Sub-picosecond injection of electrons from excited [Ru(2,2′-bipy-4,4′-dicarboxy)2(SCN)2] into TiO2 using transient mid-infrared spectroscopy

Article abstract of DOI:10.1524/zpch.1999.212.Part_1.077

We have used femtosecond pump-probe spectroscopy to time resolve the injection of electrons into nanocrystalline TiO₂ film electrodes under ambient conditions following photoexcitation of the adsorbed dye, [Ru(4,4′-dicarboxy-2,2′-bipyridine)

Full text of DOI:10.1524/zpch.1999.212.Part_1.077

Zeitschrift für
by
Oldenbourg Wissenschaftsverlag,
Bd.
München
S. 77-84
Physikalische Chemie,
212,
(1999)
©
Sub-picosecond Injection
of Electrons from Excited
[
Ru(2,2r-bipy-4,4 -dicarboxy)2(SCN)2] into
Ti02
Transient Mid-Infrared
Using
Spectroscopy*
J.
John B.
Sue
N.
Randy Ellingson1,
Asbury2,
Ferrere', Hirendra
Ghosh2,
Julian R.
Lian2 and Arthur J. Nozik1
Sprague', Tianquan
1
Center for Basic Sciences, National Renewable
Energy Laboratory,
Atlanta,
Golden, Colorado, 80401, USA
2
Department of Chemistry,
Emory University,
Received August 8, 1998; accepted December
Georgia, 30322,
0, 1998)
USA
(
1
Dynamics ofdye injection IN31 Nanocrystalline Ti02/
sensitization
I
Dye
Monocrystalline Ti021 Fast laser spectroscopy
electron
I
Ultrafast
injection
We have used femtosecond
pump-probe spectroscopy to time resolve the injection of
electrons into
film electrodes
under ambient conditions
nanocrystalline Ti02
following
of the adsorbed
photoexcitation
dye,
[
Ru(4,4'-dicarboxy-2,2'-bipyridine)2(NCS)2] (N3).
at one of the
transfer
Pumping
metal-to-ligand charge
the
peaks and
the
the
absorption
conduction
band at 1.52
pm
probing
of electrons
pm,
absorption
injected into
we
Ti02
and
in
of 4.1 to 7.0
range
have directly observed the arrival of the injected electrons.
Our measurements indicate an instrument-limited
~
50-fs
limit on the electron
upper
compared the infrared transient
of N3 in
time under ambient conditions in air. We have
injection
for
(blank)
absorption
noninjecting
systems consisting
ethanol and N3 adsorbed
of electron
to films of
ALO, and
and found no indication
nanocrystalline
Zr02,
to 7.0
injection
at probe
in the mid-IR
wavelengths
(4.1
pm). At 1.52
pm
interferences exist in the
observed transient
for the blanks.
absorption signal
1.
Introduction
Extensive efforts have been directed toward
the
understanding
photoelectro-
chemical
properties and charge
transfer
dynamics of dye-sensitized nano-
*
Presented at the "Twelfth International Conference on
Photochemical Conversion
and
of Solar
1998.
Storage
Energy", Berlin, August 9-14,
*
*
The work
here has been
previously published in Ref. [11].
reported
E-mail of
author:
correspondence
anozik@nrel.nrel.gov.
77
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

