Electronic spectroscopy of the titanium centers in titanium silicalite

Article abstract of DOI:10.1021/jp002645r

Studies of the electronic structure of titanium centers in titanium silicalite (TS-1) catalytic materials were carried out using electronic absorption and emission spectroscopy. A long-lived phosphorescent excited state with an emission maximum at 490 nm in the near UV was unambiguously assigned to the titanium. Resolved in the emission envelope was vibronic structure in 965 cm^-1 mode, which corresponds to the Si-O-Ti stretching mode in TS-1. The lowest energy excited state is significantly lower in energy than was previously suggested by diffuse reflectance absorption spectroscopy. Emission excitation spectra indicate that, contrary to previous assertions, there are electronic transitions throughout the spectral region from ～23000 to 48000 cm^-1. These observations bring into question long-standing structural arguments for the coordination of titanium in the silicalite lattice that have been made using electronic spectroscopy.

Full text of DOI:10.1021/jp002645r

J. Phys. Chem. B 2001, 105, 2687-2693
2687
Electronic Spectroscopy of the Titanium Centers in Titanium Silicalite
Allison S. Soult, Duke D. Poore´, Elizabeth I. Mayo, and A. E. Stiegman*
Department of Chemistry and The Materials Research and Technology Center (MARTECH),
Florida State UniVersity, Tallahassee, Florida 32306
ReceiVed: July 25, 2000; In Final Form: NoVember 15, 2000
Studies of the electronic structure of titanium centers in titanium silicalite (TS-1) catalytic materials were
carried out using electronic absorption and emission spectroscopy. A long-lived phosphorescent excited state
with an emission maximum at 490 nm in the near UV was unambiguously assigned to the titanium. Resolved
in the emission envelope was vibronic structure in 965 cm-1 mode, which corresponds to the Si-O-Ti
stretching mode in TS-1. The lowest energy excited state is significantly lower in energy than was previously
suggested by diffuse reflectance absorption spectroscopy. Emission excitation spectra indicate that, contrary
to previous assertions, there are electronic transitions throughout the spectral region from ∼23000 to 48000
cm-1. These observations bring into question long-standing structural arguments for the coordination of titanium
in the silicalite lattice that have been made using electronic spectroscopy.
Introduction
to 35 °C for 30 min while being stirred and then cooled to 0
°C. To the cooled solution was added 100 g of a 1.0 M
Binary titania-silica oxides have elicited a great deal of
interest due to their diverse and useful properties. For example,
amorphous titania-silica glass, with titania concentrations of
∼7-8%, is produced commercially for its extremely low
coefficient of thermal expansion.1 Conversely, crystalline titania
silicalite zeolites (TS-1) have been found to be highly active
and selective catalysts for, among other reactions, the oxidation
of phenol to catechol and hydroquinone.2,3 The properties of
these materials are due to the presence of titanium sites dispersed
in the silica matrix. The nature of these sites and how they give
rise to the useful properties exhibited by these materials is still
unclear. For this reason, a number of studies utilizing a variety
of spectroscopic techniques have been carried out to elucidate
the coordination geometry and the electronic and photophysical
properties of the titanium active sites as they reside in the matrix.
Specifically, in the case of TS-1, electronic spectroscopy,
collected by diffuse reflectance techniques or through photo-
emission, has been used to establish the coordination geometry
of the titanium site, identify the number of active sites present,
and identify the presence of extraframework titania in the final
material.2-4
tetrapropylammonium hydroxide solution (Aldrich) at 0 °C
dropwise over 2 h. The mixture was then stirred at 80-90 °C
for 3-5 h. The mixture was returned to its original volume by
the addition of deionized water (18.0 mΩ, Barnsted E-Pure
system) and then transferred to a clean, Teflon-lined autoclave,
which was heated to 175 °C for 3 days. The resulting solid was
filtered, rinsed thoroughly with distilled water, and calcined at
500 °C to remove the organic template.
Characterization. X-ray diffraction (Siemens D500 θ-2θ
diffractometer, Ni-filtered Cu radiation (KR)) confirmed that
the materials produced were authentic crystalline samples of
silicalite and TS-1 by the observation of a characteristic
diffraction pattern.5 The exact content of titanium in each
sample, expressed as a mole percent ([Ti]/([Si] + [Ti]) × 100),
was determined by elemental analysis. Samples referred to in
the text as 0.5%, 1.5%, and 2.0% titanium had exact concentra-
tions of 0.52, 1.53, and 1.97 mol % titanium, respectively. To
assess the purity of the final products, the materials were
analyzed by inductively coupled plasma mass spectroscopy
(ICP-MS). The instrumentation was a Finnigan MAT Element
ICP-MS with a Glass Expansion MicroMist nebulizer. The
nebulizer gas flow (Ar) was 1.06 L/min. The cooling gas flow
was 13 L/min, and the ICP power was 1300 W. Results: Al,
Ga, In, Ge, Pb, Sc, Y, V, Nb, Co, <1 ppb; Na, Li, Sn,Te, Zr,
Cr, Ni, Cu, Zn, <50 ppb.
