Non-aggregated hyperbranched phthalocyanines: Single molecular nanostructures for efficient semi-opaque photovoltaics

Article abstract of DOI:10.1039/c2ra21496b

A series of single molecular nanostructured hyperbranched phthalocyanines, HBMPc-COOH (M = H₂, AlCl, Co, Cu, Zn), have been synthesized, characterized, and systematically studied as efficient semi-opaque sensitizers of TiO₂ by using ultraviolet-visible absorption, steady-state and femtosecond time-resolved fluorescence, cyclic voltammetry, current-voltage and photoelectric measurements. The inherent steric effect within these hyperbranched structures effectively suppresses the aggregation of phthalocyanine rings on TiO₂ surface, providing a facile approach for improving the photovoltaic performance of phthalocyanine-based dye sensitized solar cells (DSSCs). A power conversion efficiency of 1.15% along with a high incident photon to current conversion efficiency of 66.7% at 670 nm is achieved from HBZnPc-COOH sensitized solar cells. These results are consistent with findings from the femtosecond time-resolved fluorescence study, which reveals an ultrafast and efficient multi-phasic interfacial electron injection from both the Soret and Q bands to the conduction band of TiO₂. The changing of the metal centers dramatically affects the optical, photophysical, electrochemical, and photovoltaic properties of the hyperbranched phthalocyanines. Zinc has been found to be the best metal center in the whole series for the design of hyperbranched phthalocyanine-based dyes for DSSC applications.

Full text of DOI:10.1039/c2ra21496b

RSC Advances
View Article Online
View Journal | View Issue
PAPER
Non-aggregated hyperbranched phthalocyanines:
single molecular nanostructures for efficient semi-
opaque photovoltaics
Cite this: RSC Advances, 2013, 3, 545
Yong Li,ab Peifen Lu,a Xingzhong Yan,*a Lu Jinb and Zhonghua Pengb
A series of single molecular nanostructured hyperbranched phthalocyanines, HBMPc-COOH (M = H2, AlCl,
Co, Cu, Zn), have been synthesized, characterized, and systematically studied as efficient semi-opaque
sensitizers of TiO2 by using ultraviolet-visible absorption, steady-state and femtosecond time-resolved
fluorescence, cyclic voltammetry, current–voltage and photoelectric measurements. The inherent steric
effect within these hyperbranched structures effectively suppresses the aggregation of phthalocyanine
rings on TiO2 surface, providing a facile approach for improving the photovoltaic performance of
phthalocyanine-based dye sensitized solar cells (DSSCs). A power conversion efficiency of 1.15% along with
a high incident photon to current conversion efficiency of 66.7% at 670 nm is achieved from HBZnPc-
COOH sensitized solar cells. These results are consistent with findings from the femtosecond time-resolved
fluorescence study, which reveals an ultrafast and efficient multi-phasic interfacial electron injection from
both the Soret and Q bands to the conduction band of TiO2. The changing of the metal centers
dramatically affects the optical, photophysical, electrochemical, and photovoltaic properties of the
hyperbranched phthalocyanines. Zinc has been found to be the best metal center in the whole series for
the design of hyperbranched phthalocyanine-based dyes for DSSC applications.
Received 18th July 2012,
Accepted 31st October 2012
DOI: 10.1039/c2ra21496b
www.rsc.org/advances
thus can be treated as a life window suitable for certain
practical applications, such as semi-opaque solar panels to be
Introduction
equipped on building/greenhouse windows.6,11 Besides, this
absorption window can be filled up through co-sensitization
with other visible-light-harvesting dye(s)7 or through chemical
structure modification12–14 for broader spectral coverage
toward more efficient DSSCs.
DSSCs have drawn a broad interest as a cost-effective approach
to convert solar energy into electricity.1–10 The state-of-the-art
power conversion efficiency of DSSCs has been increased very
recently from y11% for ruthenium complex based sensitizers
to 12.3% for a donor-p-bridge-acceptor zinc porphyrin dye
(YD2-o-C8).2 In spite of this, ruthenium complexes are still the
most widely used TiO2 sensitizers in DSSCs so far.3,4 Due to
the relatively high band gaps of these dyes, more than half of
the total solar irradiance especially at wavelengths greater than
650 nm cannot be effectively harvested.3,5 In contrast,
phthalocyanine dyes, as well-known low band gap absorbers
with high chemical stability and tailorability, possess very
strong absorption bands in the ultraviolet (UV) and red to near
infrared (NIR) regions, along with a gap of very weak
absorption in the visible region at 400 to 600 nm.3–9 This
absorption gap allows efficient transmission of visible light
without sacrifice of UV and red to NIR photon harvesting, and
Many phthalocyanine sensitized solar cells have been
developed in the last few years.3–9 However, their performance
is not very impressive, mainly because of the strong tendency
of phthalocyanines to aggregate on the TiO2 surface and the
lack of directionality of electron transfer in the excited state.4,5
Recently, it was found that asymmetrical structures with
‘‘push’’ and ‘‘pull’’ groups or with axially substituted ligands
enable phthalocyanines to 1) reduce the aggregation on the
TiO2 surface, 2) tune the energy level of the excited state, and
3) provide intimate electronic coupling with the TiO2 surface.3–8
As a result, good efficiencies of up to 4.6% have been obtained
from asymmetrical phthalocyanine sensitized solar cells.5 The
issues for asymmetrical phthalocyanines are their low synthetic
yields and the difficulty in purification.3–8 On the other hand,
single molecular nanostructured hyperbranched phthalocya-
nines with the desired advantages of facile synthesis and
scalable processing procedures, have shown potential applica-
tions in solar cells,11 organic light-emitting diodes,15 and high-
energy density and pulsed capacitors16 in recent years. We
aCenter for Advanced Photovoltaics, Department of Electrical Engineering &
Computer Science, South Dakota State University, Brookings, South Dakota 57007,
USA. E-mail: xingzhong.yan@sdstate.edu.
bDepartment of Chemistry, University of Missouri-Kansas City, Kansas City,
Missouri, 64110, USA
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 545

View Article Online
Paper
RSC Advances
expect that the use of hyperbranched structures can suppress
the aggregation of phthalocyanines on the TiO2 surface. To the
best of our knowledge, there has been very little work regarding
DSSC applications of hyperbranched phthalocyanine-based
dyes.
In this study, we synthesized a series of hyperbranched
phthalocyanines, HBMPc-COOH (M = H2, AlCl, Co, Cu, Zn; 1–
5) (Scheme 1), as TiO2 sensitizers in DSSCs. The optical,
photophysical, electrochemical, and photovoltaic properties of
these hyperbranched phthalocyanines were investigated.
Scheme 1 Synthetic route of 1–5.
546 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
powdery product was dried under vacuum, affording HBMPc-
CN (M = H2, AlCl, Co, Zn) in yields of .85%.
Experimental section
For HBH2Pc-CN: IR (KBr, cm21): 3400, 3300, 3050, 2210,
1720, 1600, 1480, 1280, 1220, 1120, 1080, 1020, 960, 800, 780,
720.
General remarks
N,N-Dimethylformamide (DMF) was distilled from anhydrous
MgSO4. All other reagents and solvents were used as received
from Sigma-Aldrich. 1H NMR spectra were recorded on a
Varian 500 NMR Spectrometer or on a Bruker Avance 400 NMR
Spectrometer in CDCl3 or in DMSO-d6. IR spectra were
For HBAlClPc-CN: IR (KBr, cm21): 3440, 3050, 2210, 1750,
1600, 1480, 1280, 1220, 1120, 940, 780, 730.
For HBCoPc-CN: IR (KBr, cm21): 3400, 3050, 2210, 1750,
1600, 1480, 1280, 1220, 1120, 940, 780, 730.
recorded in KBr pellets using
a Nicolet 6700 FT-IR
For HBZnPc-CN: IR (KBr, cm21): 3400, 3060, 1710, 1600,
1490, 1400, 1360, 1280, 1210, 1120, 1050, 980, 822, 700.