78
Randy J. Ellingson et
al.
semiconductor fdms. Such fdms used in
cells have
in recent
solar cells based on
crystalline
photovoltaic
demonstrated
efficient solar
conversion
highly
years [1—3]. Dye-sensitized photoelectrochemical
adsorbed to
energy
(>10%)
[
Ru(4,4'-dicarboxy-2,2'-bipyridine)2(NCS)2] (N3)
on efficient
nanocrystal-
line Ti02
at the
interface, and
rely
charge-separation
dye-Ti02
fast electron
into the Ti02
from the excited state of the adsorbed
molecule
dye
injection
must be fast to
of the excited
with
processes
nanoparticle. Injection
compete
dye
such as relaxation and
molecule. Several
degradation
the transient
of the excited and oxidized states
groups studying
absorption
of the
in the visible and near-IR, to determine the electron
dye
injection
reported
time in the N3-sensitized Ti02 (N3-Ti02) and similar
have
systems,
evidence for ultrafast (<1
and electron
[4—
ps) charge separation
injection
0]. Tachibana et
al.
transient
measurements at 750 nm
1
reported
absorption
which were ascribed to the formation of oxidized N3
at 605 nm)
;
(
pumped
of the rise time of the 750 nm
indicated
elec-
analysis
absorption
biphasic
tron
with
times of 150 fs (instrument limited) and 1.2
injection
injection
ps
[
4]. Hannappel et al. [5] measured the rise of the near-infrared absorption
of conduction band electrons at HOOnm
of N3
following photoexcitation
at the
transfer
lower-energy metal-to-ligand charge
(MLCT) absorption peak
(
550 nm)
;
the near-infrared absorption signal was attributed to free or
electrons
However, these authors believe that
following injection.
trapped
to obtain
results, it was
for these
to be
unambiguous
necessary
experiments
conducted under
the
vacuum (UHV) conditions. These authors also
ultra-high
Tachibana et al. of the 750 nm
challenged
assignment by
absorption
band to oxidized N3 [4]; this
has been answered in turn [6, 7].
challenge
Heimer et al. [8] used
a
with 30
resolution to time resolve the
ps
system
electron
into Ti02 from the
[Ru(4,4'-(COOCH2CH3)2-2,2'-bi-
injection
pyridine)(2,2'-bipyridine)2]+2;
limit on the time of —20
dye
the instrument time
an
response placed
further
upper
ps [10]. Here we present
injection
evidence,
based on time-resolved infrared (TRIR)
measurements, for ul-
absorption
trafast electron
times (50 fs±25 fs) for the N3-Ti02 electrode un-
injection
der ambient conditions.
the time-resolved mid-IR
By probing
absorption
electrons [9, 11], we are able to obtain
produced by injected
unambiguous
conclusions without the need for UHV.
Probing in
the mid-IR allows us to
for oxidized N3 and
the issue of the correct
bypass
absorption spectrum
eliminate the
of
contributions from the cationic or
possibility
any signal
11].
excited states of the
dye [9,
2. Materials and methods
Titanium dioxide
colloids were
as
previously de-
A1203 nanopar-
nanoparticle
scribed, using Degussa P25 Ti02 starting material [12].
ticle colloid was
prepared
The
prepared using a
method similar to that used for Ti02. The
78
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

79
Sub-picosecond Injection of Electrons
material for the alumina colloid was
Aluminum Oxide
C
hy-
starting
Degussa
size of 13 nm). The Zr02 colloid was
via
(
primary particle
prepared
of zirconium
and heated at 220 °C
a
drolysis
isopropoxide
for the
using procedure
of colloidal Ti02. Films were fired
published previously
preparation
at 450 °C for 45 minutes in air. Films to be used with
probe wavelengths
than 5.5 pm were
on 2.0-mm-thick CaF2 substrates, both
longer
sides
prepared
Films to be used at
on either c-cut
shorter than 5.5
were
polished.
wavelengths
or
pm
Sn02 con-
fluorine-doped
prepared
polished sapphire
substrates. Ti02 and Zr02 films were 5-8 pm thick with
ducting glass
transparency. A1203
good
films were —15
thick, and showed
greater
scat-
pm
than the Ti02 films. Immersion and
of the Ti02, A1203 and
tering
storage
ethanol solution
Zr02 films in
a
room
200
containing
adsorption
N3
temperature
pM
and 20 mM
to the
acid [14] resulted in
of the N3
films showed
chenodeoxycholic
film surface. The
dye
porous
resulting dye-sensitized
an absorbance of —1.0 at 400 nm and 550 nm.
chased from Solaronix (Lausanne, Switzerland).
measurements were made
N3 was
High purity
pur-
two different laser
Pump-probe
using
systems
described
[11].
previously
3
.
Experimental results
With the N3-Ti02
to air, transient
measurements
sample exposed
absorption
show
a
visible
and
a
visible or near-IR
a
using
pump
probe
problematic
et al. also observed such an
long-lived absorption component. Hannappel
accumulated
for visible
measurements made in
solid state
a
solvent en-
signal
probe
vironment [5];
when
under UHV did
only
measuring
samples
the accumulated
vanish. Our initial
of various
signal
investigations
samples
as well
at 550 nm and
at 1.52
showed
pumped
probed
pm
long-lived signals
as
not attributable to
electrons.
The N3-Ti02
1
shows
signals
absorption by injected
Fig.
data for N3-Ti02, N3-Zr02 and N3-ethanol
samples.
sample
shows an ultrafast rise of
and
initial
similar
absorption
subsequent
decay,
to that
et al.
However, it is clear from measure-
reported by Hannappel
[5].
ments of the N3-Zr02, where
is not
and N3-
significant injection
expected,
ethanol, where no semiconductor is present,
that other sources of transient
at 1.52 pm are
from either N3* or N3
in
absorption
present
photoproducts
this near-IR
To rule out nonlinear effects such as two
region.
photon (pump)
we have verified that the
is linear with
absorption,
absorption signal
for N3-Ti02
well be dominated
may very
pump
The transient
electron
pulse energy.
signal
the
but no clear conclusions can be drawn
by
injected
absorption,
without well-behaved blanks to
N3* or N3+.
verify the absence
of
a
from
signal arising
Due to
a
we observed when
we made additional transient absorption measurements at
at 1.52
long-lived absorption
probing
pm,
IR-wave-
longer
79
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

