Control of Recombination Pathways in TiO2 Nanowire Hybrid Solar Cells Using Sn4+ Dopants

Control of Recombination Pathways in TiO2Nanowire Hybrid Solar Cells Using Sn4+Dopants

James A.Dorman,?Jonas Weickert,?Julian B.Reindl,?Martin Putnik,?Andreas Wisnet,?

Matthias Noebels,?Christina Scheu,?and Lukas Schmidt-Mende*,?

?Department of Physics,University of Konstanz,Universita t sstr.10,Konstanz D-78457,Germany

?Department of Chemistry and Center for NanoScience(CeNS)Munich,Ludwig Maximilian University,Butenandtstr.11,81377 Munich,Germany

an undoped shell structure resulted in impressive e?ciency

TiO2-based device.Additionally,this device structure resulted in a

photovoltage decay measurements and impedance spectroscopy.

?shell systems to highlight the various e?ects of the Sn4+-doped

potential for application in the hybrid-type devices without the

structures due to the seamless interface of the metal oxide host for

material.

INTRODUCTION

While hybrid inorganic/organic based solar cells have been able to achieve impressive power conversion e?ciencies in the past few years,between13and15%for dye-sensitized1and mesostructured absorbers,2,3they are still limited due to many recombination processes that occur at the materials interface.The organic portion of these hybrid devices can vary from a dye electrolyte combination to a pure polymer or metal organic perovskites.However,the standard for the inorganic framework is a mesoporous titanium oxide(TiO2)?lm,1,2,4?6 providing a high-surface-area material for both light absorption and charge separation.Unfortunately,after the charges are separated,they must be rapidly transported away from the interface between the hole transporter material and TiO2 surface before recombination occurs.7Furthermore,the electron mobility of TiO2is relatively low,10?1cm2V?1s?1, for single-crystal anatase,8which is further decreased due to the high number of grain boundaries that the charges must cross. These low electron mobilities,and subsequent higher recombination probabilities,have been cited as one of the major limitations of using TiO2for all types of metal-oxide-based solar cells,9that is,hybrid,dye-sensitized,and perovskite-type solar cells.While the electron dynamics are a function of both hole and electron mobilities within the system,10meaning

the mobilities must be balanced for optimum device perform-

ance,11they must not be viewed as a simultaneous optimization

task.Therefore,it is important to understand the electron

dynamics within the TiO2system to reduce the e?ect of charge recombination while increasing the charge extraction e?ciency. An alternative to the mesoporous TiO2?lms that has been studied recently are rutile TiO2nanowire arrays12synthesized

on?uorine-doped tin oxide(FTO)using a hydrothermal

growth method,as?rst reported by Liu et al.13The growth of

these1D arrays are speci?c to FTO due to the crystal facets of

the substrate limiting the growth of the?lms along the[001]

axis.13Because of the single-crystalline nature of the nanowires,

the mobility of the electrons is on the order of1cm2V?1s?1.14

The increase in mobility is due to a lack of grain boundaries and

wire orientation,increasing the electron mobility while shuttling the electron to the electrode,respectively.15These

Special Issue:Michael Gra t zel Festschrift

Received:December26,2013

Revised:March14,2014

hydrothermally grown nanowires have diameters between80 and100nm,based on growth conditions,and scatter light very well,creating an opaque?lm on the FTO.13Despite their appealing properties,these nanowire arrays have been used in all di?erent types of hybrid solar cells with limited success thus far.Speci?cally,these arrays have been combined with poly(3-hexylthiophene-2,5-diyl)(P3HT)or CH3NH3PbI3perovskites and produced conversion e?ciencies of 1.5and9.4%, respectively.16,17Although the latter value is impressive,it is still signi?cantly lower than power conversion e?ciencies achieved on TiO2nanoparticle?lms,which exceed14%.2 Alternatively,a single-crystal,anatase,mesoporous thin?lm was combined with the perovskite absorber and showed conversion e?ciencies of7.3%.18Comparison of the various crystal structures shows that while the mobility of the structure is important for high-e?ciency solar cells,the recombination must be addressed for further improvements.

It is believed that the recombination processes can be reduced by controlling the position of the electrons within the nanowire arrays.First proposed by Law et.al,it is possible to direct the electrons toward the core of the nanowire by creating a cascading conduction band,where the core of the wire has a slightly lower conduction band than the shell.19To date, devices with these cascading energy levels have been produced using a highly conducting core,such as ZnO or SnO2,with a TiO2shell layer.19?22However,device performance has been limited for core?shell dye-sensitized solar cells with e?ciencies around2to3%.19,20,23An alternative approach to the formation of the cascading band structure is to dope the TiO2core with Sn4+,which is an attractive dopant due to the resulting high mobility,lower conduction band,increased electron density, and similar crystal structure.24This combination has allowed the fabrication of high-sensitivity gas sensors using a5mol% Sn concentration in TiO2.25More recently,Xu et.al have shown that the synthesis of Sn-doped TiO2nanowire arrays is possible using a modi?ed hydrothermal growth process, resulting in photocurrents greater than2mA cm?2.26However, no work has been performed on the Sn-doped TiO2core in solar-cell devices,speci?cally with a TiO2shell layer,forming this cascading conduction band and directing the electrons away from the interface.