Spectrometer with a spectral resolution of 4 cm21. Mass
spectra were taken on a Bruker TOF Mass Spectrometer or on
an Applied Biosystems Voyager-DE Pro Biospectrometry
Workstation Matrix-Assisted Laser Desorption/Ionization-
Time of Flight (MALDI-TOF) Mass Spectrometer (operating in
both linear and reflector mode). Dithranol was used as the
matrix in the measurements for 1–5. Gel Permeation
Chromatography (GPC) data were collected using a Tosoh
EcoSEC HLC-8320GPC equipped with a refractive index (RI)
detector. Absorption spectra were recorded on an Agilent 8453
UV-visible Spectrophotometer. Steady-state fluorescence spec-
tra were recorded with an Edinburgh FS920 Fluorescence
Spectrophotometer. Solutions for absorption and fluorescence
measurements were prepared using anhydrous DMF.
Synthesis of HBCuPc-CN16
1,3-bis(3,4-Dicyanophenoxy)benzene (206.91 mg, 0.57 mmol)
and 18.92 mg (0.19 mmol) of CuCl were dissolved in 5 mL of
dimethylacetamide (DMAc), and then the mixture was stirred
at 160 uC for 4 h. The mixture was poured into 100 mL of
water. The precipitate was collected by filtration, washed with
water, and dried under vacuum. After refluxing in methanol
twice, the product was filtered and rinsed with cold methanol
three times. The dark blue powdery product was dried under
vacuum, affording 182.07 mg of HBCuPc-CN. IR (KBr, cm21):
3400, 3050, 2250, 1710, 1600, 1480, 1220, 1180, 1120, 1100,
980, 820, 750.
Synthesis
Synthesis of 1–515
The hyperbranched phthalocyanines were prepared following
the synthetic steps shown in Scheme 1.
In general, HBMPc-CN (M = H2, AlCl, Co, Cu, Zn) and
potassium hydroxide (15 wt%) were dissolved in 12 mL of
water/ethanol (1/1, v/v) mixed solvent. The mixture was
refluxed for 24 h until the evolution of ammonia ceased. The
blue clear solution was poured into 60 mL of water and the pH
value of the solution was adjusted to 3–4. The dark blue
precipitate was collected by filtration and rinsed with dilute
HCl and then with water. Yield: .90%. GPC, MALDI-TOF MS,
and UV-visible absorption extinction coefficients were used for
estimating the molecular weight of the hyperbranched
structures.16 Due to the limited solubility of 1–5 in tetrahy-
drofuran (THF), it was hard to obtain precisely their molecular
weight information by GPC. GPC only provided the residual
monomer information for 2–5. The GPC data for 1 revealed a
broad molecular weight distribution with three to five rings on
average in the structure. Several trials showed that dithranol
was the best matrix for these samples, even though their
solubility in acetonitrile was rather poor (in which the MALDI-
TOF MS samples were prepared). Their MALDI-TOF MS spectra
could not comprehensively reflect the actual molecular weight
distribution either. However, the distinguishable ion peaks
with largest molecular weight demonstrated that there were
about four phthalocyanine rings within 2, 3, and 5. The
Synthesis of 1,3-bis(3,4-dicyanophenoxy)benzene16
4-Nitrophthalonitrile (2.00 g, 11.6 mmol) and 0.64 g (5.8
mmol) of resorcinol (1,3-dihydroxybenzene) were dissolved in
30 mL of dimethyl sulfoxide (DMSO). Then 1.60 g (11.6 mmol)
of potassium carbonate (K2CO3) was added into the flask, and
the mixture was stirred at room temperature for 24 h. The
mixture was poured into 600 mL of methanol/water (1/1, v/v)
mixture and stirred for 30 min. The precipitate was collected
by filtration, washed with 1000 mL of water, and dried under
vacuum. The pure product was obtained by recrystallization in
methanol. The powdery product was collected by filtration and
dried under vacuum for 24 h, affording 1.77 g of 1,3-bis(3,4-
dicyanophenoxy)benzene. Yield: 84.1%. 1H NMR (500 MHz,
CDCl3, d (ppm)): 7.76–7.82 (d, 2H, –C6H4), 7.53–7.58 (t, 1H, –
C6H4), 7.30–7.36 (m, 4H, –C6H3), 7.00–7.03 (m, 2H, –C6H3),
6.83–6.86 (t, 1H, –C6H4). MS: m/z 362.1 [Calcd. for C22H10N4O2:
[M]+ 362.3].
Synthesis of HBMPc-CN (M = H2, AlCl, Co, Zn)
In general, 1,3-bis(3,4-dicyanophenoxy)benzene (y0.6 mmol)
or its mixture with y0.3 mmol of AlCl3 or CoCl2?6H2O or
ZnCl2?xH2O, was dissolved in 5 mL of 2-dimethylaminoethanol
(DMAE), and then the mixture was refluxed for 75 min (for M =
H2, Zn) or 2 h (for M = AlCl, Co) under a slow stream of
nitrogen. The mixture was poured into 100 mL of water/
methanol (10/1, v/v) mixture. The precipitate was collected by
filtration, washed with 3 mol L21 HCl (this step was for M =
AlCl only) and then with water, and dried under vacuum. After
refluxing in methanol twice, the product was filtered and
rinsed with cold methanol three times. The dark green/blue
MALDI-TOF MS spectra also revealed
a quasi-Gaussian
distribution of the molecular weights with intervals of one
phthalocyanine ring’s molecular weight and an approximate
center at two/three rings for 2, 3, and 5. The observation above
might be attributed to the different desorption degree and
splitting/dissociation pathways of the hyperbranched struc-
tures. This made the assignments of the individual peaks
difficult. However, the above results agreed well with the GPC
ones for 1. It is worth noting that the MALDI-TOF MS spectra
for 1 and 4 only showed the monomer information. This likely
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 547

View Article Online
Paper
RSC Advances
originated from the poor desorption of the samples which
possess strong intermolecular ring–ring interactions in the
solid, including hydrogen bonding and p–p stacking of their
extremely planar structures.17 Due to the possible presence of
aggregates and observation errors, the UV-visible absorption
extinction coefficient measurements16 of 1–5 in DMF demon-
strated that there were five to eight phthalocyanine rings on
average which were likely overestimated in their structures.
tape with punched round holes was used as spacer to control
the film thickness. The films were sintered at 100 uC for 15
min and then at 475 uC for 45 min. Right before the
temperature decreased to 80 uC, the TiO2 films were immersed
into the dye solution and kept at room temperature for 12 h.
The quartz plates were then taken out, flushed with anhydrous
DMF followed with absolute ethanol, and dried by nitrogen
flow and then under vacuum in an oven at y50 uC for 6 h
before test.
1
For 1: H NMR (400 MHz, DMSO-d6, d (ppm)): 8.27–8.32 (d,
HOOCPh–H), 7.95 (s, Pc–H), 7.44–7.57 (m, HOOCPh–H), 7.16–
7.23 (m, Pc–H), 6.86–6.98 (m, Ar–H). IR (KBr, cm21): 3300,
3080, 1880, 1800, 1720, 1600, 1480, 1280, 1220, 1120, 1090,
1020, 970, 880, 740. GPC (relative to polystyrene standards):
Mw = 5748.8, Mn = 3486.1, Mw/Mn = 1.65.
The measurements were carried out using the same system
and procedures as described in detail in our previously
reported work.14,18,19 The instrument response function (IRF)
was estimated to be y250 fs at full width at half maximum.
The fluorescence data were fitted with a multi-exponential
decay/rise model, in which the fluorescence signal F(t) can be
theoretically expressed by a convolution of the IRF r(t) with a
molecule-response function f(t),
1
For 2: H NMR (400 MHz, DMSO-d6, d (ppm)): 8.00–8.20 (d,
HOOCPh–H), 7.78 (s, Pc–H), 7.49–7.51 (d, HOOCPh–H), 7.48–
7.49 (s, HOOCPh–H), 7.35–7.36 (d, Pc–H), 7.26–7.27 (d, Pc–H),
7.15–7.19 (dd, Ar–H), 7.03 (d, Ar–H), 6.97 (s, Ar–H), 6.92–6.93
(d, Ar–H). IR (KBr, cm21): 1880, 1800, 1720, 1610, 1490, 1280,
1220, 1125, 1080, 970, 895, 800, 760. MS: m/z 2033.50, 2361.95,
2781.94, 3200.90, 3649.47, 3915.29, 4832.10.
