Improved efficiency of smooth and aligned single walled carbon nanotube/silicon hybrid solar cells† Xiaokai Li, a Yeonwoong Jung, b Kelsey Sakimoto, a Teng-Hooi Goh, a Mark A. Reed bc and Andr´ e D. Taylor * a Smooth and aligned single walled carbon nanotube (SWNT) thin films with improved optoelectronic performance are fabricated using a superacid slide casting method. Deposition of as made SWNT thin film on silicon (Si) together with post treatments result in SWNT/Si hybrid solar cells with unprecedented high fill factor of 73.8%, low ideality factor of 1.08 as well as overall dry cell power conversion efficiency of 11.5%. Broader context The unique optical, electrical and mechanical properties as well as solution processability of carbon nanotubes render it a promising candidate for the next generation electronic devices. The development of technologies that can simultaneously optimize several key and, in some cases orthogonal, parameters such as conductivity, transparency, morphology and mechanical properties is very challenging yet of vital importance. We have developed a superacid slide casting method to achieve that goal. In addition, studies that combine carbon nanotube with silicon, a well-characterized semiconductor, could provide valuable insight into how photo-generation, transport, and dissociation of excitons and charge carriers function in large ensembles of CNTs. Optimizing this interface could serve as a platform for many next generation solar cell devices including CNT/polymer, carbon/polymer, and all carbon solar cells. Introduction Carbon nanotubes (CNTs), especially single walled carbon nanotubes (SWNTs), have extraordinary properties such as high mobility (10 5 cm 2 V 1 s 1 ), 1 on-off current ratio (>10 5 ), 2 and current carrying capacity (>10 9 A cm 2 ). 3,4 Recent improvements in synthesis, 5 large-scale single chirality separation, 6 and rapid solution processing and assembly techniques 7,8 further improves the viability and versatility of CNTs in energy storage and conversion devices. In fact, CNT has already been used to contact many emerging solar materials. 9 Studies that combine CNTs with silicon, a well-characterized semiconductor, could provide valuable insight into how photo-generation, transport, and dissociation of excitons and charge carriers function in large ensembles of CNTs. 10 Indeed, optimizing this interface could serve as a platform for many next generation solar cell devices including CNT/polymer, 11 carbon/polymer, 12,13 and all carbon solar cells. 14–16 A CNT/Si hybrid solar cell typically consists of an n-type (or p-type) single-crystalline silicon coated by a thin lm of p-type 17 (or n-type) 18 CNTs. The transparent CNT lm functions as a charge carrier collecting conductive electrode and establishes a built-in potential 19 interfacing with Si, which separates photo-generated carriers to yield a photocurrent. p-CNT/n-Si hybrid solar cells have received more intensive study as CNTs naturally develop p-type characteristics when exposed to air. 20 Toward designing high efficiency SWNT/Si photovoltaic devices, the systematic optimiza- tion of the materials, the optoelectronic/morphological properties of the SWNT lms, and the SWNT/Si interfaces are essential. Single-walled carbon nanotubes (SWNTs) 21 present more advantages over doubled 19,22 (or multi-) 23 walled carbon nano- tubes owing to their tunable/direct band gap energies matching with a wide range of the solar spectrum and better charge carrier transport properties. 7,10,14,24 For heterogeneous mixtures of metallic and semiconducting SWNT lms, thinner and sparse lms are desirable in terms of optical properties since light absorbed in the carbon nanotube layer does not contribute to the power conversion efficiency due to the presence of metallic nanotubes which rapidly quench excitons in their vicinity. 16 Simultaneously, in terms of electrical properties, dense thick SWNT lms are more favorable to allow a larger diffusion length of minority carriers over the thickness of the SWNT lms via the percolated networks, 25 to minimize series resistance, 17 and to improve the effective p-SWNT/n-Si interfa- cial area. 26 a Department of Chemical & Environmental Engineering, Yale University, New Haven, Connecticut 06511, USA. E-mail: [email protected]b Department of Electrical Engineering, Yale University, New Haven, CT 06511, USA c Department of Applied Physics, Yale University, New Haven, CT 06520, USA † Electronic supplementary information (ESI) available: Microscope images, ideality factor calculation, sheet resistance and transmittance results, J–V characteristics in the semi-logarithmic scale at the forward linear region. See DOI: 10.1039/c2ee23716d Cite this: Energy Environ. Sci., 2013, 6, 879 Received 6th October 2012 Accepted 20th December 2012 DOI: 10.1039/c2ee23716d www.rsc.org/ees This journal is ª The Royal Society of Chemistry 2013 Energy Environ. Sci., 2013, 6, 879–887 | 879 Energy & Environmental Science PAPER Downloaded by Yale University Library on 15/04/2013 03:52:43. Published on 20 December 2012 on http://pubs.rsc.org | doi:10.1039/C2EE23716D View Article Online View Journal | View Issue
Transcript
Energy &Environmental Science
PAPER
Dow
nloa
ded
by Y
ale
Uni
vers
ity L
ibra
ry o
n 15
/04/
2013
03:
52:4
3.