This paper focuses on the impact of these cascading energy levels in the Sn-doped TiO2core?shell nanowires arrays on the charge recombination and overall photon conversion e?ciency of the devices.The TiO2and Sn:TiO2|TiO2core?shell nanowire arrays are synthesized using a two-step process consisting of a rutile core,via the hydrothermal growth,and a thin shell,deposited from a TiCl4bath.Electron?hole recombination is probed using multiple spectroscopic techni-ques to accurately elucidate the charge kinetics and dynamics for the nanowire arrays.A competing recombination pathway is proposed for the doped core?shell structure due to the cascading electronic structure to visualize how the recombina-tion processes are controlled.Finally,the e?ects of the Sn4+ dopant on the device performance are reported,resulting in a

signi?cant e?ciency improvement.

■EXPERIMENTAL METHODS

Hydrothermal Growth.TiO2nanowires were grown on FTO substrates(Solaronix,15Ω/□)cut into1.4×1.4cm2 squares.The FTO substrates were?rst ultrasonicated in water with dishwashing detergent,acetone,and isopropanol for30 min.Nanowire growth proceeded by placing the FTO substrates in a Te?on-lined autoclave with15mL of H2O,15 mL of concentrated HCl(Sigma-Aldrich),and525μL of titanium tert-butoxide(Sigma-Aldrich,>99%)after stirring for 15min.The autoclaves were placed in an oven for3h at180°C for nanowire growth.After the reaction was complete,the autoclaves were rapidly cooled in a water bath.The nanowire arrays were then removed from the solution and washed in DI H2O three times to remove any excess salt ions.Sn4+doping was achieved through the addition of tin tert-butoxide(Sigma-Aldrich,>99%)dissolved in concentrated HCl in a2:1ratio, with volumes between20and160μL for various dopant concentrations in solution.The Sn:HCl precursor was added directly to the growth solution,after the Ti precursor,using the same reaction conditions as previously described.The Sn:TiO2 growth performed here is slightly modi?ed from that reported by Xu et al.26Dopant concentrations referred to in this study are based on the mol percentage in the initial reaction solution because the concentration in the nanowires is too low for accurate measurements.

TiCl4Surface Treatment.To remove surface defects and form the core?shell architecture,a TiCl4(Sigma-Aldrich) surface treatment was performed after the hydrothermal growth.The thickness of the TiO2shell can be controlled by varying the TiCl4concentration in the growth solution.For this study,the TiCl4concentration was set at100mM to form a thick shell layer,which completely covered the wire surface.All samples were submerged in the TiCl4solution and placed in a water bath for a uniform heating pro?le.The solution was heated to70°C and allowed to react for3h.Finally,the shell was converted to the oxide form by annealing the wires at450°C for30min.The wire arrays were allowed to cool naturally over a period of～2h to reduce any stress that may form after rapid cooling.

Solar-Cell Fabrication.Solar-cell fabrication proceeded by immersing the surface-treated nanowire arrays in squaraine dye (SQ2)(Mitsubishi Chemicals)dissolved in ethanol(0.2M)for 3h.The SQ2dye was chosen to extract charge from both the interfacial and hole transport materials.27,28Excess dye was removed by rinsing the nanowire arrays with isopropanol and dried under a N2gas.The samples were then pretreated in chlorobenzene(Sigma-Aldrich)before a poly(3-hexylthio-phene-2,5-diyl)(P3HT,RMI-001EE Rieke materials),40mg/ mL in chlorobenzene,was spin coated.The P3HT-coated samples were heated to120°C for5min in ambient air to remove any excess solvent from the?lm.Finally,a130nm Ag ?lm was thermally evaporated through a shadow mask to complete the device fabrication.

Device Characterization.The structure and composition of the nanowire arrays were characterized using a scanning electron microscopy(SEM)(Zeiss Neon40EsB)operated at5 keV accelerating voltage.High-resolution transmission electron microscopy measurements were collected using a Jeol JEM 2011operated at200kV and a FEI Titan(S)TEM80?300 operated at300kV.Samples were suspended on a400-mesh carbon-coated TEM grid.Additionally,the crystal structure was characterized using the Al Kαemission line of an X-ray di?raction(XRD).Nanowire?lms were scanned at a rate of 1°/s for2θbetween20and70°.UV?vis measurements were performed in a Cary5000series UV?vis-NIR spectrometer (Agilent Technologies).Samples were placed in the center of an attached integrating sphere and masked using a Te?on sample holder with a0.78cm2hole for uniform sample illumination.The absorption spectra were scanned from800to

300nm at a rate of1nm s?1and slit size of5nm and referenced to100and0%transmission.Additionally,the spectra were normalized so that the absorption at800nm was 0.

Current density?voltage(J?V)and external quantum e?ciency(EQE)measurements were performed using a Keithley2400SourceMeter controlled through a self-written LabView program.Cells were illuminated via a LOT-Oriel LS0106solar simulator(AM1.5G,100mW cm?2)through a shadow mask with a resulting active area of0.125cm2.Light intensities were calibrated with a certi?ed Si reference solar cell (Fraunhofer Institute)with a KG5?lter.To control the illumination wavelength for the EQE measurements,light from a150W Xe lamp was passed through a LOT-Oriel Omni150 monochromator.All EQE spectra were normalized to the measured J?V current for accurate comparison.For the light-intensity measurements,the illumination light was passed through a series of neutral density?lters before illumination to obtain light intensities between10and95mW cm?2.