ð
?
F(t)~
r(t{t)f (t)dt
0
For 3: 1H NMR (400 MHz, DMSO-d6, d (ppm)): 8.06–8.15 (br,
HOOCPh–H), 7.55–7.65 (br, Pc–H), 7.5 (s, HOOCPh–H), 7.46–
7.47 (d, HOOCPh–H), 7.23–7.24 (d, Pc–H), 7.16–7.19 (dd, Ar–
H), 7.13–7.14 (d, Pc–H), 7.01–7.02 (d, Ar–H), 6.96 (s, Ar–H),
6.91–6.92 (d, Ar–H). IR (KBr, cm21): 3400, 3100, 1860, 1790,
1720, 1600, 1480, 1280, 1220, 1120, 1100, 980, 890, 800, 760.
MS: m/z 2056.04, 2445.53, 2845.93, 3264.81, 4068.79, 4906.68.
where r(t) is a Gaussian function with laser pulse width and f(t)
is given by:
ꢀ
ꢁ
X
t
f (t)~
Ai exp {
ti
i
where the factor Ai represents the relative weight (or
amplitude) of corresponding component, whose sign can
distinguish rising or decay process; and ti is the rising or decay
time constant.
1
For 4: H NMR (400 MHz, DMSO-d6, d (ppm)): 8.21–8.23 (d,
HOOCPh–H), 7.75 (s, Pc–H), 7.40–7.58 (m, HOOCPh–H), 7.31–
7.36 (m, Pc–H), 7.15–7.20 (d, Pc–H), 7.03–7.11 (m, Ar–H), 6.89–
6.91 (s, Ar–H). IR (KBr, cm21): 3400, 3080, 1900, 1810, 1730,
1600, 1480, 1420, 1280, 1220, 1140, 1100, 1060, 980, 900, 740.
For 5: 1H NMR (400 MHz, DMSO-d6, d (ppm)): 12.7–14.2 (br,
PhCOO–H), 7.80–7.85 (d, HOOCPh–H), 7.51 (s, HOOCPh–H),
7.49–7.50 (d, HOOCPh–H), 7.31–7.35 (dd, Ar–H), 7.27 (s, Pc–H),
7.19–7.20 (d, Pc–H), 7.15–7.17 (d, Pc–H), 7.03 (d, Ar–H), 6.98 (d,
Ar–H), 6.96 (s, Ar–H). IR (KBr, cm21): 3400, 3200, 3050, 1870,
1795, 1710, 1600, 1480, 1280, 1220, 1120, 1050, 980, 890, 740.
MS: m/z 2095.20, 2441.19, 2880.05, 3317.73, 4967.82.
Electrochemical cyclic voltammetry measurements
Electrochemical cyclic voltammetry measurements were per-
formed on a VersaSTAT 3 electrochemical working station
(Princeton Applied Research). The cell comprised inlets for a
platinum working electrode (MF-2013) of 1.6 mm in diameter
and a platinum wire counter electrode. The reference electrode
was an Ag/AgCl electrode (a 3.5 M KCl aqueous solution). Its
potential was internally calibrated using a ferrocene/ferroce-
nium (Fc/Fc+) redox couple [E1/2(Fc/Fc+) = +567 mV]. Typically,
a 0.2 M solution of [NBu4][PF6] (tetrabutylammonium hexa-
fluorophosphate) in anhydrous DMF containing sample (y3
mg mL21) was purged with nitrogen for at least 10 min, and
then the voltammograms were recorded at ambient tempera-
Femtosecond time-resolved fluorescence measurements
A femtosecond fluorescence upconversion (FFU) technique
was employed to investigate the fluorescence dynamics of the
solutions of 1–5 and the films of 5/TiO2. The solutions were
prepared with anhydrous, deoxygenated DMF at a concentra-
tion of y1.0 mg mL21. One-minute ultrasonication treatment
was performed on the solutions to accelerate the dissolution
process. The solutions were kept sealed and still at room
temperature for at least 2 days. The clear solutions over the
precipitate were then collected for test or for further use.
Quartz plates were used as substrates to prepare the solid
films of 5/TiO2. The quartz plates were cleaned in an
ultrasonic bath sequentially by hot detergent, hot deionized
water, acetone, and isopropyl alcohol, each for 10 min, and
then dried by nitrogen flow. Transparent single-layered
mesoporous TiO2 (y5 mm thick) was coated by applying a
nanoparticle TiO2 paste (Ti-Nanoxide HT/SP, Solaronix) onto
the quartz plates using a doctor blade technique. Scotch Magic
ture. The scan rate was 50 mV s21
.
Device fabrication and characterization5–7
Fluorine-doped tin dioxide (FTO, with a thickness of y400 nm
and a sheet resistance of 8 V/sq) pre-coated glass purchased
from Hartford Glass Co. was used as the substrate for cell
fabrication. FTO glass substrates were cleaned using the same
procedures as for quartz plates. Dye solutions were prepared at
a concentration of 0.2 mg mL21 using the same procedures as
for those prepared for the fluorescence dynamics test. A
compact TiO2 underlayer was deposited by spin coating at
3000 rotations per minute (RPM) for 30 s from a 0.1 M solution
of titanium diisopropoxide bis(acetylacetonate) in ethanol on
the FTO glass substrates followed by sintering first at 100 uC
548 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
for 15 min and then at 450 uC for 45 min. A transparent single-
or double-layered mesoporous TiO2 was coated by applying the
nanoparticle TiO2 paste (Ti-Nanoxide HT/SP, Solaronix) onto
the compact TiO2 using a doctor blade technique. Scotch
Magic tape with punched round holes was used as spacer to
control the film thickness. The film was sintered at 100 uC for
15 min and then at 475 uC for 45 min. This method yielded a
film thickness of y5 mm for the single layer and y10 mm for
the double layer. A scattering TiO2 layer (y4 mm thick) was
made by doctor blading another anatase paste (WER4-O,
Dyesol) on the mesoporous TiO2 using the same procedures as
above. The TiO2 was sintered at 100 uC for 15 min and then at
500 uC for 45 min. The quality of these TiO2 films for
photoanodes was improved by further deposition of TiO2 from
aqueous TiCl4 solution. A 0.2 M TiCl4 aqueous solution was
freshly diluted from a 1 M TiCl4 stock solution which was
prepared at 0 uC to prevent precipitation of TiO2 due to the
highly exothermic hydrolysis reaction. The dilute TiCl4
solution was applied onto the electrodes (50 mL cm22). The
electrodes were left overnight at room temperature in a closed
chamber, spun at 1000 RPM for 20 s, and immediately washed
with distilled water while wet. After further rinsing with
absolute ethanol, the TiO2 electrodes were dried by nitrogen
flow. After being fired again at 450 uC for 40 min in air, and
right before the temperature of the TiO2 electrodes decreased
to 80 uC, these films were immersed into the dye solutions and
kept at room temperature for 12 h. The films were taken out,
flushed with anhydrous DMF followed with absolute ethanol,
and dried by nitrogen flow. The photoanodes were then dried
under vacuum in an oven at y50 uC for 6 h. An electrolyte
solution was prepared by dissolving 0.6 M 1-butyl-3-methyli-
midazolium iodide, 0.04 M iodine, 0.025 M LiI, 0.05 M
guanidinium thiocyanate, and 0.28 M tert-butylpyridine in a
15/85 (v/v) mixture of valeronitrile and acetonitrile. Counter
electrodes were prepared by sputter deposition of 40 nm thick
films of Pt on pre-cleaned FTO glass substrates using a CrC-
150 Sputtering System. A parafilm mask was used as both
spacer and sealing material for the two electrodes. The two
electrodes were connected by the parafilm mask at 80 uC in air
on a hotplate platform. The electrolyte solution was injected
from two pre-cut channels of the mask and the final cells were
sealed by hot glue. The active area was 0.31 cm2 for all the
cells.