Publ
ishe
d on
20
Dec
embe
r 20
12 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C2E
E23
716D
View Article OnlineView Journal | View Issue
aDepartment of Chemical & Environmental
Connecticut 06511, USA. E-mail: andre.taylbDepartment of Electrical Engineering, YalecDepartment of Applied Physics, Yale Univer
† Electronic supplementary informationideality factor calculation, sheet resistcharacteristics in the semi-logarithmic sDOI: 10.1039/c2ee23716d
Cite this: Energy Environ. Sci., 2013, 6,879
Received 6th October 2012Accepted 20th December 2012
DOI: 10.1039/c2ee23716d
www.rsc.org/ees
This journal is ª The Royal Society of
Improved efficiency of smooth and aligned singlewalled carbon nanotube/silicon hybrid solar cells†
Xiaokai Li,a Yeonwoong Jung,b Kelsey Sakimoto,a Teng-Hooi Goh,a Mark A. Reedbc
and Andre D. Taylor*a
Smooth and aligned single walled carbon nanotube (SWNT) thin films with improved optoelectronic
performance are fabricated using a superacid slide casting method. Deposition of as made SWNT thin
film on silicon (Si) together with post treatments result in SWNT/Si hybrid solar cells with
unprecedented high fill factor of 73.8%, low ideality factor of 1.08 as well as overall dry cell power
conversion efficiency of 11.5%.
Broader context
The unique optical, electrical and mechanical properties as well as solution processability of carbon nanotubes render it a promising candidate for the nextgeneration electronic devices. The development of technologies that can simultaneously optimize several key and, in some cases orthogonal, parameters such asconductivity, transparency, morphology and mechanical properties is very challenging yet of vital importance. We have developed a superacid slide castingmethod to achieve that goal. In addition, studies that combine carbon nanotube with silicon, a well-characterized semiconductor, could provide valuable insightinto how photo-generation, transport, and dissociation of excitons and charge carriers function in large ensembles of CNTs. Optimizing this interface couldserve as a platform for many next generation solar cell devices including CNT/polymer, carbon/polymer, and all carbon solar cells.
Introduction
Carbon nanotubes (CNTs), especially single walled carbonnanotubes (SWNTs), have extraordinary properties such as highmobility (�105 cm2 V�1 s�1),1 on-off current ratio (>105),2 andcurrent carrying capacity (>109 A cm�2).3,4 Recent improvementsin synthesis,5 large-scale single chirality separation,6 and rapidsolution processing and assembly techniques7,8 furtherimproves the viability and versatility of CNTs in energy storageand conversion devices. In fact, CNT has already been used tocontact many emerging solar materials.9 Studies that combineCNTs with silicon, a well-characterized semiconductor, couldprovide valuable insight into how photo-generation, transport,and dissociation of excitons and charge carriers function inlarge ensembles of CNTs.10 Indeed, optimizing this interfacecould serve as a platform for many next generation solar celldevices including CNT/polymer,11 carbon/polymer,12,13 and allcarbon solar cells.14–16
(ESI) available: Microscope images,ance and transmittance results, J–V
cale at the forward linear region. See
Chemistry 2013
A CNT/Si hybrid solar cell typically consists of an n-type (orp-type) single-crystalline silicon coated by a thin lm of p-type17 (orn-type)18 CNTs. The transparent CNT lm functions as a chargecarrier collecting conductive electrode and establishes a built-inpotential19 interfacing with Si, which separates photo-generatedcarriers to yield a photocurrent. p-CNT/n-Si hybrid solar cells havereceived more intensive study as CNTs naturally develop p-typecharacteristics when exposed to air.20 Toward designing highefficiency SWNT/Si photovoltaic devices, the systematic optimiza-tion of the materials, the optoelectronic/morphological propertiesof the SWNT lms, and the SWNT/Si interfaces are essential.