Electronic Spectroscopy. Emission spectroscopy was carried
out on a Spex Fluorolog II equipped with 0.22 m double
monochromators (Spex 1680) and a 450 W Hg/Xe lamp. Front
face (22.5°) collection was used to collect both emission and
emission excitation spectra. Cutoff filters were used to suppress
second-order excitation lines. It was determined that reproduc-
ible emission spectra were attained only after calcination of the
samples at 750 °C under a flow of oxygen (UHP grade) and
that it was imperative that they not experience ambient
laboratory conditions again prior to data collection. As a result,
We report here a detailed study of the electronic spectroscopy
of the titanium centers in TS-1, using both emission and diffuse
reflectance electronic spectroscopies. The results of this study
yield insights into the electronic structure and excited-state
properties of the reactive titanium site. They also provide a
quantitative assessment and a critical reappraisal of the structural
arguments for tetrahedral coordination of the titanium thought
to be supported by electronic spectroscopy.
Experimental Section
Synthesis. Silicalite and TS-1 were made by published
procedures.5,6 In a typical preparation, 5.0 g of titanium
isopropoxide (Aldrich, 99.999%) was added dropwise to 48.4
g of tetraethyl orthosilicate (Aldrich, 99.999%) under vigorous
stirring and a constant flow of nitrogen. The solution was heated
10.1021/jp002645r CCC: $20.00 © 2001 American Chemical Society
Published on Web 03/15/2001

2688 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Soult et al.
they were taken from a tube furnace directly into an antechamber
under a dry N2 purge, where they were mounted in an APD
model DE-202 cryostat, shrouded, and evacuated. The samples
were carefully aligned to maximize intensity in the detector prior
to collection of spectra to ensure the best reproducibility of
emission intensities. All reported spectra were corrected for the
lamp profile and the detector response. Spectra reported in
wavenumber units were corrected in the standard way for the
band-pass variability in spectra collected at fixed wavelength
resolution.7
Vibronic Analysis. Vibrational modes taken from the λmax
of the resolved vibronic bands in the emission spectrum were
determined from a spectrum collected at 12 K with 350 nm
excitation and 0.25 mm slits. The error in each mode was
estimated as that of the instrument resolution, which has a
dispersion of 1.70 nm/mm.8 At the slit width used, this yields
a resolution of 0.85 nm or (17 cm-1 at 20000 cm-1. Since the
reported vibrational mode is the difference between two resolved
peaks, application of the propagation of error to this difference
yields an error of (24 cm-1 for the experimentally determined
mode.
Franck-Condon analysis was carried out using a published
procedure.9 A reduced mass of 12 amu for a Ti-O asymmetric
stretch was assumed, the ground- and excited-state vibrational
frequencies were taken to be the same (νg ) νe), and a frequency
of 965 cm-1 was found to give the best agreement with the
data. The normal coordinate change was varied to best reproduce
the intensity distribution of the experimentally observed vibronic
progression.
Figure 1. Emission spectra of silicalite (340 nm excitation) and 0.5%
TS-1 (350 nm excitation) at (A) room temperature, (B) 77 K, and (C)
12 K.
Results
Luminescence Properties of Silicalite and Titanium Sili-
calite. The room temperature emission spectra of silicalite and
TS-1 (0.5 mol % Ti) are shown in Figure 1A. The pure silicalite
sample shows a non-Gaussian emission band with a maximum
centered around 417 nm. The band shape is excitation-
wavelength-dependent, suggesting that it contains emissions
from more than one emitting site. By contrast, TS-1 shows two
well-resolved peaks at 422 and 499 nm. Comparison with the
control spectrum clearly indicates that the 422 nm band
corresponds directly to the intrinsic emission from the silicalite
matrix, which suggests that the low-energy emission (499 nm)
is particular to TS-1 and, hence, may originate specifically from
titanium. As would be expected, the spectra become more
intense and better resolved as the temperature decreases (Figure
1). At 77 and 12 K, shoulders at 467 and 545 nm are resolved
in the silicalite spectrum along with the sharp peak at 422 nm.
For TS-1, the 499 nm band shows evidence of a vibrational
progression at 77 K (Figure 1B), which becomes well resolved
at 12 K (Figure 1C).
