carbon
of the acetylacetonyl moiety. If the metal migrates onto the π
system or is bound through the â carbon, it is much closer to
observed spacing of ∼600 cm-1
.
Bands observed in the blue region of the spectrum have a
spacing characteristic of CrF. No electronic state for CrF in
this region has been previously reported; however the charac-
teristic spacing is a strong indication that this a previously
unknown state of CrF. The E0 appears to occur at 23 154 cm-1
,
but spectral congestion could prevent identification of other
bands. The 0,1 band occurs at 22 498 cm-1. This gives rise
to an ground state vibrational energy of 656 cm-1 which is
within experimental error of the 664 cm-1 reported for the
ground state vibrational frequency of CrF. Likewise the 0,3
band is observed at 21 814 cm-1
.
Three bands on the Rayleigh edge of the laser are observed
at 31 777, 31 180, and 30 051 cm-1. While there are not enough
features to allow for conclusive identification, these bands occur
in the region expected for the D6Σ f X6Σ transition of CrF.
Unassigned Luminescence. While many of the features
observed in the gas phase luminescence and molecular beam
spectra can be assigned, there are still many features that cannot
be assigned. Many of these features are difficult to resolve due
to spectral congestion or low signal to noise ratios. Others,
such as the band systems centered at 37 000 and 13 950 cm-1
,
are well resolved but have no literature precedent. The
simplicity of these features suggests that they may be due to a
diatomic species, perhaps CrF, CrC, or CrO, but no further
assignment can be made. A summary of CrF band systems
observed and suspected CrF band systems is in Table 3.
3. Assignments from the Pt(hfac)3 Spectra. Lumines-
cence from Platinum Carbide. All major luminescence
features present in Figure 3 can be assigned to the A f X
transition of PtC. There are no features present that we can
assign to either Pt atoms or PtF. Unfortunately, there is no
literature precedent for PtF luminescence; therefore, we cannot
be certain that PtF is not present. It is possible that Pt atoms
may be present in nonluminescent states. Spectral congestion
in the hot band region from 445 to 530 nm may be obscuring
some small features; however, all clearly identifiable peaks are
assignable to PtC luminescence, as is clear from the calculation
in Figure 4. The effect of inert buffer gas is attributable to
collisional cooling of the PtC photofragment.
4. Mechanisms of Fragmentation for Metal hfac Com-
pounds. Formation of MF and M. The emission of the
diatomic metal fluorides CuF, NiF, and CrF occurs under the
collision free conditions of the molecular beam. This observa-
(40) Behnke, G. T.; Nakamoto, K. Inorg. Chem. 1967, 6, 440.
(41) Lewis, F. D.; Miller, A. M.; Salvi, G. D. Inorg. Chem. 1995, 34, 3173.
(42) Kimiya, S.; Kochi, J. K. J. Am. Chem. Soc. 1977, 99, 11.

2886 Inorganic Chemistry, Vol. 37, No. 12, 1998
Talaga et al.
Figure 5. Proposed intramolecular rearrangements that explain the observed photofragmentation patterns of the metal-hfac compounds studied.
Table 4. Spectroscopic Parameters Used in the Intensity Fits To
the fluorines on the CF3 groups than it would be in the standard
bidentate binding mode. This mechanism is discussed in more
Determine the Vibrational and Rotational Temperatures of PtC45
state
Te
ωe
ωe xe
Tvib
detail in the following section.
A second mechanism which seems more likely for Cu, Ni,
and Cr is illustrated in the top pathway in Figure 5. The single
photon absorption at 308 nm has been assigned as a π f π*
ligand-centered transition.43,44 One photon excites the π f π*
system on the ligand and another photon excites an LMCT,
breaking a M-O bond. The resulting half-free ligand is bound
through one oxygen. The unbound side can then rotate about
the (now) formally C-C single bond, placing the CF3 group in
direct contact with the metal. The metal abstracts the fluorine,
and the resulting MF chemiluminesces. This mechanism
involves the least motion of the atoms in the molecule. If a
collision occurs before the abstraction takes place, free copper
can be produced. However, this collision is not required for
production of luminescent Ni or Cr.
The production of luminescent metal atoms may arise directly
from the M(hfac)n fragmentation process, or it may be a result
of absorption of the excitation laser light by either atoms in
nonluminescent states or by MF which then fragments to
produce M* + F. One note on that regard is that laser-induced
fluorescence excitation spectra of Cr(hfac)3 show atomic line
resonances at 373.08 and 373.20 nm from ground state Cr.45
This result indicates that Cr atoms are being produced in the
ground state upon multiphoton fragmentation of the Cr(hfac)3
molecule.
X
A
0
1051.18
818.8
4.87
5.5
18 510
868
transition of PtC. The adjustable parameters were the popula-
tions of vibrational levels in the excited states. The vibrational
state populations are shown as the insert in Figure 4. The
parameters used for the fits to the various states of PtC are
summarized in Table 4.