For photovoltage decay(PVD)and photocurrent decay (PCD)measurements,a pulsed laser(18Hz,532nm)was focused through the shadow mask onto the sample.The sample was background-illuminated with a LOT-Oriel LS0106solar simulator with variable light intensities(10?95mW cm?2). Signals were recorded with a Tekscope DPO7254digital oscilloscope.Termination resistances1MΩand50Ωwere used for PVD and PCD measurements,respectively.Data were ?tted with monoexponential decays to estimate the character-istic decay lifetimes.

Impedance spectroscopy(IS)measurements were conducted with an Ecochemie Autolab potentiostat/galvanostat.The samples were placed in a dark,grounded metal box,and the spectra were collected with and without a white light diode at ～100mW cm?2.Additionally,the measurements were performed at biases between?0.2to0.6V with a small AC perturbation using frequencies between0.1Hz and1MHz. The resulting spectra were modeled using ZView software

based on circuit model from Boix et.al.29

■RESULTS AND DISCUSSION

The nanowire arrays were synthesized using the hydrothermal method and measured to have diameters of roughly80?100 nm and lengths of～1μm after a3h growth period(Figure1a). Additionally,the nanowire arrays were indexed to the rutile crystal phase(JCPDS no.21-1276)using XRD(Figure1b). The wires are single crystalline and form a compact TiO2layer at the FTO surface before the wires are elongated in the[001] growth direction.Furthermore,no impurities can be seen in the EDX spectrum,indicating that high-quality nanowires were achieved.Finally,no changes in the nanowire structures were observed after the incorporation of low concentrations of the Sn precursor into the growth solution.A TEM image is shown

in Figure1c to highlight the nonconformal but completely encapsulated with up to a2nm TiCl4shell layer with the presence of nanoparticles around5nm in diameter.The thickness of this layer can be varied by adjusting the concentration and growth time.It is believed that this combined shell layer plays a key role in the device performance, discussed later,because the thin shell layer separates the core from the P3HT,forming an intermediate energy level in the conduction band,while increasing the surface area of the structure with the formation of the nanoparticles.However,the importance of the shell conformity and surface area is not fully understood and will be studied at a later date in combination with atomic layer deposition.The core and shell layer are indexed to rutile(3.24?)and anatase(3.51?,JCPDS no.21-1272)based on the lattice spacing,respectively.No di?erence was observed in the morphology of the Sn:TiO2and pure TiO2 nanowire structures after the hydrothermal growth.Because of the low concentration of incorporated Sn,as reported by Xu et. al for a pure SnCl4additive,it was not possible to measure the concentration of the Sn4+in the nanowire array.However, based on the XPS results reported by Xu,it is believed that

the Figure1.(a)Side view of the TiO2nanowire arrays with a height of ～1μm and diameters around80?100nm.(b)Nanowires can be index to the rutile crystal phase(JCPDS no.21-1276).The sample was scanned while still on the FTO substrate,which was also indexed for reference.(c)HRTEM image shows the TiO2coating that is achieved after the100mM TiCl4treatment.The?lm is nonconformal but covers the wires completely with thicknesses between1and2nm.The HRTEM image highlights the lattice planes of both the rutile core (3.24?,JCPDS no.21-1276)and the anatase shell layer(3.51?, JCPDS no.21-1272).

Sn concentration ranges between 0.1and 0.5mol %in the nanowires.26

First,to verify that the Sn dopant is important to the device performance,solar cells of pure TiO 2and SnTiO 2(5mol %)nanowire arrays were fabricated,with and without the TiCl 4shell.Nanowires arrays were decorated with SQ2for increased light absorption and in ?ltrated with P3HT to form the hybrid interface.The resulting J ?V curves,shown in Figure 2a,

indicate that the Sn 4+drastically increases the device perform-ance due to an increase in photocurrent,while the open-circuit voltage (V OC )is relatively unchanged for the uncoated samples.The pure TiO 2device resulted in the worst device performance due to low V OC and short-circuit current (J SC ),while the incorporation of the Sn 4+dopant increased the generated current while maintaining the measured V OC .The improvement in the J SC is attributed to an increase in charge density within the wires due to the Sn 4+dopant,which is shown in the IS results discussed later.However,the device performance can be further improved after the formation of a core ?shell Sn:TiO 2|TiO 2device,which produces currents around 7mA cm ?2and V OC values around 0.65V,which is in the range of mesoporous TiO 2?P3HT devices.28The increase in photocurrent is attributed purely to the Sn 4+because no structural change was observed,as mentioned previously.Furthermore,the increase was shown to be independent of the morphology based on the absorption spectra (Figure 2b).It can be seen in the Figure that there is little di ?erence between the SQ2absorption (550?720nm)and TiO 2absorption (<400nm)for both sample types.The band gaps for the two sample types,based on the absorption of a direct transition semiconductor (rutile TiO 2),are both calculated to be 3eV (not shown).After Sn 4+incorporation,there is no discernible change in the band gap,indicating that the dopant does not play a signi ?cant role in the composition of the material but instead assists in the charge transport of the nanowires.