Diffuse Reflectance UV-Vis Spectroscopy. The diffuse
reflectance UV-vis spectra were collected on a Perkin-Elmer
Lambda 900 UV-vis-near-IR spectrophotometer equipped
with a 160 mm integrating sphere. The samples were taken from
a 500 °C tube furnace into a purged antechamber that also ported
to the sample compartment of the N2-purged spectrometer. The
spectra were collected as total reflectance spectra and displayed
in Kubelka-Munk units.10
Time-Resolved Emission. Fluorescence was induced using
355 nm excitation pulses produced from a frequency-tripled
DCR-2a 10 Hz pulsed Nd:YAG laser system with a 9 ns pulse
time width and a 8 mm physical diameter. A Jarrel-Ash 0.25 m
(f/3.6) Ebert monochromator, model 82-410 (Scientific Mea-
surement Systems, Inc.), was the dispersive element prior to
the detector, which is a Hamamatsu R928 extended red high-
sensitivity side-on-type photomultiplier tube with all dynodes
employed. The signal from the PMT was coupled to the
oscilloscope with 5.0 kΩ load resistor. A LeCroy model 9410
dual-channel 150 MHz digital oscilloscope was used in the
collection and handling of data.
The dependence of the emission spectra of TS-1 on titanium
concentration is shown in Figure 2. Two things are evident from
this series of spectra. The first is that, consistent with emission
from titanium, the intensity of the 499 nm peak increases relative
to the 417 nm band as the amount of titanium increases. Second,
the vibrational structure observed at low temperature is resolved
only in the lower concentrations of titanium.
Data analysis was carried out on the emission decays,
collected at 274 and 77 K over a range of monitoring
wavelengths from 450 to 600 nm. Fits of the overall emission
decay curve were carried out using a monoexponential (I/I0 )
ae-t/τ ) and a biexponential (I/I0 ) ae-t/τ + be-t/τ′) decay
function, where I is the intensity, t is time, and τ is the lifetime.
Decays were fit starting from 5% after the initial spike to avoid
convolution with the exciting light. The monoexponential fits
were uniformly poor (R2 e 0.981), while fits to the biexponential
functions were considerably better with R2 g 0.997 and 0.998
obtained for the 274 and 77 K data, respectively. Lifetimes
collected at 77 K were significantly better due to the increased
intensity at that temperature, which yielded better signal-to-
noise ratios.
Analysis of the Vibrational Structure in the Luminescence
Peak of TS-1. Figure 3 shows the vibrationally resolved 12 K
luminescence spectrum of TS-1 (0.5% Ti) with the frequencies
of the resolved modes indicated. Energy differences between
each mode in the progression, which correspond to the frequency
of the active vibration, were calculated, and an average value
of 966 ((24) cm-1 was determined. The infrared spectrum of
silicalite shows intense vibrations at 1227, 1105, and 805 cm-1
,
corresponding to modes of the silica lattice. As titanium is added
to the framework to form TS-1, a peak unique to this material

Electronic Spectroscopy of the Ti Centers in TS-1
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2689
Photophysical Properties of TS-1. The unambiguous de-
termination of the luminescence lifetime of the titanium emission
in TS-1 is complicated by the presence of the overlapping
silicalite emission, which contributes to the overall spectrum.
Consistent with this, the time-resolved luminescence decays of
TS-1 (355 nm excitation, 0.5 mol %) were poorly fit by a
monoexponential decay function while a biexponential function
yielded significantly better fits as indicated by the R factor and
the residuals.13 Emission lifetimes for both components are very
long, indicating that phosphorescence is occurring from both
the titanium and the silicalite. Decay curves were recorded over
the emission envelope from 450 to 600 nm. The relative
contribution of the two lifetimes (obtained from the preexpo-
nential factors of the biexponential fit) to the total emission
decay varied systematically over the emission envelope (Table
1). The short-lived component (τ1) becomes the dominant
contribution as the wavelength increases until, at 600 nm, it
accounts for about 90% of the total decay. On the basis of this
observed trend, we assign the short-lived component to the
titanium emission and the longer-lived component to the intrinsic
silicalite background emission.
Lifetimes collected at 77 K, which are better determined due
to the higher intensity of the emission, indicate that the titanium
emission (τ1) is relatively constant over the emission envelope,
giving an average value of 7.48 ( 0.59 ms while the long
component shows significant variation ((20%). This is con-
sistent with our observation that the silicalite emission appears
to contain multiple emitting species.
Figure 2. Emission spectra for TS-1 samples containing 0.5, 1.5, and
2 mol % Ti (77 K, 350 nm excitation).