Boltzmann fits to the vibrational populations from the
luminescence studies were carried out to determine the vibra-
tional temperature of PtC. The fit shows that the ejected PtC
is vibrationally hot with a calculated temperature on the order
of 1000 K. The ratio of the A state V ) 0 population to the V
) 1 population implies a temperature of 870 K. However, the
populations of the V ) 2, 3, 4, 5 states are much higher than
predicted for a Boltzmann population at this temperature.
Selective population of the higher levels may occur. The PtC
fragment produced from the Pt(hfac)2 photofragmentation is
vibrationally hot, as is the case for CuF from Cu(hfac)2. The
vibrational “temperature” is substantially lower than those
observed for CuF (∼1700 K). The amount of energy disposed
of in electronic and vibrational excitation of the PtC fragment
is approximately 30 000 cm-1
.
The alternative η1 binding mode that is specifically known
for â-diketonate complexes of platinum is important in the
production of PtC. It is a simple matter to eliminate relatively
stable fragments from an already PtC-bonded species to produce
the final PtC diatomic as shown in Figure 5. It cannot be ruled
out that binding through the â carbon of the acetylacetonyl
moiety could be involved in the mechanism for the production
of metal fluorides. The different patterns (carbides versus
fluorides) observed in different metals could be due to the
differences in metal binding strength to the â carbon. Platinum
may bind particularly well to the â carbon, which would explain
why this intermediate has been identified previously. The
production of PtC has a large driving force; the bond strength
of PtC is 50 830 cm-1 (145.3 kcal/mol).46 This strong metal
carbon bond may substantially weaken the carbon-carbon
bonds, which results in the mechanism which produces PtC.
However, if the metal-â carbon bond is weak, then perhaps
fluorine abstraction could dominate.
Formation of PtC. Perhaps the most intriguing and unex-
pected result of these studies is that of the platinum carbide
production. In our original copper hfac paper we raised the
possibility that an alternative binding mode, a metal-carbon
η1 bond that has been observed for Pt(hfac)2, was involved in
the photofragmentation mechanism.1 The data and calculations
discussed below prove that PtC is an important photofragmen-
tation product.
The observed emission spectrum in Figure 3 is in excellent
agreement with the known A f X emission of PtC. A fit to
the spectrum was obtained by using literature values26,27 for
diatomic molecular parameters to calculate the excited and
ground state vibrational energy levels. The vibrational line
strengths were calculated using wave packet propagation and
the time dependent theory of electronic spectroscopy. Figure
4 shows the fit to the low-pressure spectrum of the A f X
(43) Wexler, D.; Zink, J. I. Inorg. Chem. 1996, 35, 4064.
(44) Fackler, J. P.; Cotton, F. A.; Barnum, D. W. Inorg. Chem. 1963, 2,
97.
(46) Donne´es Spectroscopiques RelatiVes aux Mole´cules Diatomiques, 17th
(45) D. S. Talaga. Unpublished work.
ed.; Rosen, B., Ed.; Pergamon Press, Inc: New York, 1970; pp 515.

Luminescent Photofragments of hfac Complexes
Inorganic Chemistry, Vol. 37, No. 12, 1998 2887
Conclusions
The spectra from the Ni and Cr compounds contain many
atomic lines in addition to the metal fluoride bands. In our
earlier study of Cu(hfac)2 atomic lines were observed in the
presence of collisions but copper fluoride was found in the
absence of collisions. Collisions are not necessary for atomic
metal emission in the cases of Ni(hfac)2 and Cr(hfac)3.
Materials deposited by the laser photochemistry of these
complexes would be expected to show impurities. Photochemi-
cally derived deposits of metals from Cu(hfac)2, Cr(hfac)3, or
Ni(hfac)2 are expected to show fluorine incorporation due to
the gas phase component of the deposition process. Similarly
the photochemical production of platinum metal from Pt(hfac)2
is expected to show carbon contamination6 because of the
production of PtC in the gas phase portion of the deposition
process. In all cases, however, the contributions of surface
reactions will also be important.
Bare metal atoms and diatomic metal fluorides are common
photofragments from metal hfac compounds. The spectroscopic
studies of UV photofragmentation of M(hfac)n under a range
of gas phase conditions with no surface interactions imply that
MF is formed by intramolecular fluorine abstraction by the
metal. The MF molecule is both rotationally and vibrationally
hot (T ∼ 1700 K). Although the one-photon transition in the
M(hfac)n complexes at 308 nm has been assigned as ligand-
centered π f π*, there is likely some metal contribution to the
orbitals involved in the multiphoton transition. Collisions are
required for luminescent metal production from Cu(hfac)2 but
not from Cr(hfac)3 or Ni(hfac)2.
The production of PtC as a luminescent photofragment from
Pt(hfac)2 implies that the type of coordination of the ligand to
the metal is very important in determining the photoproduct
observed spectroscopically. Pt(hfac)2 is known to bind to
the â carbon of the ligand, and Pt is the only metal for which
metal carbide luminescence during photofragmentation is ob-
served.
Acknowledgment. This work was supported by a grant from
the National Science Foundation (CHE-9509275).
IC971340X