Next,the concentration-dependent device performance was investigated to quantify the e ?ect of various Sn 4+amounts within the system.All devices from this point on have a TiO 2shell layer of up to 2nm deposited around the TiO 2nanowire core through the TiCl 4treatment because it was seen to be necessary to achieve increased V OC values while maximizing the J SC .The results from the AM1.5G J ?V measurements for the four device types are collected in Table 1,speci ?cally the characteristic J SC ,V OC ,?ll factor (FF),and power conversion e ?ciency (η)for each structure.The TiO 2|TiO 2device showed the lowest e ?ciency of ～1.9%due to a J SC that is roughly 2mA cm ?2lower,while the V OC and FF are within measurement error of the Sn:TiO 2|TiO 2samples.Alternatively,there is a peak in the performance of the Sn:TiO 2|TiO 2samples,with e ?ciencies of 2.5%or a 33%increase over the pure TiO 2|TiO 2structure,between 2.5and 5mol %,after which the J SC begins to decrease.The increase in e ?ciency is attributed to the increase in J SC ,with a maximum of 7.24mA cm ?2for the 5mol %sample.Additionally,it is unknown why the J SC at 10mol %doping is reduced,but it is likely due to the increased charge density producing a higher recombination rate across the TiO 2?P3HT interface.

To elucidate the origin of the generated charge carriers,we measured the EQE response as a function of the

excitation

Figure 2.(a)Measured J ?V curves for the model TiO 2and Sn:TiO 2(5mol %)nanowire arrays with and without a TiO 2shell layer.With the incorporation of Sn 4+into the system,the extracted current signi ?cantly increases as compared with the pure TiO 2system while maintaining the V OC that is typically for the TiO 2?P3HT material combination.(b)UV ?vis absorption spectrum for the TiO 2and Sn:TiO 2nanowire systems after decoration with SQ2.There is little di ?erence between the two systems,indicating that the di ?erence in generated current is due to the Sn 4+and not the dye loading.

Table 1.Collect Figures of Merit for the Core and Core-Shell Devices a

concentration J SC (mA/cm 2)

V OC (V)FF (%)η(%)ατRecom (s)PVD τRecom (s)IS PCD ?t (s)TiO 2|TiO 2

?5.620.66151 1.890.970 4.48×10?5 6.43×10?4 2.12×10?5Sn:TiO 2|TiO 2(2.5mol %)?7.000.64556 2.510.956 5.24×10?5 6.60×10?4 1.98×10?5Sn:TiO 2|TiO 2(5mol %)?7.240.63052 2.500.941 5.22×10?57.51×10?4 2.26×10?5Sn:TiO 2|TiO 2(10mol %)

?6.88

0.642

53

2.36

0.954

4.01

×10?5

7.08

×10?4

1.99

×10?5

a

Increase in short circuit current (J SC )is responsible for the improved e ?ciency (η)for the Sn:TiO 2|TiO 2samples,while the open circuit voltage (V OC )and ?ll factor (FF)are stable.Additionally,the device performance factor (α)and charge recombination lifetimes (τRecom )are shown for both PVD,at open circuit,and IS,at the maximum power point.Finally,the extraction times are included based on the PCD ?ts.

wavelength (Figure 3).In this Figure,three areas are highlighted to indicate the material in which the photon is

absorbed prior to charge generation,that is,the TiO 2(325?425nm),P3HT (425?600nm),and SQ2(600?750nm),based on the absorption data previously discussed.On direct comparison,it can be seen that the Sn 4+doping increases the quantum e ?ciency by a minimum of 5%over the full spectrum.Speci ?cally,for the 5mol %sample,where the largest J SC is generated,the core ?shell structures produces roughly a 25%increase in produced charge for the SQ2and P3HT regions,while the EQE of the TiO 2region jumps from 10to 29%.Furthermore,the TiO 2peak appears to be a function of the dopant concentration,increasing proportionally in magnitude and blue shifting with the Sn 4+concentration.It is interesting to note that this absorption peak is not seen for the undoped TiO 2|TiO 2arrays,while it is seen for all Sn 4+-doped samples,indicating that the charge generation of the TiO 2is related to the Sn 4+dopants.While it is not immediately clear why the charge generation occurs for the doped TiO 2and not the pure samples,one possible explanation could be that the Sn 4+dopants creates a n ?/n interface that can separate charge within the core ?shell structure,similar to that seen in Si-based solar cells at the p ?n interface.Finally,the increase across the visible spectrum is attributed to a bene ?cial energy coupling between the TiO 2and P3HT layers caused by the cascading energy levels that are present within the nanowires.This proposed cascading structure channels the electrons from the interface,where there is a high concentration of holes available for recombination,due to interfacial trap states produced through the disordered P3HT layer.28Therefore,while additional generation trapping mechanisms are occurring during illumination,it is believed that the increase in device performance is due to the cascading structure,reducing the charge recombination at the TiO 2?P3HT interface.