Electronic Spectroscopy of TS-1. The emission excitation
spectrum for TS-1 (0.5% Ti, 12 K), monitored on the red edge
of the titanium emission at 600 nm, is shown in Figure 4. The
onset of significant intensity occurs at approximately 26700
cm-1 and then increases gradually to 35100 cm-1 with a
resolved shoulder at 31228 cm-1. Above 35000 cm-1 the
intensity increases sharply to the limits of our excitation range
at 40000 cm-1. Importantly, there is measurable intensity over
the frequency range from 26000 to 40000 cm-1, indicating that
the titanium is being electronically excited throughout this range.
Figure 5 shows the electronic absorption spectra, collected as
diffuse reflection spectra and plotted in Kubelka-Munk units,
for silicalite and for TS-1 containing 0.5% and 1.5% titanium.
The spectrum for pure silicalite shows a single strong absorption
with a λmax at 46657 cm-1. This spectrum is qualitatively similar
to that of TS-1 containing 0.5 mol % titanium, which has a
single absorption maximum at 47188 cm-1. In both cases there
is no measurable absorption below ∼33000 cm-1. At 1.5 mol
% titanium, however, detectable spectral intensity begins to
appear below 37000 cm-1. The onset of absorption occurs at
Figure 3. Emission spectra of 0.5% TS-1 (s) and silicalite (---) at 12
K (350 nm excitation). Vibrational frequency is the average of all peaks
except 22942 cm-1, which is poorly resolved. Spectra were not scaled.
The relative intensities displayed are those recorded.
∼27000 cm-1 and increases gradually until, at ∼33500 cm-1
,
it increases sharply. More importantly, the emission excitation
spectrum reproduces well the absorption spectrum in the range
emerges at ∼960 cm-1. This spectral band has been reported
previously and, in fact, has been commented on rather exten-
sively.3 It has been assigned to a titanium-silica mode, though
more recent arguments have suggested that it is predominantly
a silica (SiO-) mode “perturbed” by the presence of a titanium-
(IV) ion.11 Not observable in the infrared spectrum due to
interference from the silicalite, but seen in the Raman spectrum,
between 25000 and 40000 cm-1
.
Discussion
Assignment of the Titanium Emission in TS-1. As indicated
by the data, the luminescence spectrum of TS-1 consists of two
distinct emitting regions: a sharp, high-energy feature centered
at 422 nm and a broad low-energy band at 499 nm. The high-
energy band can be clearly assigned to intrinsic emission from
silicalite itself. The exact origin of the silicalite luminescence
is unclear, although the excitation dependence of the band
envelope suggests that several emitting species are present.
Under the processing conditions employed, this emission is
unlikely to contain a significant contribution from adsorbed
is a titania-related mode at ∼1100 cm-1 12
. This peak, however,
is much too high in frequency to be a reasonable candidate for
the resolved progression. The absence of any other bands in
the 850-1050 cm-1 region of the vibrational spectrum, coupled
with the good agreement in energy between the vibronic
progression in the emission spectrum and the infrared spectrum,
suggests that it can be assigned with considerable certainty to
the titania-silica mode.

2690 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Soult et al.
TABLE 1: Emission Liftimes for TS-1
274 K
77 K
monitoring
wavelength (nm)
a
b
a
b
a
b
a
b
τ1
%
τ2
%
τ1
%
τ2
7.51
19.0
%
450
500
550
600
0.0788
0.0532
0.0773
75
82
89
88
0.377
0.267
0.464
15
18
11
7.49
7.00
8.13
8.00
66
84
93
94
34
16
7
48.0
62.0
0.0867
0.487
12
6
average
0.0816 ( 0.021
0.399 ( 0.010
7.66 ( 0.52
34.1 ( 25.2
a Lifetimes in milliseconds. b Percents represent the contribution to the total decay for each component of the biexponential fit: % ) (a/(a + b))
× 100 and (b/(a + b)) × 100 from I/I0 ) ae-t/τ1 + be-t/τ2
.
electronic structure and energetics of the reactive sites in these
materials. It also allows a critical and quantitative assessment
to be made of prior assignments of the structure of the titanium
sites in the silicalite lattice based on diffuse reflectance electronic
spectroscopy.