Table 1 Steady-state UV-visible absorption and fluorescence data for 1–5 in
anhydrous DMF
a
b
Sample
lmax (nm)
lem (nm)
1
2
3
4
5
330
355
339
342
352
—
363
—
—
—
615
616
603
612
615
634
—
—
—
—
666
686
668
676
676
699
—
—
—
—
—
—
427
433
—
707
699
706
—
784
765
—
—
755
690
a
b
Absorption peak maximum. Emission peak maximum.
with medium vibronic bands at 606 to 616 nm, all of which are
corresponding to the characteristic p–p* transition of mono-
meric phthalocyanines.20–22 Furthermore, the position of both
Soret and Q bands absorption peaks of 1–5 showed depen-
dence on the nature of the central metal ions. A red shift of
both the Soret and Q bands was observed when the metal
center was changed in the sequence of Co–Cu–Zn–AlCl. A same
trend was previously observed by Liu et al. from a series of
water-soluble sodium salts of sulfonated phthalocyanines
[MPc(SO3Na)4, M = H2, AlCl, Co, Cu, Zn].21 The sharpness of
the absorption peaks indicates the non-aggregating nature of
the phthalocyanine rings in these hyperbranched structures in
DMF solution.22 With strong absorption in the UV and red
region, 1–5 showed a window with very weak absorption at
from 400 to 600 nm, which may be utilized for some practical
photovoltaic applications, such as solar cells for a green-
house.6,11 This absorption gap may also be complemented by
other visible light harvesting materials for more efficient
photovoltaic cells.7
Steady-state fluorescence was used to provide information
for understanding the photo-generation of charge carriers and
intra/intermolecular energy transfers within 1–5.14 Fig. 2
shows the emission and excitation spectra of 1–5. Upon
excitation at the Soret band of 400 nm, 1, 2, and 5 individually
exhibited a strong Q band emission peak at 690 to 707 nm
along with a medium Q band emission peak at 755 to 784 nm.
The Q band emission for phthalocyanines corresponds to the
Current–voltage characteristics were measured using an
Agilent 4155C Semiconductor Parameter Analyzer under 1 sun
air mass (AM) 1.5 global illumination. Incident photon to
current conversion efficiency (IPCE) was recorded using
Newport’s QE/IPCE Measurement Kit with a 74125 Oriel
Cornerstone 260 J m Monochromator.
Results and discussion
Steady-state absorption and fluorescence
The steady-state absorption and fluorescence spectra of 1–5
were recorded in anhydrous DMF and the data are compiled in
Table 1. As a quick characterization, the absorption spectra of
1–5 (Fig. 1) displayed intense Q-band peaks at 666 to 699 nm
Fig. 1 UV-visible absorption spectra (normalized to their own Q band maxima)
of 1–5 in anhydrous DMF.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 549

View Article Online
Paper
RSC Advances
(T1). This is consistent with the findings from the femtosecond
time-resolved fluorescence measurements. The sharpness of
the emission peaks indicates that it is the individual
phthalocyanine ring which fluoresces.24 The very strong Q
band emission peaks for 1, 2, and 5 indicate that each
phthalocyanine ring is isolated from each other in one
hyperbranched molecule, which inhibits quenching due to
self-assembly.25 From Fig. 2, it was further noted that the Q
band emissions showed no excitation dependence, and that
the excitation spectra at corresponding emission maxima
approximately resemble the absorption spectra. Besides, the
excitation spectra profiles did not vary with the detecting
emission wavelengths. These results confirmed the fact that
there was a single emitting species, i.e. the phthalocyanine
core, in the hyperbranched structures.25
There is a great difference in the fluorescence spectra of 4 [a
CuPc], from those of the others. Q band emission of CuPc
derived from upconversion transfer resulting from a plasmon-
enhanced triplet–triplet annihilation of molecules was once
observed for a CuPc film deposited on Au (111) using
tunneling electrons from a scanning tunneling microscope.23
Other than this single observation under very unusual
conditions, Q band emission of CuPc has never been observed
by common emission analysis methods such as the steady-
state fluorescence spectroscopy due to the extremely low
intrinsic Q band fluorescence quantum yield from a CuPc
molecule.23,26 The fluorescence lifetimes of common fluores-
cent compounds including H2Pc, AlPc, ZnPc, are at the time
scale of 0.5 to 20 ns.27–29 Consistent with previous results,
strong Q band emission was detected for 1, 2, and 5 in solution
by the steady-state fluorescence spectroscopy. Al and Zn are
both non-transition metals without incomplete d shells, while
Co and Cu are transition metals with incomplete d shells.
When Co2+ or Cu2+ is coordinated with the four pyrrole
nitrogen atoms within the phthalocyanine ligand, the d orbital
electrons get involved in the electronic interaction between the
metal and the phthalocyanine p-conjugated system. For CuPc,
the open-shell paramagnetic Cu2+ ion facilitates spin–orbit
coupling. As a result, CuPc exhibits fast ISC with a time scale
of subpicoseconds to picoseconds,11,30 and thus very short Q
band fluorescence lifetime and very small fluorescence
quantum yield. In the present case, the Q band fluorescence
lifetime of 4 was determined to be 0.75 ps (corresponding to
the ISC from S1 to T1) by the femtosecond time-resolved
fluorescence method, which is three to four orders of
magnitude shorter than that of common fluorescent species
and thus beyond the detection capability of steady-state
fluorescence spectroscopy.27 This is the reason for the absence
of steady-state Q band emission for 4. As for 3 [a CoPc], the Q
band fluorescence lifetime of CoPc has been reported to be
y2.1 ps,31 which is longer than the Q band fluorescence
lifetime of 4, but much shorter than that of other metal
centered phthalocyanines (1, 2, and 5), therefore, very weak
steady-state fluorescence (both Soret band and Q band) was
observed for 4.
Fig. 2 Fluorescence spectra of 1–5 in anhydrous DMF. Excitation spectra, if
shown, were normalized to the maxima of their own emission spectra under the
excitation of Soret absorption peak.
S1–S0 transition. We can conclude that the energy from the
Soret band excitation was efficiently transferred to the Q bands
for 1, 2, and 5. Under the same conditions, not only a weak Q
band emission peak at 706 nm but also a broad weak Soret
band emission peak at y427 nm was observed for 3. In
comparison to 1, 2, and 5, this excitation seems more localized
on the Soret band, as may be attributed to the electron-
withdrawing properties of the Co center. Only a broad, weak
Soret band emission peak at y433 nm was detected for 4,
which is in line with previously published results for typical
CuPc.23 There was a very small portion of the energy from the
Soret band excitation contributing to the broad, weak
emission of 4. Most of the excitation energy was transferred
to the Q band and led away by a very quick channel, i.e. the
intersystem crossing (ISC) from singlet state (S1) to triplet state
It is worth noting that the absence of Q band emission (S1–
S0 transition) in steady-state fluorescence spectra doesn’t have
to mean the absence of S2–S1 transition. In fact, the Q band
550 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
emission signals of 4 have been detected by the following
femtosecond time-resolved fluorescence measurements.
emission at 480 nm, and the fitting parameters are compiled
in Table 2. As indicated by t1 in Table 2, the ultrafast decay
component of 102–494 fs with large amplitude for 1–5 is
assigned to intrasystem energy conversion from the excited
electronic state S2 to S1 (Soret band to Q band energy
transfer),11 which might be coupled with intrasystem energy
vibrational redistribution18,33 in the present cases. It is worth
noting that the conversion rate (1/t1) for the hyperbranched
metal-free phthalocyanine 1, is the slowest, while this rate for
AlCl, Co, Cu and Zn centered hyperbranched phthalocyanines
1–5 is at least 1.8 times as fast as that of 1. It seems that the
presence of metal centers in the hyperbranched phthalocya-
nines can significantly accelerate the ultrafast energy transfer
process from the Soret band to the Q band. The slow decay t2
in the time scale of 7.26–9.43 ps for 1–5 with small amplitude
might be designated to the lifetime of Soret band fluorescence.
Femtosecond time-resolved fluorescence
FFU as a powerful technique has been widely used to
investigate ultrafast processes, including excitation energy
migration,11,13,14,18,19,32,33 charge separation and transfer,11,18
and interfacial electron injection.11,32 In the present work, we
employed the FFU system to reveal the dynamics of ultrafast
photo-induced initial processes of 1–5 in solution and the
interfacial electron injection processes at the 5/TiO2 interface.
Both the Soret and Q bands fluorescence dynamics of 1–5 in
solution were investigated under the excitation of 400 nm. All
the fluorescence data were fitted well with bi-exponential
decays or decay/rise. Fig.