Next,the four di ?erent sample types were probed using intensity-dependent J ?V measurements to determine the extent to which the diodes are limited by recombination and space-charge buildup.This performance factor can be easily extracted using the following intensity-dependent equation

β∝α

I P (1)

where I is the measured current,P is the intensity of the excitation source,αis the diode performance factor,and βis a ?tting factor.By taking the log of both sides of eq 1,αcan be extracted from the slope of the line.The device performance factor decreases from 1as the recombination increases typically related to the introduction of defects into the system.The light-intensity-dependent measurements for the four device types are plotted and o ?set for clarity in Figure 4a with the four αfactors

shown in Table 1.It can be seen that after Sn 4+incorporation αdecreases from 0.97to between 0.94and 0.95.While these values only slightly vary,the trend does indicate the Sn 4+is playing a role in the intensity-dependent response,especially at the doping concentrations that are expected.Furthermore,αcan be used to discuss the monomolecular recombination because the population of electrons in trap states,under short circuit conditions,can be directly related to measured current.30On the basis of the decrease in α,it is assumed that the monomolecular recombination is increasing as Sn 4+

is

Figure 3.EQE spectra are shown for the di ?erent Sn 4+doping concentrations types.The spectra are divided into three wavelength ranges to highlight the current generated from the TiO 2,P3HT,and SQ2.An increase of at least 5%can be observed with the formation of the Sn:TiO 2|TiO 2core ?shell

system.

Figure 4.(a)Log ?log intensity dependent current response is shown for the various Sn 4+concentrations for intensities between 10and 95mW cm ?2.The curves are o ?set from their original position for readability.The device performance factors,which were extracted from the slope of each line,are tabulated in Table 1.(b)Extract PVD lifetimes are plotted against the incident light intensity,showing a decrease in lifetimes with increase in intensity.The maximum lifetime is obtained for the Sn:TiO 2|TiO 2(2.5mol %)structure due to the formation of a cascading energy level,causing the electrons to migrate toward the core of the nanowires.

incorporated into the system due to the presence of empty energy states below the TiO2conduction band.This is not apparent in the J?V curves,which show an increase in J SC in the presence of Sn4+because the increase in J SC can be associated with the increasing charge density and increased charge mobility associated with the Sn:TiO2core.

To quantify the recombination,PVD and PCD measure-ments were conducted under illumination intensities between 10and95mW cm?2.The recombination lifetime(PVD)and charge extraction time(PCD)were obtained from the decay curves.The collected data are tabulated in Table1,and the PVD data are plotted in Figure4b.It can be seen from the data that the recombination lifetime is dependent on the Sn4+ concentration,with the longest lifetimes measured for the2.5 mol%sample,roughly a13?17%increase over the pure TiO2| TiO2arrays depending on the light intensity.The fastest recombination is measured for the Sn:TiO2with10mol% doping,up to15%lower than the pure TiO2sample.These results indicate that there is a decrease in the amount of charges that are recombining across the TiO2?P3HT interface with lower doping concentrations.However,as the dopant concentration increases,the charge lifetime decreases due to increased recombination,suggesting that the dopant ions begin to cause stress within the lattice and increase the number of trap states that are available.While the PCD extraction times are roughly a factor of two faster than the recombination lifetimes,there is no speci?c trend in relation to the dopant concentration.On the basis of these measured values,it can be assumed that the cascading conduction band and increased core mobility decreases recombination probability due to the reduced probability of electrons being located near the TiO2?P3HT interface.

To con?rm the recombination lifetimes and further under-stand the e?ects of the doping and core?shell structure,IS measurements were performed on all devices.The bias across the device was incremented from?0.2to0.6V in0.05V steps under both light and dark conditions to elucidate the charge dynamics under various operating conditions.The impedance

curves were?t using a modi?ed circuit diagram based on the model by Boix et al.29(Figure5a).In this model,the time constant element consists of the recombination resistance (R Recom)and a chemical capacitance(Cμ),which can then be used to calculate a recombination lifetime(τRecom=R Recom*Cμ) for the illuminated device.Additionally,the Cμterm describes the charge density within the nanowires without incident light. Furthermore,a geometrical capacitance(C G)is included in a parallel circuit to describe the structure of the nanowire array and assist in?tting the model.The IS measurements were conducted to focus on the dynamics of the solar cell under working conditions that cannot be measured directly in the light intensity and decay measurements.Therefore,the recombination lifetime of the solar cells at the maximum power point(0.4V)is included in Table1.The IS modeling shows a similar trend in recombination lifetimes,as seen in the PVD measurements;speci?cally there is a maximum at5mol% Sn doping that is17%longer than the pure TiO2sample.The di?erence in magnitude between the two measurements is due to the applied bias voltages during measurements,that is,the open circuit and the maximum power points for the PVD and IS,respectively.This results in a longer recombination lifetime at the maximum power point because the charges are extracted and decreases their dwell time at the interfacial region.Finally, the Cμunder dark conditions was extracted to look at the inherent charge densities of the system(Figure4).The plot shows a proportional increase in the charge density,up to two times increase,with the amount of Sn4+added to the growth solution further con?rming the presence of the dopant within the system and the role it plays in the device performance. These results must be collected to form an energy level schematic to quantify the role of the Sn4+in the recombination process,as shown in Figure6.First,the basic TiO2?P3HT system is shown in Figure6a,highlighting the natural alignment between the two materials.Included in the Figure is the driving force of the system,and the theoretical V OC is the di?erence in energy between the HOMO of the P3HT and the conduction band.It is important to?rst summarize the results in a more qualitative manner,beginning with the J?V measurements and ?nishing with the IS data,to understand the e?ect of the dopant and shell structure.Speci?cally,two important conclusions were drawn from the data.First,the charge-carrier density is increased after the incorporation of Sn4+into the system due to the extra Sn4+electrons supplied to the system. This increased density causes a shift in the Fermi level to slightly higher energies,as shown in Figure6b.Second,the Sn4+ increases the monomolecular recombination through the addition of empty states below the conduction band,increasing the driving force for charge separation.The formation of these states below the conduction band is shown as a shaded