There have been two previous reports of luminescence from
titanium in TS-1. A note by Ichihashi et. al. reported a single
broad emission centered at 455 nm for a sample containing 2%
titanium.14 This spectrum appears, on the basis of our data and
analysis, to be an unresolved average between the titanium
emission and the silicalite background. A more complete study
by Lamberti et. al. on TS-1 (2.03 wt % Ti, collected at room
temperature) reported two emissions at 430 and 495 nm.4 While
the positions of these bands are in close agreement with our
observations, they were assigned to two distinct titanium
emissions arising from different sites in the silicalite lattice. As
we have shown, a careful comparison of the emission data for
TS-1 and a titanium-free control (Figure 1) reveals that the high-
energy peak is inherent to the silicalite and only the low-energy
band is attributable to titanium emission. Lamberti et al.
disregarded the contribution from the silicalite emission to the
luminescence of TS-1 because they observed it to be an order
of magnitude weaker than the TS-1 emission. It is well-known
that reproducible intensities are extremely difficult to achieve
in emission spectra collected from microcrystalline solids.15
Optical noise caused by random fluctuations in the optical
transfer from the scattering media can lead to large variations
in the intensity from sample to sample. As can be seen from
the overlay in Figure 3, with careful optical alignment, the
intensity, position, and shape of this band is completely
superimposable on the high-energy emission in TS-1, proving
that it contributes significantly to the overall spectrum.16
Figure 4. Emission (350 nm excitation) and emission excitation spectra
(monitored at 600 nm) of 0.5% TS-1 measured at 12 K.
While the assignments made in this previous luminescence
study are incorrect, the question of how many different titanium
sites occur in TS-1 is an important one. There are crystallo-
graphically distinct positions for which titanium can potentially
substitute for silicon in the lattice. In addition to the lattice sites,
there is also the possibility of extraframework and surface
species, though recent NMR studies have indicated that most
of the titanium resides in lattice sites.17 If all the sites are
emissive and reside in very different coordination environments,
then several distinct emission peaks would be expected.
Conversely, if the coordination geometries of the emitting sites
are very similar, then a single broad, excitation-dependent
emission peak might be expected. For the concentration range
of titanium investigated, no titanium emission, apart from the
499 nm band, is observed, with all other spectral features being
accounted for by the silicalite background. Furthermore, while
the intensities of the silicalite background and the titanium
emission vary dramatically with respect to each other as a
function of excitation wavelength, the 499 nm titanium emission
band is itself quite excitation invariant over an excitation range
from 280 to 350 nm. In short, notwithstanding the sensitivity
Figure 5. Electronic spectra collected as diffuse reflectance spectra
and presented in Kubelka-Munk units for silicalite and 0.5% and 1.5%
Ti TS-1.
organic species and, therefore, probably arises from trace metal
impurities or from naturally occurring defect sites in the silicalite
lattice. The low-energy emission (499 nm) can be assigned with
certainty as emanating from the titanium. This assignment is
supported by the absence of this band in pure silicalite, the
dependence of its relative intensity on titanium concentration,
and the observation of vibronic structure in a mode assignable
to a Si-O-Ti stretch. The observation and assignment of the
titanium emission is important for several reasons. On a
fundamental level, it increases our understanding of the

Electronic Spectroscopy of the Ti Centers in TS-1
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2691
cm-1 mode originating from each of these estimated origins.
Regardless of the origin, very good agreement is obtained
between the observed and calculated progressions, further
supporting our vibronic assignment. As can be seen from the
spectrum, progressions generated at both origins give a good
reproduction of the intensity distribution of the unobscured part
of the spectrum, though better agreement is obtained with the
low-energy edge of the emission from the progression origi-
nating at 22942 cm-1 18
. The normal coordinate change required
to generate either of these progressions is reasonable, with values
of 0.17 and 0.19 Å used for the 22942 and 23907 cm-1 origins,
respectively. A Franck-Condon progression generated for an
origin one additional mode to higher energy (24872 cm-1
)
proved to be much poorer at reproducing the spectrum, even
with large values for the normal coordinate change. This,
coupled with the fact that this origin would lie well under the
onset of absorption, suggests that it is probably not a reasonable
candidate for E00.
Figure 6. Franck-Condon progression based on E00 ) 22942 cm-1
and E00 ) 23907 cm-1
.
of emission spectroscopy, no clear evidence for multiple titanium
emitting sites can be gleaned from the spectra. While this may
suggest that only one distinct titanium site emits, such an
interpretation is probably not warranted. We cannot, for
example, completely rule out the presence of other, weak
titanium emissions in the area of spectral congestion imparted
by the silicalite background. Alternatively, close similarities in
the coordination geometries of the individual sites, irrespective
of their crystallographic distinction, may render them indistin-
guishable in the spectrum. Indeed, the loss of resolvable vibronic
structure at higher titanium concentrations may be due to the
superposition of closely related emissions emanating from the
different sites, although this observation may also be accounted
for by changes in the coupled vibrations of the lattice with
increasing titanium substitution.
Spectroscopic Assignments and Franck-Condon Analysis.
In the absence of a well-characterized coordination geometry,
rigorous spectroscopic assignments are impossible. The fact that
the titanium is in the +4 oxidation state means that the excited
states are all LMCT in nature. The long lifetime (80 µs at room
temperature) coupled with the weakness of the electronic
absorption in the region where direct excitation would be
expected suggests that the emitting state is strongly nonallowed.