3 illustrates the Soret band
fluorescence dynamics of 1–5 with excitation at 400 nm and
Fig. 3 Soret band fluorescence dynamics of 1–5 in anhydrous DMF with excitation at 400 nm and emission at 480 nm. The red lines are the plotted IRF and the blue
ones are the fitted curves.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 551

View Article Online
Paper
RSC Advances
Table 2 Bi-exponential fitting parameters for the Soret band fluorescence
dynamics of 1–5 in anhydrous DMF with excitation at 400 nm and emission at
480 nm
Table 3 Bi-exponential fitting parameters for the Q-band fluorescence
dynamics of 1–5 in anhydrous DMF with excitation at 400 nm
Sample
t1 (fs)
A1 (%)a
t2 (ps)
A2 (%)
Sample
t1 (fs)
A1 (%)
t2 (ps)
A2 (%)
1b
36.0
156
—
32.0
70.0
245.4
253.2
—
266.0
244.2
296
196
—
0.75
66.0
54.6
46.8
—
34.0
55.8
2b
1
2
3
4
5
494
198
102
184
266
72.4
66.7
98.8
85.9
83.3
7.26
8.78
8.20
9.43
8.58
27.6
33.3
1.22
14.1
16.7
3b,c
4d
5d
a
b
The minus sign represents a rising behavior. With emission at
c
d
700 nm. No emission signal was detected. With emission at 690
nm.
However, the possibility of excitation energy transfer (EET)
which occurs between segments, and thermal equilibrium by
energy exchange with the solvent, in a time scale of a few ps to
several tens of ps,18,33,34 could not be excluded. The EET
process may be interpreted by a resonance energy transfer
model between the weakly coupled segments.33
Fig. 4 displays the Q band fluorescence dynamics of 1–5
except 3 with excitation at 400 nm, and the fitting parameters
are listed in Table 3. It is worth noting that there were
extremely weak emission signals in the region of from 700 to
800 nm (Q band emission) with unidentified characteristics
for HBCoPc–COOH (3), as might further prove that the
excitation is more localized on the Soret band. Also, there
might be ultrafast energy relaxation in very quick channels
within the IRF or other mechanisms induced by a high-
intensity laser beam such as the possible metal-to-ligand
charge transfer process in 3. For the Q band emission of other
hyperbranched phthalocyanines (1–5 except 3), the ultrafast
rising time of 32.0 to 156 fs represents the intrasystem
conversion from the Soret band to the Q band, whose time
scale should be consistent with the ultrafast decay time for
Soret band emission. For the hyperbranched copper phthalo-
cyanine (4), a relatively slow decay (t2) of 0.75 ps appeared
immediately after the rising time for Q band emission. This
relaxation path can be assigned to the ISC from S1 to T1. This
is in good agreement with the results for another hyper-
branched copper phthalocyanine in solution.11 For the H2,
AlCl and Zn centered hyperbranched phthalocyanines (1, 2,
and 5) in solution, a relatively slow decay t2 at the time scale of
66–296 ps appeared in their Q band emission dynamics right
after the ultrafast rising time. It can be attributed to the
relaxation of the first singlet state, as the existence of a lower
Fig. 4 Q band fluorescence dynamics of 1–5 except 3 in anhydrous DMF with excitation at 400 nm. The red lines are the plotted IRF and the blue ones are the fitted
curves.
552 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
Fig. 5 Anisotropy decay profiles for the Q-band emissions of 1, 2 and 5 in anhydrous DMF with excitation at 400 nm. The red lines are the plotted IRF.
excited singlet state for 1, 2, and 5 has been confirmed by their
steady-state emission peaks at longer wavelengths as shown in
Fig. 2. This process might include the EET process.18,33
Compared to the metal-free hyperbranched phthalocyanine
1, the existence of the metal centers of AlCl and Zn accelerated
this relaxation process.
As shown in Fig. 5, the anisotropy for the Q band
fluorescence dynamics of 1, 2 and 5 is almost a constant of
y0.2, which indicates the symmetry breaking in these
delocalized systems and that the excitation of one ring might
be transferred to other rings.17 Otherwise, with the exciton
located on one ring, a residual anisotropy value of 0.1 in such a
planar structure should be expected.17,35 This provides
evidence for the electron hopping process in these hyper-
branched systems, because the symmetry breaking could
possibly induce more traps.17
The femtosecond time-resolved fluorescence of 5/TiO2 film
was also investigated to reveal the interfacial electron injection
kinetics. Under the Soret band excitation at 400 nm, both the
Soret and Q bands fluorescence dynamics of 5/TiO2 film were
recorded. The fluorescence data were fitted well with bi-
exponential decays. Fig. 6 shows both the Soret band (480 nm)
Fig. 6 Fluorescence dynamics of 5/TiO2 film with excitation at 400 nm. The red lines are the plotted IRF and the blue ones are the fitted curves.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 553

View Article Online
Paper
RSC Advances
region, and the reductive cyclic voltammograms of 1–5 are
shown in Fig. 7. Note that no remarkable oxidation waves were
observed in the positive potential range, and instead, only
small bumps in the positive potential range were found. The
positions of these small bumps were in approximate agree-
ment with the corresponding onset oxidation potentials which
were estimated from a combined analysis of the onset
reduction potentials with the optical band gaps of 1–5. The
much smaller oxidation waves (bumps) on the cyclic voltam-
mograms may be tentatively interpreted by the hyperbranched
molecular structure of 1–5. The phthalocyanine rings in one
molecule of 1–5 were surrounded by non-conjugated bulky
substituents. As is known, the electrochemical oxidization is a
process occurring at the surface of the electrode.36 It might be
difficult for the phthalocyanine cores to reach to the electrode
and get oxidized there.36
For DSSCs, HOMO/LUMO energy levels of the dyes are
required to be appropriate in order for both effective electron
injection to the conduction band of TiO2 and effective electron
acceptance from the iodide/triiodide (I2/I32) electrolyte.3,37
For efficient electron injection into the conduction band of
TiO2, the LUMO energy levels of the dyes should be sufficiently
higher than the conduction band edge of TiO2.3,37 While for
efficient dye regeneration, the HOMO energy levels of the dyes
should be sufficiently lower than the potential of the
electrolyte redox.3,37 The hyperbranched phthalocyanines 1–5
displayed LUMO energy levels of 23.58 to 23.86 eV, which are
all higher than the conduction band edge (24.2 eV) of TiO2,38
demonstrating the possibility of efficient electron injection.
On the other hand, the HOMO energy levels of 1–5 are in the
range of 25.32 to 25.63 eV, all of which are lower than the
redox potential of I2/I32 (reported values range from 24.84 to
24.97 eV in organic solvents10), ensuring the possibility of
efficient electron transfer from the electrolyte.
Table 4 Bi-exponential fitting parameters for the fluorescence dynamics of 5/
TiO2 film with excitation at 400 nm
Emission wavelength
t1 (fs)
A1 (%)a
t2 (ps)
A2 (%)
480 nm
690 nm
241
454
86
76
3.1
3.6
14
24
and Q band (690 nm) fluorescence decay profiles of 5/TiO2,
and the fitting parameters are compiled in Table 4. For the
Soret band fluorescence dynamics, the ultrafast decay compo-
nent of 241 fs with large amplitude, as indicated by t1 in
Table 4, is assigned to the intrasystem energy conversion from
the excited electronic state S2 to S1,11 which might be coupled
with intrasystem energy vibrational redistribution.18,33 It is
worth noting that the conversion rate (1/t1) is faster than that
for 5 in solution, indicating a possibility of the co-existence of
the energy migration from S2 to S1 and the energy migration
from S2 directly to the conduction band of TiO2.11 The slow
decay t2 in a time scale of 3.1 ps with small amplitude might
still be designated to the lifetime of Soret band fluorescence or
EET process, which is faster than that in solution as well. For
the Q band fluorescence dynamics, the ultrafast decay process
(t1) in a time scale of 454 fs can be attributed to the interfacial
electron injection from the S1 state of the hyperbranched
phthalocyanine 5 to the conduction band of TiO2. The
relatively slow decay t2 appeared in the time scale of 3.6 ps,
which is much faster than that in solution, can still be
assigned to the relaxation of the first singlet state which might
be coupled with EET process between the rings. In summary,
efficient and ultrafast multi-phasic electron injection from
both the Soret and Q bands of the hyperbranched zinc
phthalocyanine 5 to the conduction band of TiO2 was
demonstrated.