region Figure5.(a)All IS spectra were?tted using the circuit diagram,which is based on the model proposed by Boix et al.29The resistance across the interface(R Recom)and the capacitance(Cμ)are used to extract the recombination lifetime(τRecom)and charge densities under light and dark conditions.(b)Charge density(Cμ,dark)is shown for the various Sn4+doping concentrations.An increase is seen in the magnitude of the charge density with the incorporation of Sn4+.

in the Figure because the exact change in the conduction band is unknown.It is only when a shell layer is deposited that the cascading band structure is formed (Figure 6c).The empty states below the conduction band and the increased Fermi level cause the formation of a continuous cascading conduction band,maintaining the V OC due to the TiO 2shell layer.Alternatively,the charge driving force is dependent on the lower core conduction band,which forces the electrons to migrate toward the core where the mobility is higher,increasing charge extraction while reducing the probability of recombina-tion across the interface.Therefore,it is di ?cult to link the improved device performance directly to the cascading conduction band because this energetic structure directly a ?ects other aspects of the dynamics within the system.However,it can be concluded that the incorporation of the Sn 4+into the nanowire core plays an important role in the device,and without it the reduced recombination and increased current generation would not occur without this core ?shell architecture.

The recombination pathways of the electron ?hole pairs across the interface are also indicated on the band structures in Figure 6a ?c.For the TiO 2and an Sn:TiO 2,the recombination pathways are similar because both materials are in direct contact with the P3HT layer.Additionally,the two recombination pathways can be modeled using the following relationship:

τ

=

=?+v k 1

[e ][h ]MO (2)

where v is the rate of recombination,τis the recombination lifetime,k MO is the recombination constant with MO representing TiO 2or Sn:TiO 2,and [e ?]and [h +]are the concentration of electrons and holes in the metal oxide and P3HT layers,respectively.On the basis of this equation,it can be seen that the rate of recombination increases as the electron density increases.However,this is not seen for the Sn:TiO 2|TiO 2core ?shell sample,indicating that an alternative mechanism is occurring.Speci ?cally,with the deposition of the shell layer and the formation of the cascading conduction band,the recombination relationship follows the more complex mechanism shown here

τ=

=++??v h k x k 1

[](()[e ][e ])CS CS

Sn:TiO Sn:TiO TiO TiO 2222(3)

where the core recombination constant (k (x )Sn:TiO 2)is now a function of the distance between the TiO 2?P3HT interface based on the solution for quantum tunneling across a barrier,that is,k ∝e ?δx ,where δis a constant function of the energy barrier of the TiO 2shell layer.All other parameters are the same as in eq 2.Because of the cascading energy levels,the two recombination processes simultaneously compete for the holes in the P3HT layer,while the electron density is higher in the core than the shell,where the recombination decreases due to the thickness of the TiO 2shell layer.Unfortunately,the rate constants are di ?cult to calculate from the current data set,and more experiments are planned for future.However,the role of the core ?shell system in the recombination dynamics can be appreciated when the device performance,electron lifetimes,and recombination kinetics are considered simultaneously.Because of core ?shell Sn:TiO 2|TiO 2architecture,the nano-wires have a higher electron density that allows for more charge to be extracted from the system while producing the cascading conduction band and complicating the overall recombination kinetics.

■

CONCLUSIONS

In this work,core ?shell TiO 2|TiO 2and Sn:TiO 2|TiO 2nanowire arrays were grown via hydrothermal growth and TiCl 4surface treatment.Solar cells were fabricated by decorating the surface of the core ?shell structure with SQ2dye,to allow for the collection of charges from the dye and HTM,and in ?ltrated with P3HT.The Sn 4+was incorporated into the core of the wire to address two limiting factors for the current generation of hybrid-based devices,speci ?cally high recombination and low charge mobility.After the incorporation of Sn 4+into the nanowire core,a signi ?cant increase of 33%in device performance was observed primarily due to the increase in J SC .The UV ?vis spectra indicated the increase was solely due to the Sn 4+dopant as both sample sets showed almost identical absorption in the TiO 2and SQ2absorption ranges.Furthermore,the EQE spectra showed increased charge generation across the full absorption spectra after the addition of Sn 4+,with a signi ?cant increase in the TiO 2absorption range due to the activation of the metal oxide after dopant incorporation.Additionally,the Sn 4+does not alter the charge generation in the P3HT and SQ2layers.Instead,the dopant reduces the recombination and allows for more charges to be extracted once these materials are excited.Light-intensity measurements resulted in a decreasing device performance factor with the dopant ion due to increased monomolecular recombination caused by the addition of empty states below the TiO 2conduction band.However,the formation of a cascading conduction band in the core ?shell Sn:TiO 2|TiO 2increased the lifetime of the charges within the system by up to 17%,as determined using PVD spectroscopy.The trend in recombina-tion lifetimes was con ?rmed with IS spectroscopy,which resulted in an identical lifetime improvement,as with the PVD measurements.In addition to the lifetimes,the IS spectroscopy also indicated an increase in charge density after the addition of Sn 4+into the system,con ?rming the incorporation of the dopant ion.Finally,a recombination mechanism was proposed for the di ?erent systems to illustrate the e ?ect of the Sn 4+doping for core and core ?shell

systems.