The unambiguous assignment of the zero-point energy (E00) of
the emitting state is also difficult due to the fact that the silicalite
background obscures the high-energy edge of the titanium
emission. We expect, on the basis of the position and apparent
bandwidth of the emission, that E00 should fall between 22000
and 24000 cm-1, underneath the silicalite background. Consis-
tent with this expectation, the excitation spectrum has measur-
able, albeit weak, intensity in this region (Figure 4). As
mentioned, the emitting state is nonallowed and a direct
transition into it would be very weak. The highest energy
vibronic band that can be determined with confidence is at
∼21904 cm-1 (Figure 4). At higher energy there appears to be
another mode of the progression, which can be seen as a
shoulder on the red edge of the silicalite emission at ∼23000
cm-1. While this peak is in a reasonable position to be the origin
of the progression, it is certainly possible that other modes lie
to higher energy and are obscured.
In general, the spectroscopic and photophysical properties of
the titanium in TS-1 are analogous to those observed for other
high-valent early transition metals in a silica matrix. In
particular, both vanadium(V) and chromium(VI) dispersed in
amorphous silica have long-lived phosphorescent LMCT states,
and both show resolved vibronic progressions in the emission
spectra corresponding to M-O-Si stretching modes.19,20 Taken
as a series, the emission energy decreases going from titanium
to chromium as the metal becomes easier to reduce. While it is
not possible to infer molecular structure from such an analogy,
the similarities are striking enough to suggest that perhaps some
particular structural feature, common to all of these metal sites,
gives rise to the observed spectroscopic and photophysical
properties.
Electronic Spectroscopy and the Assignment of Ti Coor-
dination Geometry. Since the discovery of TS-1 (and related
titania-silica materials), a great deal of effort has been expended
to elucidate the coordination geometry of the titanium center.
X-ray analyses (XAFS and XANES) clearly indicate that the
titanium is not octahedral, with coordination numbers of 4-5
usually being reported.3 The initial suggestion that titanyl groups
(TdO) are present has been disputed by a number of investiga-
tors, and the current view is that Ti(IV) is substituted isomor-
phically for Si(IV) in the silica network and resides in a
tetrahedral or distorted tetrahedral environment.21-23
Among the arguments used to support the assignment of
tetrahedral coordination to the titanium sites in TS-1 are those
based on diffuse reflectance UV-vis spectroscopy. As a result,
this technique has grown to serve as a general analytical tool
for determining the presence of extraframework titania in
samples of TS-1.24-27 The basis of these structural arguments,
as put forth by Bocutti et al., involves an application of
Ju¨rgensen’s concept of optical electronegativity.24 In employing
this approach, the energy (ν (cm-1)) of a LMCT transition is
predicted from the optical electronegativity equation (shown
below), where ø(M) is a function of the metal and its
coordination geometry and ø(L) is a function of the ligand.28,29
ν (cm-1) ) 30000[ø(L) - ø(M)]
If we assign the mode at ∼23000 cm-1 as either the 0-0 or
the 0-1 mode, then the most intense and best resolved
vibrational mode in the center of the envelope at 19082 cm-1
would be the 0-4 or 0-5 transition. If we start from that point
and add four and five 965 cm-1 quanta, we estimate a zero-
point energy of either 22942 or 23907 cm-1, respectively. Figure
6 shows the Franck-Condon progressions calculated for a 965
In the analysis of TS-1, the value of ø(M) used for titanium-
(IV) was 2.05 and 1.85 for octahedral and tetrahedral coordina-
tion, respectively, and ø(L) for the oxide ligand was estimated
at 3.45.30 These values predict a LMCT transition at 48000 cm-1
for tetrahedral coordination and 42000 cm-1 for octahedral
coordination. The authors concluded from these predictions and
the absence of an observable absorption in the diffuse reflectance

2692 J. Phys. Chem. B, Vol. 105, No. 14, 2001
Soult et al.
between 30000 and 42000 cm-1 that the titanium did not sit in
an octahedral site. While this is a reasonable conclusion, it is
more compellingly deduced from X-ray data.31 The authors
further concluded that the “remarkable” agreement between the
energies of the experimentally determined absorption maximum
and the predicted maximum proves that the titanium is in a
tetrahedral environment in the silicalite lattice.24
David Gormin for his assistance in determining the phospho-
rescence lifetimes. We thank Dr. Ted Zateslo and Dr. Afi Sachi-
Kocher of the National High Magentic Field Laboratory for
assistance with the ICP-MS measurements. We thank Dr. Bruno
Notari for helpful discussions. The work was conducted under
support provided by the National Science Foundation, Division
of Materials Research, under Grant DMR-9623570.