80
Randy J. Ellingson et
al.
The
were measured in air, and the linear absorbance at
samples
lengths.
2
110 cm-1 (4.74
was 30 mO.D. for the N3-Ti02
and 70 mO.D.
pm)
for the N3-A1203
sample
the
1
kHz
we used
a
sample. Using
system,
pump pulse
energy of
1.1
at 400 nm and 100 fs
to excite
of N3-
pJ
pulsewidth
samples
Ti02, N3-A1203, N3-Zr02, and unsensitized Ti02. Probing the resultant ab-
at 4.63
and 4.9
pm, we
find no transient absorbance for either
sorption
pm
the N3-A1203 or N3-Zr02
and
a
small absorbance for the
samples, a
only
unsensitized Ti02. N3-Ti02 shows
transient absorbance maximum of
~
3
mO.D. at 4.63
and ~6 mO.D. for 4.9
Results are shown in
pm
The transient mid-IR
pm.
2.
can be well fit
a
Fig.
signals
by single exponential
rise convoluted with the instrument
time constants of 50 fs
b. The risetime of 100 fs does not
function. The best fits
response
25 fs, as shown
yield
±
the solid lines in
2
a
and
by
Fig.
as
satisfactory a
fit of the data,
produce
as shown
mined in
the dashed line. The instrument
function was deter-
response
by
thin silicon wafer
a
before or after
a
kinetics scan at each
right
A
in silicon is shown in
2b; this
wavelength.
typical response
the integration of
Fig.
signal
can be well
a
function with
represented by
gaussian
FWHM of 185 fs. The transient absorbance for the unsensitized Ti02 is
—
0.6 mO.D., -10% of the absorbance for N3-Ti02 at 4.9
These results
pm.
show conclusive and direct evidence of the arrival of electrons within the
Ti02 following photoexcitation of adsorbed molecules of N3.
As shown in
3, when
injection
at 6.6
pm we
obtain similar results
Fig.
probing
electron
for N3-Ti02, no
for N3-A1203, and
showing
IR
injection
for unsensitized Ti02. In addition, the data shows
a
fast
slight
absorption
80
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