Photovoltaic performance
Electrochemical properties
As the HOMO/LUMO energy levels of 1–5 were determined to
be appropriate for charge extraction, all the five hyper-
branched phthalocyanines were used as sensitizers of TiO2
in DSSCs. The J–V curves of the solar cells were illustrated in
Fig. 8 and Fig. 9(a). Table 6 summarizes the J–V characteristics,
including the short circuit current density (JSC), open circuit
voltage (VOC), fill factor (FF), and power conversion efficiency
(g) of the DSSCs under 1 sun AM 1.5 G illumination.
The electrochemical redox behaviors of 1–5 in anhydrous DMF
solution were investigated by cyclic voltammetry (CV). The
highest occupied molecular orbital (HOMO) and the lowest
unoccupied molecular orbital (LUMO) energy levels of 1–5 as
well as their band gaps were determined by CV in associate
with their UV-visible absorption spectra (Fig. 1) in DMF. All the
redox data are compiled in Table 5. All the five hyperbranched
phthalocyanines displayed reversible n-doping/de-doping
(reduction/reoxidation) processes in the negative potential
Table 5 Optical and redox properties of 1–5 in anhydrous DMF
Eogpt (eV)a
E
(V)b
E
(V)c
HOMOd (eV)
LUMOe (eV)
abs
onset
onset
red
onset
ox
Sample
l
(nm)
1
2
3
4
5
722
705
691
698
699
1.72
1.76
1.80
1.78
1.77
21.20
21.17
21.22
21.20
20.94
+0.52
+0.59
+0.58
+0.58
+0.83
25.32
25.39
25.38
25.38
25.63
23.60
23.63
23.58
23.60
23.86
a
b
c
The optical band gap was estimated from UV-visible absorption onset by Eogpt = 1240/l
(eV). Onset reduction potential vs. Fc/Fc+. Onset
abs
onset
d
e
onset
ox
onset
red
onset
ox
oxidation potential was calculated by E
LUMO = 2(E
= E
+ Eogpt
.
HOMO was calculated by HOMO = 2(E
+ 4.8) (eV). LUMO was calculated by
onset
red
+ 4.8) (eV).
554 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
Fig. 8 (a) J–V curves (empty: dark; filled: illuminated) of the DSSCs of 5
fabricated (a) with a single-layered mesoporous TiO2 (y5 mm), and without the
scattering TiO2 layer; (b) with a double-layered mesoporous TiO2 (y10 mm) and
a scattering TiO2 layer, under 1 sun AM 1.5 G illumination.
and a 5 mm thick mesoporous layer, exhibited a moderate
efficiency of 0.832%, with a VOC of 0.53 V, a JSC of 2.05 mA
cm22, and a FF of 0.767. For further optimization, devices with
a comprehensive TiO2 electrode consisting of a y100 nm thick
compact layer, a 10 mm thick mesoporous layer, and a 4 mm
thick scattering layer, with TiCl4 treatment, were fabricated for
dye-soaking. Compared to the simplified cell structure, the
comprehensive cell structure had a thicker mesoporous TiO2
layer (10 mm), and one extra scattering TiO2 layer (particle size
y400 nm). Moreover, the TiO2 film in the comprehensive
structure was treated with TiCl4 solution. A higher efficiency of
1.15% was obtained with a VOC of 0.491 V, a JSC of 3.20 mA
cm22, and an only slightly lower FF of 0.731. The impressive
56% increase of the JSC was attributed to the doubling of the
thickness of the mesoporous TiO2 layer which offered more
dye loading, as well as to the enhancement of light absorption
resulting from the addition of the TiO2 light scattering layer.39
A 0.04 V decrease of the VOC was observed as the thickness of
the TiO2 film was increased. This effect has been widely
observed for DSSCs based on TiO2 nanoparticles.38 It is due to
the enlargement of the surface area which provides additional
charge-recombination sites, and the enhancement of the dark
current which results from the direct reduction of the
electrolyte by the conduction band electrons of TiO2.40,41 It is
known that the major problem for using phthalocyanine dyes
to sensitize TiO2 is their intrinsic tendency toward aggrega-
tion.42 In comparison to a monolayer of dye molecules that is
directly connected to TiO2 surface, aggregated phthalocya-
nines are less effective to inject electrons to TiO2.42 Moreover,
the excess of outer dye molecules induce inefficient light
absorption.42 The aggregation dramatically reduces the power
conversion efficiency of phthalocyanine-based DSSCs.42 In the
present cases, both the efficiency and fill factor are as high as
those of some monomeric zinc phthalocyanine-sensitized
solar cells with 3a,7a-dihydroxy-5b-cholonic acid (CDCA) as a
co-adsorbent, indicating the non-aggregating nature of the
Fig. 7 Reductive cyclic voltammograms of 1–5 in anhydrous DMF.
The hyperbranched zinc phthalocyanine (5) was first
fabricated into DSSCs because 5 has been characterized the
most comprehensively, especially by the femtosecond time-
resolved fluorescence study of the 5/TiO2 films. The fluores-
cence dynamics study revealed ultrafast and efficient multi-
phasic electron injection from 5 to TiO2, from which a decent
photovoltaic performance of 5 in DSSCs was anticipated. The
device structure optimization was performed on 5 and then
the optimized structure was applied to all other hyper-
branched phthalocyanines. Therefore, it appears necessary to
describe the performance of 5 first. To facilitate the elucida-
tion of the effect of cell structure on the cell performance, the
J–V curves of 5 are shown in Fig. 8. Devices with a simplified
TiO2 electrode, consisting of a y100 nm thick compact layer
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 555

View Article Online
Paper
RSC Advances
Fig. 9 (a) J–V curves (empty: dark; filled: illuminated) and (b) incident photon to current conversion efficiencies (IPCE) of the DSSCs of 1–5 fabricated with a double-
layered mesoporous TiO2 and a scattering TiO2 layer under 1 sun AM 1.5 G illumination.
hyperbranched phthalocyanine molecules on the surface of
TiO2.9
consequence of the successful suppression of the aggregation
tendency of phthalocyanine rings on TiO2 surface by the
hyperbranched structure via non-conjugated bulky substitu-
ents.5 The non-conjugated bulky substituents surrounding the
The comprehensive cell structure which yielded relatively
high power conversion efficiency for 5 was adopted for the
fabrication of DSSCs of all other hyperbranched phthalocya-
nines 1–4. Fig. 9(a) summarizes the J–V curves of the DSSCs
fabricated from 1–5 with the same comprehensive cell
structure. Cells of the hyperbranched metal-free phthalocya-
nine 1 showed an efficiency of 0.234%, with a VOC of 0.434 V, a
JSC of 0.837 mA cm22, and a FF of 0.644. In contrast, DSSCs
fabricated from a previously reported sterically hindered
monomeric metal-free phthalocyanine dye showed no photo-
current due to a low excited singlet state compared with the
conduction band of TiO2.43 Cells of the hyperbranched
aluminum phthalocyanine chloride 2 displayed an efficiency
of 0.294%, with a VOC of 0.424 V, a JSC of 1.05 mA cm22, and a
FF of 0.662. Cells of the hyperbranched cobalt phthalocyanine
3 displayed a very low efficiency of 0.096%, with a VOC of 0.324
V, a JSC of 0.425 mA cm22, and a FF of 0.700. Cells of the
hyperbranched copper phthalocyanine 4 exhibited an effi-
ciency of 0.233%, with a VOC of 0.420 V, a JSC of 0.805 mA
cm22, and a FF of 0.688. The generally high fill factors,
ranging from 0.644 to 0.731 for cells of 1–5, are likely the
2
phthalocyanine rings may block their interaction with I3 to
some extent, resulting in a limited recombination enhance-
ment compared to N719 dye.5 The VOCs of these solar cells
varied in the range of 0.324 to 0.491 V, when the metal centers
in the hyperbranched phthalocyanines are changed. It is
known that the VOC of DSSCs corresponds to the difference
between the Fermi level of the electron in the solid and the
redox potential of the electrolyte.3,37 However, lower VOCs for
cells of 1–5 than those for the cells of N719 were observed. In
order to examine the origin of the lower VOC for a monomeric
zinc phthalocyanine dye namely PcS6 than for N719, Mori
et al. carried out electron lifetimes and densities measure-
ments with various amounts of adsorbed dyes.5 In their work,
only small difference was observed in the VOC-versus-electron
density plot, suggesting a negligible shift of the conduction
band edge position among the cells.5 Thus, they ascribed the
low VOC to the shorter electron lifetime caused by a smaller
blocking effect (less compact dye layer).5 Similar phenomena
have recently been observed in porphyrin-based DSSCs as
well.44,45 The smaller blocking effect should also be the reason
for the observed low VOC values in the present cases since the
relatively large hyperbranched structure of these dyes may
make it even more difficult to cover the whole surface of TiO2.