There is a fundamental problem with this argument that can
be addressed with the spectroscopic data reported here. The
optical electronegativity expression predicts the energy of the
LMCT transitions for only two high-symmetry geometries:
tetrahedral and octahedral. However, it cannot be used to exclude
other, lower symmetry coordination environments that could
easily have LMCT transitions at the same energy as the high-
symmetry species. More importantly, it is clear from the
spectroscopy that the primary assertion that no electronic
absorption is observed below 42000 cm-1 for the titanium site
in TS-1 is simply incorrect.24,25 The emission excitation spectra
clearly show that the emitting state of the titanium is excited at
wavelengths between 32000 and 42000 cm-1. The reason that
absorption bands in the diffuse reflectance electronic spectrum
corresponding to these transitions are not observed until higher
concentrations of titanium are reached is due to the weakness
of the transitions (which are strongly nonallowed), coupled with
the inherent insensitivity of diffuse reflectance techniques.32
These results clearly show that electronic spectroscopy does not
support simple tetrahedral coordination of the titanium in
silicalite. Furthermore, it is not possible to infer a structure from
the electronic spectroscopy, though, consistent with X-ray
analysis, a lower symmetry, four-coordinate species may very
well explain the spectra. It also indicates that the observation
of intensity in the diffuse reflection spectrum in the region below
42000 cm-1 may not necessarily be diagnostic of extraframe-
work titania, especially in samples with higher titanium
concentrations where the weak titanium-based transitions be-
come observable. Conversely, the observation of a strong
absorption in that spectral region, especially at lower titanium
concentrations, is unlikely to be from the titanium site and may
be indicative of extraframework titania or other impurities.
Finally, with regard to the resolved vibrational structure, it is
clear that the ∼960 cm-1 vibrational mode in TS-1 is not
accurately described as a Si-O- stretch “perturbed” by the Ti-
(IV), as has been suggested.11,25,33 The fact that it is an active
progression in the emission spectrum indicates that it is strongly
coupled to the electronic manifold of the titanium and is more
correctly viewed as a Ti-O-Si stretch. Notably, recent
vibrational studies of crystalline titanium silicates tend to support
this interpretation.34
References and Notes
(1) Nordberg, M. E. U.S. Patent 2326059, 1943.
(2) Notari, B. AdV. Catal. 1996, 41, 253.
(3) Vayssilov, G. N. Catal. ReV. Sci. Eng. 1997, 39, 219.
(4) Lamberti, C.; Bordiga, S.; Arduino, D.; Zecchina, A.; Geobaldo,
F.; Spano, G.; Genomi, F.; Villian, F.; Vlaic, G. J. Phys. Chem. B 1998,
102, 6382.
(5) Taramasso, M. U.S. Patent 4410501, 1983.
(6) Van Der Pol, A. J. H. P.; van Hoof, J. H. C. Appl. Catal., A 1992,
92, 93.
(7) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Ple-
num: New York, 1983; p 42.
(8) Ingle, J. D.; Crouch, S. R. Spectrochemical Analysis; Prentice
Hall: Upper Saddle River, NJ, 1988; p 71.
(9) McCoy, E. F.; Ross, I. G. Aust. J. Chem. 1962, 15, 573.
(10) Blitz, J. P. In Modern Techniques in Applied Molecular Spectros-
copy; Mirabella, F. M., Ed.; Wiley: New York, 1998; p 189.
(11) Zecchina, A.; Spoto, G.; Bordiga, S.; Padovan, M.; Leofanti, G.;
Petrini, G. In Catalysis and Adsorption of Zeolites; Ohlmann, G., Ed.;
Elsevier: Amsterdam, 1991; p 671.
(12) Deo, G.; Turek, A. M.; Wachs, I. E.; Huybrechts, D. R. C. Zeolites
1993, 13, 365.
(13) We obtain good fits to a biexponential decay function which
accounts well for the two emitting regions in this system. Closer analysis
is likely to reveal that the decay is more complex and better fit to a
distribution function, as is well-known in heterogeneous porous media. [For
a lucid discussion of decays in heterogeneous porous solids see: Castellano,
F.; Meyer, G. J. In Progress in Inorganic Chemistry; Karlin, K. D., Ed.;
Wiley: New York, 1997; Vol. 44, p 167 and references therein.]
(14) Ichihashi, Y.; Yamashita, H.; Anpo, M.; Souma, Y.; Matsumura,
Y. Catal. Lett. 1998, 53, 207.
(15) Hurtubise, R. J. Solid Surface Luminescence Analysis; Marcel
Dekker: New York, 1981; p 60.