Sub-picosecond Injection of Electrons
81
-
0.001
-1.5
-1.0
-0.5
0.0
0.5
1.0
15
Delay (ps)
.
Q.2
-1.0
-0.5
0.0
0.5
1.0
1.5
Delay (ps)
of
2
.
TRIR
data for
4.63
(a),
and
neither
4.9
follow-
N3-A120,
nor N3-
Fig.
absorption
probe wavelengths
(a)
µ
(b)
pm
400 nra, 100 fs
As seen in
ing photoexcitation by
pulses.
Zr02 show evidence of injected electron absorption. The system
function from
response
free-carrier
in silicon shown in (b) is seen to match
the rise of
due to
injection
absorption
very closely
the
for N3-TÌO,; unsensitized Ti02 shows
a
small
absorption signal
absorption
carriers photoexcited
within the electrode. The rise times indicate an
directly
time -50 fs (after Ref. [11]).
81
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

82
Randy J. Ellingson et al.
N3-sens¡tized
Unsensitized
N3-sensitized ALO,
T¡02
Ti02
0
.008
.006
q
0
0.000
0.0
0.5
1.0
1.5
Delay (ps)
6.6 prn probe for N3-Ti02, N3-A120, and unsensitized Ti02
Fig. 3. TRIR absorption of
following photoexcitation by
for N3-Ti02 is followed by an equally fast initial decay. The source of this decay has not
been
it is believed to be due to cooling or trapping of the
a
4
00 nm, 100 fs pulses. The fast (~50 fs) rise of
absorption
decisively assigned, though
injected electrons (after Ref. [11]).
initial
the rise; this
occurs on
a
—50 fs timescale.
decay following
decay
The
either the
of the initial
or
is not determined, but is consistent with
origin
decay
of the
electrons. The free electron
cooling
trapping
injected
cross-section is
to the
of states and relax-
absorption
proportional
density
ation of electrons toward the conduction band minimum is
to result
expected
in reduced IR
duced IR
dence of the IR
of electrons could also result in
measurement of the
indicates an absorbance which increases with
a re-
absorption. Trapping
Our
absorption
absorption.
preliminary
spectral depen-
probe wavelength.
4
.
Discussion
TRIR
we have measured the electron
Using
absorption spectroscopy,
injec-
tion time for N3-Ti02.
of N3-ethanol, N3-Ti02, N3-
Pumping samples
A1203, and N3-Zr02 at 400
nm or 550
nm,
we
the
probed
absorption by
and also in
pm
electrons
infrared femtosecond
at 1.52
TRIR
pm
injected
using
pulses
the
of 4.1 to 7.0
Our results from the 1.52
range
pm.
experiments
show the
of
as well as
N3-
presence
long-lived absorption,
absorption by
ethanol and N3-Zr02.
from N3-Zr02 is
inefficient,
Any injection
no free or
likely quite
and N3-ethanol
electrons
;
therefore, the
of the
provides
trapped
absorp-
tion
tion of N3 or
for these
could be due to
signals
samples
possibly to lower-lying N3
products
photodegrada-
excited state
absorption. Hannappel
82
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