Figure 6.Energy-level alignment diagrams are shown for (a)a pure TiO 2,(b)Sn:TiO 2,and (c)a core ?shell Sn:TiO 2|TiO 2nanowires device.The ?gures highlight the changing charge separation driving force (V DF ),the shifting Fermi energy (E F ),and the di ?erent recombination pathways that can occur across the interface.Two di ?erent mechanisms are shown for the Sn:TiO 2devices,where empty states are added below the conduction band due to the dopant,and the Fermi level is increased due to the increasing charge density,causing the conduction band to be lowered.These mechanisms are also seen in the core ?shell Sn:TiO 2|TiO 2system based on the electronic characterization that was performed.

AUTHOR INFORMATION

Corresponding Author

*Tel:+497531885409.E-mail:lukas.schmidt-mende@uni-konstanz.de.

Notes

The authors declare no competing?nancial interest.■ACKNOWLEDGMENTS

We acknowledge the support from the Alexander von Humboldt Foundation for the postdoctoral research fellowship, the German Research Foundation(DFG)for the projects “Identi?cation and overcoming of loss mechanisms in nano-580 structured hybrid solar cells?pathways to more e?cient devices”,“SPP1355:Elementary processes of organic photo-voltaics”,the“Nanosystems Initiative Munich”,and the REFINE research consortium funded by the Carl Zeiss Foundation.We would also like to thank Matthias Hagner for his support in the Nanostructure Laboratory at the University of Konstanz.

■REFERENCES

(1)Yella,A.;Lee,H.W.;Tsao,H.N.;Yi,C.;Chandiran,A.K.; Nazeeruddin,M.K.;Diau,E.W.G.;Yeh,C.Y.;Zakeeruddin,S.M.; Gra t zel,M.Porphyrin-Sensitized Solar Cells with Cobalt(II/III)?Based Redox Electrolyte Exceed12%Efficiency.Science2011,334, 629?634.

(2)Burschka,J.;Pellet,N.;Moon,S.-J.;Humphry-Baker,R.;Gao,P.; Nazeeruddin,M.K.;Gratzel,M.Sequential Deposition as a Route to High-Performance Perovskite-Sensitized Solar Cells.Nature2013, 499,316?319.

(3)Liu,M.;Johnston,M. B.;Snaith,H.J.Efficient Planar Heterojunction Perovskite Solar Cells by Vapour Deposition.Nature 2013,501,395?398.

(4)Hagfeldt,A.;Boschloo,G.;Sun,L.;Kloo,L.;Pettersson,H.Dye-Sensitized Solar Cells.Chem.Rev.2010,110,6595?6663.

(5)Chandiran,A.K.;Comte,P.;Humphry-Baker,R.;Kessler,F.;Yi,

C.;Nazeeruddin,M.K.;Gra t zel,M.Evaluating the Critical Thickness of TiO2Layer on Insulating Mesoporous Templates for Efficient Current Collection in Dye-Sensitized Solar Cells.Adv.Funct.Mater. 2013,23,2775?2781.

(6)Ferragut,R.;Aghion,S.;Moia,F.;Binda,M.;Canesi,E.V.; Lanzani,G.;Petrozza,A.Infiltration and Selective Interactions at the Interface in Polymer-Oxide Hybrid Solar Cells.J.Phys.Conf.Ser.2013, 443,012051.

(7)Weickert,J.;Dunbar,R.B.;Hesse,H.C.;Wiedemann,W.; Schmidt-Mende,L.Nanostructured Organic and Hybrid Solar Cells. Adv.Mater.2011,23,1810?1828.

(8)Tang,H.;Prasad,K.;Sanjines,R.;Schmid,P.E.;Levy,F. Electrical and Optical Properties of TiO2Anatase Thin Films.J.Appl. Phys.1994,75,2042?2047.

(9)Leijtens,T.;Lim,J.;Teuscher,J.;Park,T.;Snaith,H.J.Charge Density Dependent Mobility of Organic Hole-Transporters and Mesoporous TiO2Determined by Transient Mobility Spectroscopy: Implications to Dye-Sensitized and Organic Solar Cells.Adv.Mater. 2013,25,3227?3233.

(10)Huynh,W.U.;Dittmer,J.J.;Alivisatos,A.P.Hybrid Nanorod-Polymer Solar Cells.Science2002,295,2425?2427.

(11)Mandoc,M.M.;Koster,L.J.A.;Blom,P.W.M.Optimum Charge Carrier Mobility in Organic Solar Cells.Appl.Phys.Lett.2007, 90,133504.

(12)Feng,X.;Zhu,K.;Frank,A.J.;Grimes,C.A.;Mallouk,T.E. Rapid Charge Transport in Dye-Sensitized Solar Cells Made from Vertically Aligned Single-Crystal Rutile TiO2Nanowires.Angew.Chem. 2012,124,2781?2784.