(16) Lamnberti et al. [ref 13] also report an emission centered at 500
nm for the tetrehedral model compound Ti(OSi(CH3)3)4. This observation
was used to assign the low-energy band as originating from a tetrahedral
site in the lattice. We observed no emission from this compound either in
cyclohexane at room temperature or in a methylcyclohexane glass at 77 K.
(17) Labouriau, A.; Ott, K. C.; Rau, J.; Earl, W. L. J. Phys. Chem. B
2000, 104, 5890.
(18) Better agreement can be realized by relaxing the constraint that νg
) νe [Yersin, H.; Otto, H.; Zink, J. I.; Gliemann, G. J. Am. Chem. Soc.
1980, 102, 951]; however, given the complexity of the spectra and the
qualitative nature of this Franck-Condon analysis, this higher level of
calculation seems unwarranted.
(19) Hazenkamp, M. F.; Blasse, G. J. Phys. Chem. 1992, 96, 3442.
(20) Tran, K.; Hanning-Lee, M. A.; Biswas, A.; Stiegman, A. E.; Scott,
G. W. J. Am. Chem. Soc. 1995, 117, 2618.
Conclusion
(21) Notari, B. In Studies in Surface Science and Catalysis 37; Grobet,
J. M., Ed.; Elsevier: Amsterdam, 1988; p 413.
(22) Perego, G.; Bellussi, G.; Corno, C.; Taramasso, M.; Buonomo, F,
Esposito, A. In Studies in Surface Science and Catalysis 28; Murakami,
Y., Iijima, A., Ward, J. W., Eds.; Elsevier: Amsterdam, 1986; p 129.
The titanium site in TS-1 catalyst materials has a long-lived
phosphorescent excited state with a maximum at 490 nm in the
near-UV. This state is significantly lower in energy than was
previously suggested by diffuse reflectance absorption spec-
troscopy. Emission excitation spectra indicate that there are weak
electronic absorptions throughout the spectral region from
∼23000 to 48000 cm-1. These observations bring into question
long-standing structural arguments for the coordination of
titanium in the silicalite lattice that have been made using
electronic spectroscopy.
(23) Crocker, M.; Herold, R. H. M.; Roosenbrand, B. G.; Kees, A. E.;
Wilson, A. E. Colloids Surf., A 1998, 139, 351.
(24) Boccuti, M. R.; Rao, K. M.; Zecchina, A.; Leofanti, G.; Petrini, G.
In Structure and ReactiVity of Surfaces; Norterra, C., Zecchina, A., Costa,
G., Eds.; Elsevier: Amsterdam, 1989; p 133.
(25) Zecchina, A.; Spoto, G.; Bordiga, S.; Ferrero, A.; Petrini, G.;
Leofanti, G.; Padovan, M. In Zeolite Chemistry and Catalysis; Jacobs, P.
A., Ed.; Elsevier: Amsterdam, 1991; p 251.
(26) Geobaldo, F.; Bordiga, S.; Zecchina, A.; Giamello. E.; Leofanti,
G.; Petrine, G. Catal. Lett. 1992, 16, 109.
(27) Duprey, E.; Beaunier, P.; Springuel-Huet, M.-A.; Bozon-Verduraz,
F.; Fraissard, J.; Manoli, J.-M.; Bre´geault, J.-M. J. Catal. 1997, 165, 22.
Acknowledgment. We thank Dr. Eric Lochner of the
Materials Research and Technology Center (MARTECH) at
Florida State University for providing X-ray analysis and Dr.

Electronic Spectroscopy of the Ti Centers in TS-1
J. Phys. Chem. B, Vol. 105, No. 14, 2001 2693
(28) Jørgensen, C. K. In Progress in Inorganic Chemistry; Lippard, S.
J., Ed.; Wiley: New York, 1979; p 101.
(31) On, D. T.; Bonneviot, L.; Bittar, A.; Sayari, A.; Kaliaguine, S. J.
Mol. Catal. 1992, 74, 233.
(29) Lever, A. B. P. Inorganic Electronic Spectroscopy; Elsevier:
Amsterdam, 1984; p 218.
(30) While this value of the optical electronegativity for the oxo ligand
is probably reasonable, it should be noted that there are a number of
problems in assigning optical electronegativites for oxygen [Muller, A.;
Diemann, E.; Jørgensen, C. K. Struct. Bonding 1973, 14, 23].
(32) Reference 10, p 188.
(33) Camblor, M. A.; Corma, A.; Perez-Pariente, J. J. Chem. Soc., Chem.
Commun. 1993, 557.
(34) Su, Y.; Balmer, M. L.; Bunker, B. C. J. Phys. Chem. B 2000, 104,
8160.