The VOC of the DSSC of 5 was the highest, likely due to the
more electron-donating nature of ZnPc which might have
provided a more intimate contact with the TiO2 surface
compared to those of other metal centered hyperbranched
phthalocyanines. The short circuit current density of the solar
cells diverges from a lowest value of 0.425 mA cm22 for the Co
counterpart 3 to a highest value of 3.20 mA cm22 for the Zn
counterpart 5. Such a remarkable result is probably due to a
significantly increased directional electron-transfer property of
the zinc phthalocyanine dye by the electron-pushing zinc
metal center, and the severely decreased directional electron-
transfer property of the cobalt phthalocyanine dye by the
Table 6 Parameters of the DSSCs fabricated with 1–5 and measured under 1
sun AM 1.5 G illumination
Cell area
(cm2)
JSC
IPCEmax
(%)
Dye
VOC (V)
(mA cm22
)
FF
g (%)
1a
2a
3a
4a
5a
5b
0.312
0.312
0.312
0.312
0.312
0.312
0.434
0.424
0.324
0.420
0.491
0.530
0.837
1.05
0.425
0.805
3.20
0.644
0.662
0.700
0.688
0.731
0.767
0.234
0.294
0.096
0.233
1.15
6.28
23.5
5.16
8.72
66.7
2.05
0.832
a
Cells have been fabricated with a double-layered mesoporous TiO2
b
(y10 mm) and a scattering TiO2 layer. Cells have been fabricated
with a single-layered mesoporous TiO2 (y5 mm) and without a
scattering TiO2 layer.
556 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013

View Article Online
RSC Advances
Paper
electron-withdrawing cobalt metal center, as compared to the
hyperbranched metal-free phthalocyanine 1.5 The JSCs of all
the solar cells including those for 5 are generally low. This is
eventually limited by the narrow absorption of these hyper-
branched phthalocyanines.
Conclusions
A series of nanostructured non-aggregated hyperbranched
phthalocyanines (1–5) have been synthesized, characterized,
and applied as TiO2 sensitizers into red to NIR photon
harvestable semi-opaque DSSCs with efficiencies of up to
1.15%. The photovoltaic properties of these materials have
been significantly affected by the metal centers. The changing
of the metal centers shifted the absorption peaks, resulting in
slightly different harvesting capabilities of the solar cells at
different wavelengths. The metal centers have changed the
HOMO/LUMO energy levels and affected the energy transfer
processes within the phthalocyanines, which induced dra-
matic differences in their photovoltaic performances in DSSCs.
The effects of the metal centers have also been observed in
porphyrin-based DSSCs.46–48 Zinc was found to be the best
metal center in the whole series for designing new hyper-
branched phthalocyanine-based DSSCs.
Notably, this work developed a facile avenue of utilizing
hyperbranched structure to solve the aggregation issue of
phthalocyanine-based dyes on TiO2 surface which has hin-
dered them from high photovoltaic performances in DSSCs.
Ultrafast and efficient multi-phasic electron injection from
both the Soret band and Q band of 5 to the conduction band of
TiO2 was demonstrated. DSSCs with higher power conversion
efficiency are anticipated by broadening the spectral response
range of hyperbranched zinc phthalocyanine-based dyes in the
visible region through chemical structure modification.
The photovoltaic performances of 1–5 as discussed above
are consistent with the observations from both the steady-state
and femtosecond time-resolved fluorescence studies. In
particular, the time-resolved fluorescence results have pro-
vided information to correlate with the photovoltaic perfor-
mances of these materials. First, the femtosecond time-
resolved fluorescence results provided an insight into the
energy flowing after photoexcitation. It has been demonstrated
that the excitons are delocalized in the non-aggregated
structure of 1, 2, and 5 by their ultrafast depolarization
behaviors (Fig. 5).17,35 The excitation with each ring seems to
have the same energy delocalization feature. For the three
materials, after Soret band excitation, energy can transfer
fluently from the Soret band to the Q band, and their Q band
fluorescences are strong and long-lived, indicating a long
lifetime of the excitons. This is usually a prerequisite for an
efficient charge separation/injection. While for 3 and 4, the
weak or no steady-state Q band emission and/or the ultrafast
ISC process restrained the efficiency of electron injection in
the devices. The ultrafast ISC process is highly competitive to
electron injection. Our previous study revealed that the low-
lying triplet-states of CuPc can hardly be harvested by TiO2 due
to the energy level mismatch.11 Since the triplet-states, which
hold a dominant share of the excitons, could not be effectively
harvested, the JSCs and thus the overall power conversion
efficiencies of the DSSCs sensitized by 3 would be low. For 4,
after photoexcitation, the energy is more localized around the
metal center and likely induces metal-to-ligand charge
transfer, resulting in inefficient electron injection. This should
be the reason for the worst photovoltaic performance in
DSSCs. Besides, the photo-oxidation of Co2+ to Co3+ might also
have dissipated the energy of the primary photoproducts
(charges). Second, ultrafast and efficient electron injection
processes have been demonstrated at the interface of 5/TiO2 by
femtosecond time-resolved fluorescence study, from which a
high IPCE value of the devices of 5 can be anticipated prior to
the device fabrication and testing.
Acknowledgements
Financial support for this work was provided by the NSF
(ECCS-0723114), NASA (NNX09AP67A), NSF DMR (1153514),
and SDSU EE PhD program.
References
¨
1 B. O’Regan and M. Gratzel, Nature, 1991, 353, 737.
2 A. Yella, H.-W. Lee, H. N. Tsao, C. Yi, A. K. Chandiran, M.-
K. Nazeeruddin, E. W.-G. Diau, C.-Y. Yeh, S. M Zakeeruddin
¨
and M. Gratzel, Science, 2011, 334, 629.
Fig. 9(b) shows the IPCEs of the DSSCs of 1–5 with the
comprehensive cell structure. The IPCE data are consistent
with the results from the absorption spectra as well as the J–V
characteristics. The sharpness of the IPCE peaks further
confirmed the non-aggregating nature of the phthalocyanine
rings on the TiO2 surface for 2, 3, and 5.9 With HBZnPc–COOH
(5), the IPCE at the maximum absorption of the Q band
reached the highest value of 66.7%. This is consistent with the
observation of efficient and ultrafast multi-phasic electron
injection at the 5/TiO2 interface from the femtosecond time-
resolved fluorescence studies. Further significant improve-
ment of the performance of the DSSCs based on hyper-
branched zinc phthalocyanines is possible, however, via
structure modification to broaden the spectral response range.
3 A. Hagfeldt, G. Boschloo, L. Sun, L. Kloo and H. Pettersson,
Chem. Rev., 2010, 110, 6595.
4 (a) H. Imahori, T. Umeyama and S. Ito, Acc. Chem. Res.,
2009, 42, 1809; (b) X. Li, H. Wang and H. Wu, Struct.
Bonding, 2010, 135, 229; (c) M. V. Martınez-Dıaz, G. de la
Torre and T. Torres, Chem. Commun., 2010, 46, 7090.
5 S. Mori, M. Nagata, Y. Nakahata, K. Yasuta, R. Goto,
M. Kimura and M. Taya, J. Am. Chem. Soc., 2010, 132, 4054.
6 P. Y. Reddy, L. Giribabu, C. Lyness, H. J. Snaith,
C . V i j a y k u m a r , M . C h a n d r a s e k h a r a m ,
M. Lakshmikantam, J.-H. Yum, K. Kalyanasundaram,
´
´
¨
M. Gratzel and M. K. Nazeeruddin, Angew. Chem., Int.
Ed., 2007, 46, 373.