Sub-picosecond Injection of Electrons
83
et al.
made in
that measurements of the visible transient
[
5] reported
absorption
which
a
solvent or air environment exhibited an accumulated
signal
attributed to "side reactions". To overcome this
accumulated
they
spurious
made all solid state measurements in
a
UHV environment. We
signal, they
have found that
further into the
measurement of the TRIR
IR; in the 4.1 to 7.0 pm range,
probing
electrons
simplifies
absorption by injected
by
the
of the UHV
long-lived absorption components
we observed no ac-
eliminating
requirement
apparatus;
cumulated or
in this
wavelength range.
In
our results indicate that we eliminate the
of
addition,
possibility
any
contribution to the IR
from the
non-injecting dye
of 4.63,
probe wavelengths
excited or
molecules.
absorption signal arising
dye
cationic states; the former could arise from
of our transient
data for
Analysis
absorption
4
.9, and 6.6 pm, shows an instrument-limited electron injection time for
N3-Ti02 of -50 fs. For A1203 with
a
bandgap of -10 eV, and Zr02 with
a
of
5
eV [15 18], the conduction band
is
to lie too
bandgap
edge
expected
far
to allow
from N3*. As
our TRIR
negative
injection
expected,
absorption
measurements indicate that while N3-Ti02
an
elec-
produces
IR-absorbing
tron
the
within the semiconductor, no such
is
for
absorption
population
absorption
present
N3-A1203 and N3-Zr02
The broad IR
non-injecting
samples.
of N3-Ti02 could be due to free electrons,
electrons, or both. Data
trapped
of the
taken at 6.6
pm suggests a cooling
or
electrons on
trapping
injected
the —50 fs timescale. Efforts to measure the time-resolved
injected-electron
IR
for
N3-Ti02 are underway.
absorption spectrum
Acknowledgement
R.J.E., S.F., J.R.S. and A.J.N. of NREL were supported by the U.S. Depart-
ment of Office of
Research, Division of Chemical Sciences.
Energy,
J.A., H.N.G. and T.L. of EU were supported by the Petroleum Research
Fund (administered the American Chemical
the Univer-
Energy
by
Society),
No. 9733796. We thank Brian
Emory
for his assistance
Research Committee, and the National Science Foundation CAREER
sity
award under
grant
Fluegel
absorption measurements.
with the 1.52 pm transient
References
1
2
3
.
.
.
B.
and M. Griitzel, Nature 353 (1991) 737.
O'Regan
M. K. Nazeeruddin, P. Pecha and M. Grátzel, Chem. Commun. (1997) 1705.
Conversion
M. Grátzel, "Photoelectrochemical Solar
by Dye
Sensitization",
Energy
in Future Generation Photovoltaic Technologies, R. D. McConnell, Ed., A.I.P. Conf.
Proa, 404, American Institute of Physics, Woodbury NY, 1997.
4
.
.
Y. Tachibana, J. E. Moser, M. Grátzel, D. R. Klug and J. R. Durrant, J.
Chem.
Phys.
1
00 (1996) 20056.
T. Hannappel, B. Burfeindt, W. Storck and F. Willig, J. Phys. Chem.
799.
5
101 (1997)
6
83
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM

Randy J. Ellingson et al.
84
6
.
J. E. Moser, D. Noukakis, U. Bach, Y. Tachibana, D. R. Klug, J. R. Durrant, R.
Humphry-Baker and
M. Grâtzel, J.
Phys.
Chem.
102 (1998) 3649.
7
.
T.
J.
C. Zimmermann, B. Meissner, B. Burfeindt, W. Storck and F. Willig,
102 (1998) 3651.
Hannappel,
Chem.
Phys.
8
.
T. A. Heimer and E. J. Heilweil, J. Phys. Chem.
101 (1997) 10990.
9.
.
A
.
Gosh, J.
.
and T. Lian, J.
Chem.
(1998) (in press).
Asbury
Phys.
1
1
1
0.
more recent measurement with
time resolution in the subpicosecond
improved
time of —500 fs (E. J. Heilweil,
communication).
private
A.
revealed an
range
1. R. J. Ellingson, J. B. Asbury, S. Ferrere,
Chem.
102 (1998) 6455.
injection
.
N. Ghosh, J. R. Sprague, T. Lian and
J. Nozik, J.
. .
Vlachopoulos
Phys.
2.
Nazeeruddin, A. Kay, I. Rodicio, R. Humphry-Baker, E. Midler, P. Liska, N.
and M. Grâtzel, J. Am. Chem. Soc. 115 (1993) 6382.
1
1
1
1
3. A. Zaban, S. Ferrere, J. Sprague and B. A. Gregg, J. Phys. Chem. 101 (1997) 55.
4. A. Kay and M. Grâtzel, J. Phys. Chem. 97 (1993) 6272.
5. J. G. Bendoraitis and R. E. Salomon, J. Phys. Chem. 69 (1965) 3666.
6. P. Clechet, J. Martin, R. Oliver and C. Vallouy, C. R. Acad. Sci.
C
282 (1976) 887.
Phys. 48 (1977)
1
7. H. H.
Kung, H.
S. Jarrett, A. W.
and A. Ferretti, J.
Sleight
Appi.
2463.
1
8. T. A. Heimer and G. J. Meyer, J. Lumin. 70 (1996) 468.
84
Brought to you by | University of Saskatchewan
Authenticated
Download Date | 3/19/18 10:50 AM