(13)Liu,B.;Aydil,E.S.Growth of Oriented Single-Crystalline Rutile TiO2Nanorods on Transparent Conducting Substrates for Dye-Sensitized Solar Cells.J.Am.Chem.Soc.2009,131,3985?3990. (14)Hendry,E.;Koeberg,M.;O’Regan,B.;Bonn,M.Local Field Effects on Electron Transport in Nanostructured TiO2Revealed by Terahertz Spectroscopy.Nano Lett.2006,6,755?759.

(15)Zhu,K.;Neale,N.R.;Miedaner,A.;Frank,A.J.Enhanced Charge-Collection Efficiencies and Light Scattering in Dye-Sensitized Solar Cells Using Oriented TiO2Nanotubes Arrays.Nano Lett.2006, 7,69?74.

(16)Liao,W.-P.;Hsu,S.-C.;Lin,W.-H.;Wu,J.-J.Hierarchical TiO2 Nanostructured Array/P3HT Hybrid Solar Cells with Interfacial Modification.J.Phys.Chem.C2012,116,15938?15945.

(17)Kim,H.-S.;Lee,J.-W.;Yantara,N.;Boix,P.P.;Kulkarni,S.A.; Mhaisalkar,S.;Gra t zel,M.;Park,N.-G.High Efficiency Solid-State Sensitized Solar Cell-Based on Submicrometer Rutile TiO2Nanorod and CH3NH3PbI3Perovskite Sensitizer.Nano Lett.2013,13,2412?2417.

(18)Crossland, E.J.W.;Noel,N.;Sivaram,V.;Leijtens,T.; Alexander-Webber,J. A.;Snaith,H.J.Mesoporous TiO2Single Crystals Delivering Enhanced Mobility and Optoelectronic Device Performance.Nature2013,495,215?219.

(19)Law,M.;Greene,L.E.;Radenovic,A.;Kuykendall,T.;Liphardt, J.;Yang,P.ZnO-Al2O3and ZnO-TiO2Core-Shell Nanowire Dye-Sensitized Solar Cells.J.Phys.Chem.B2006,110,22652.

(20)Karlsson,M.;Jo g i,I.;Eriksson,S.K.;Rensmo,H.;Boman,M.; Boschloo,G.;Hagfeldt,A.Dye-Sensitized Solar Cells Employing a SnO2-TiO2Core-shell Structure Made by Atomic Layer Deposition. Chimia2013,67,142?148.

(21)Greene,L.E.;Law,M.;Yuhas,B.D.;Yang,P.ZnO-TiO2Core-Shell Nanorod/P3HT Solar Cells.J.Phys.Chem.C2007,111,18451.

(22)Kay,A.;Gra t zel,M.Dye-Sensitized Core?Shell Nanocrystals: Improved Efficiency of Mesoporous Tin Oxide Electrodes Coated with a Thin Layer of an Insulating Oxide.Chem.Mater.2002,14,2930?2935.

(23)Prasittichai,C.;Hupp,J.T.Surface Modification of SnO2 Photoelectrodes in Dye-Sensitized Solar Cells:Significant Improve-ments in Photovoltage via Al2O3Atomic Layer Deposition.J.Phys. Chem.Lett.2010,1,1611?1615.

(24)Tiwana,P.;Docampo,P.;Johnston,M.B.;Snaith,H.J.;Herz,L. M.Electron Mobility and Injection Dynamics in Mesoporous ZnO, SnO2,and TiO2Films Used in Dye-Sensitized Solar Cells.ACS Nano 2011,5,5158?5166.

(25)Zakrzewska,K.;Radecka,M.;Rekas,M.Effect of Nb,Cr,Sn Additions on Gas Sensing Properties of TiO2Thin Films.Thin Solid Films1997,310,161.

(26)Xu,M.;Da,P.;Wu,H.;Zhao,D.;Zheng,G.Controlled Sn-Doping in TiO2Nanowire Photoanodes with Enhanced Photo-electrochemical Conversion.Nano Lett.2012,12,1503?1508. (27)Weickert,J.;Auras,F.;Bein,T.;Schmidt-Mende,L.Character-ization of Interfacial Modifiers for Hybrid Solar Cells.J.Phys.Chem.C 2011,115,15081?15088.

(28)Weickert,J.;Zimmermann, E.;Reindl,J. B.;Pfadler,T.; Dorman,J.A.;Petrozza,A.;Schmidt-Mende,L.Synergistic Effects of Interfacial Modifiers Enhance Current and Voltage in Hybrid Solar Cells.APL Mater.2013,1,042109.

(29)Boix,P.P.;Lee,Y.H.;Fabregat-Santiago,F.;Im,S.H.;Mora-Sero,I.;Bisquert,J.;Seok,S.I.From Flat to Nanostructured Photovoltaics:Balance Between Thickness of the Absorber and Charge Screening in Sensitized Solar Cells.ACS Nano2011,6,873?880.

(30)van de Lagemaat,J.;Frank,A.J.Effect of the Surface-State Distribution on Electron Transport in Dye-Sensitized TiO2Solar Cells:Nonlinear Electron-Transport Kinetics.J.Phys.Chem.B2000, 104,4292?4294.