7 J.-J. Cid, J.-H. Yum, S.-R. Jang, M. K. Nazeeruddin,
´
¨
E. Martınez-Ferrero, E. Palomares, J. Ko, M. Gratzel and
T. Torres, Angew. Chem., Int. Ed., 2007, 46, 8358.
This journal is ß The Royal Society of Chemistry 2013
RSC Adv., 2013, 3, 545–558 | 557

View Article Online
Paper
RSC Advances
8 L. Giribabu, C. V. Kumar, V. G. Reddy, P. Y. Reddy, C.
S. Rao, S.-R. Jang, J.-H. Yum, M. K. Nazeeruddin and
28 B. Ballesteros, S. Campidelli, G. de la Torre, C. Ehli, D.
M. Guldi, M. Prato and T. Torres, Chem. Commun., 2007,
2950.
¨
M. Gratzel, Sol. Energy Mater. Sol. Cells, 2011, 95, 306.
¨
9 J.-H. Yum, S.-r. Jang, R. Humphry-Baker, M. Gratzel, J.-
29 S. M. Bishop, A. Beeby, H. Meunier, A. W. Parker, M. S.
C. Foley and D. Phillips, J. Chem. Soc., Faraday Trans., 1996,
92, 2689.
J. Cid, T. Torres and M. K. Nazeeruddin, Langmuir, 2008,
24, 5636.
10 M. Wang, N. Chamberland, L. Breau, J.-E. Moser,
R. Humphry-Baker, B. Marsan, S. M. Zakeeruddin and
30 G. J. Dutton, W. Jin, J. E. Reutt-Robey and S. W. Robey,
Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 073407.
31 R. R. Millard and B. I. Greene, J. Phys. Chem., 1985, 89,
2976.
32 L. Luo, C.-F. Lo, C.-Y. Lin, I.-J. Chang and E. W.-G. Diau, J.
Phys. Chem. B, 2006, 110, 410.
33 X. Yan, T. Goodson III, T. Imaoka and K. Yamamoto, J.
Phys. Chem. B, 2005, 109, 9321.
34 O. Ohno, Y. Kaizu and H. Kobayashi, J. Chem. Phys., 1985,
82, 1779.
¨
M. Gratzel, Nat. Chem., 2010, 2, 385.
11 Y. Li, M. Yan, M. Jiang, R. Dhakal, P. S. Thapaliya and
X. Yan, J. Photonics Energy, 2011, 1, 011115.
´
12 M. K. R. Fischer, I. Lopez-Duarte, M. M. Wienk, M.
´
´
¨
V. Martınez-Dıaz, R. A. J. Janssen, P. Bauerle and
T. Torres, J. Am. Chem. Soc., 2009, 131, 8669.
13 Q. Wang, Y. Li, X. Yan, M. Rathi, M. Ropp, D. Galipeau and
J. Jiang, Appl. Phys. Lett., 2008, 93, 073303.
14 Y. Li, Y. Bian, M. Yan, P. S. Thapaliya, D. Johns, X. Yan,
D. Galipeau and J. Jiang, J. Mater. Chem., 2011, 21, 11131.
15 T.-W. Lee, Y. Kwon, J.-J. Park, L. Pu, T. Hayakawa and
M. Kakimoto, Macromol. Rapid Commun., 2007, 28, 1657.
16 M. Guo, X. Yan, Y. Kwon, T. Hayakawa, M. Kakimoto and
T. Goodson III, J. Am. Chem. Soc., 2006, 128, 14820.
17 M. Guo, X. Yan and T. Goodson III, Adv. Mater., 2008, 20,
4167.
18 Y. Xie, Y. Li, L. Xiao, Q. Qiao, R. Dhakal, Z. Zhang, Q. Gong,
D. Galipeau and X. Yan, J. Phys. Chem. C, 2010, 114, 14590.
19 X. Zhang, Y. Li, D. Qi, J. Jiang, X. Yan and Y. Bian, J. Phys.
Chem. B, 2010, 114, 13143.
20 Y. Zhang, P. Ma, P. Zhu, X. Zhang, Y. Gao, D. Qi, Y. Bian,
N. Kobayashi and J. Jiang, J. Mater. Chem., 2011, 21, 6515.
21 M. O. Liu, C. Tai, M. Sain, A. T. Hu and F. Chou, J.
Photochem. Photobiol., A, 2004, 165, 131.
22 J. M. Fox, T. J. Katz, S. V. Elshocht, T. Verbiest,
M. Kauranen, A. Persoons, T. Thongpanchang, T. Krauss
and L. Brus, J. Am. Chem. Soc., 1999, 121, 3453.
23 T. Uemura, M. Furumoto, T. Nakano, M. Akai-Kasaya,
A. Saito, M. Aono and Y. Kuwahara, Chem. Phys. Lett., 2007,
448, 232.
35 C. Galli, K. Wynne, S. M. LeCours, M. J. Therien and R.
M. Hochstrasser, Chem. Phys. Lett., 1993, 206, 493.
36 Y. Wei, Q. Zhang, Y. Jiang and J. Yu, Macromol. Chem. Phys.,
2009, 210, 769.
¨
37 M. Gratzel, J. Photochem. Photobiol., C, 2003, 4, 145.
38 F. D. Angelis, S. Fantacci and A. Selloni, Nanotechnology,
2008, 19, 424002.
39 P. Joshi, L. Zhang, Q. Chen, D. Galipeau, H. Fong and
Q. Qiao, ACS Appl. Mater. Interfaces, 2010, 2, 3572.
¨
40 E. Ghadiri, N. Taghavinia, S. M. Zakeeruddin, M. Gratzel
and J.-E. Moser, Nano Lett., 2010, 10, 1632.
41 D. Kuang, S. Ito, B. Wenger, C. Klein, J.-E. Moser,
¨
R. Humphry-Baker, S. M. Zakeeruddin and M. Gratzel, J.
Am. Chem. Soc., 2006, 128, 4146.
42 X. Wang and J. C. Yu, Macromol. Rapid Commun., 2004, 25,
1414.
43 S. Eu, T. Katoh, T. Umeyama, Y. Matano and H. Imahori,
Dalton Trans., 2008, 5476.
44 Y.-C. Chang, C.-L. Wang, T.-Y. Pan, S.-H. Hong, C.-M. Lan,
H.-H. Kuo, C.-F. Lo, H.-Y. Hsu, C.-Y. Lin and E. W.-G. Diau,
Chem. Commun., 2011, 47, 8910.
45 C.-L. Wang, C.-M. Lan, S.-H. Hong, Y.-F. Wang, T.-Y. Pan,
C.-W. Chang, H.-H. Kuo, M.-Y. Kuo, E. W.-G. Diau and C.-
Y. Lin, Energy Environ. Sci., 2012, 5, 6933.
46 C.-Y. Lin, C.-F. Lo, M.-H. Hsieh, S.-J. Hsu, H.-P. Lu and E.
W.-G. Diau, J. Ch. Chem. Soc., 2010, 57, 1136.
47 W. M. Campbell, A. K. Burrell, D. L. Officer and K.
W. Jolley, Coord. Chem. Rev., 2004, 248, 1363.
48 L.-L. Li and E. W.-G. Diau, Chem. Soc. Rev., DOI: 10.1039/
c2cs35257e.
24 A. Ogunsipe, D. Maree and T. Nyokong, J. Mol. Struct., 2003,
650, 131.
25 J. Leclaire, R. Dagiral, S. Fery-Forgues, Y. Coppel,
B. Donnadieu, A.-M. Caminade and J.-P. Majoral, J. Am.
Chem. Soc., 2005, 127, 15762.
26 P. S. Vincett, E. M. Voigt and K. E. Rieckhoff, J. Chem. Phys.,
1971, 55, 4131.
27 J. R. Lakowicz, Principles of Fluorescence Spectroscopy, 3rd
edn, Springer, Singapore, 2006.
558 | RSC Adv., 2013, 3, 545–558
This journal is ß The Royal Society of Chemistry 2013