Single-walled carbon nanotubes (SWNTs)21 present moreadvantages over doubled19,22 (or multi-)23 walled carbon nano-tubes owing to their tunable/direct band gap energies matchingwith a wide range of the solar spectrum and better chargecarrier transport properties.7,10,14,24 For heterogeneous mixturesof metallic and semiconducting SWNT lms, thinner andsparse lms are desirable in terms of optical properties sincelight absorbed in the carbon nanotube layer does not contributeto the power conversion efficiency due to the presence ofmetallic nanotubes which rapidly quench excitons in theirvicinity.16 Simultaneously, in terms of electrical properties,dense thick SWNT lms are more favorable to allow a largerdiffusion length of minority carriers over the thickness of theSWNT lms via the percolated networks,25 to minimize seriesresistance,17 and to improve the effective p-SWNT/n-Si interfa-cial area.26
The successful fabrication and assembly of SWNT lms withthe aforementioned desired properties requires a special SWNTprocess that can maintain good control of the dispersedSWNTs. Current SWNT dispersion techniques are not idealbecause chemical functionalization disrupts the electronicproperties of the SWNTs27 and stabilization via polymers/surfactants introduces non-SWNT materials into the nalproduct.28 Sonication, widely used in the SWNT process, is alsonot appropriate as it reduces the length of the SWNTs, and thusinhibits the formation of the SWNT percolation network.29
Recently, we demonstrated that SWNT lms prepared by theMayer rod coating method possess optoelectronic propertiesthat compete with the best SWNT thin lms fabricated by anyother techniques so far.30 This superiority comes from the use ofa non-covalent stabilizer, carboxymethyl cellulose (CMC), whichdisperses the SWNTs during the Mayer rod process. Thesubsequent removal of the CMC stabilizer aer the lmformation minimizes the contact resistance between the indi-vidual SWNTs andmaintains the collective electronic propertiesof the SWNT network.
At the interface of SWNT/Si hybrid solar cells there are agreat multitude of tiny nanotube–silicon junctions whereSWNTs are in physical contact with the Si surface. Within thecarbon nanotube network, many SWNTs overlap and suspendon each other without contacting the Si. As a result, the solarcell performance suffers as surface recombination takes place atthe exposed bare Si surfaces.19,21 Wadhwa et al.31 introduced anionic liquid electrolyte between the SWNT networks thatinduces a depletion layer in the silicon. The electric eldaccumulates holes at the surface and also repels electrons.Photogenerated holes can thus diffuse along the Si with reducedsurface recombination prior to an encounter with a nanotubewhere they are collected. Jia et al.21 introduced the occupation ofthe bare Si surface between the porous SWNT network withnitric acid solution. The acid can induce an inversion layer inthe device leading to an enhanced performance.32,33 We suggesta more effective approach could be to fabricate a highly smoothand dense SWNT network to maximize the effective p–n junc-tion interface. Yet fullling this task without jeopardizing thetransparency and conductivity of the SWNT network has beenextremely challenging with few related advanced materialapproaches in the literature.
In this paper, we report the engineering of high-efficiencySWNT/Si solar cells by utilizing a sliding method on superaciddispersed SWNTs to create SWNT lms with the followingadvantages for photovoltaic applications: (a) combination ofhigh transparency and conductivity for improved light absorp-tion and carrier transportation; (b) extremely low surfaceroughness for a seamless interface with Si for efficient carrierdissociation; (c) mechanical robustness with high exibility andease of fabrication/deposition. Here we show a systematic studyon the performance of SWNT/Si solar cells using two distinctmethods (Mayer rod and superacid slide casting methods) forfabrication of SWNT lms. We also describe the effects ofvarious post-fabrication methods (i.e. doping) on the solar cellperformance, and present the device parameters of optimizedsolar cells. These SWNT/Si photovoltaic devices have a ll factor
880 | Energy Environ. Sci., 2013, 6, 879–887
(FF) of 73.8%, ideality factor of 1.08 and efficiencies for dry cellsup to 11.5%. For the wet cells, we suggest that the observedextraordinary high short circuit currents are due to lightbending from the liquid on the surface of the hydrophobicSWNT thin lms.
Experiment procedure
SWNT thin lm fabrication methods: superacid slide coatingmethod: 3 mL of chlorosulfonic acid (Fluka, purum 98%) isadded to 14 mg of single welled carbon nanotube (SWNT)(kindly supplied by SouthWest NanoTechnologies) and stirredvigorously for 3 days, inside a glove box lled with nitrogen. Adrop of the prepared SWNT ink is sandwiched between twoglass slides. The slides are manually pressed together until thedesired lm thickness is achieved (typically 30 to 50 nm). Thetwo slides are rapidly slid across each other in opposite direc-tions, producing a SWNT lm on each glass slide. Each slide iseither dried slowly inside a glove box or gently dipped intodeionized water to remove any acid leached out of the lm. Theresulting lm is oated on the top of the deionized waterallowing transfer to another substrate.
SWNT/Si device fabrication: 500 nm-thermal oxide coveredn-type Si (100) wafer (1–10 U cm) are patterned with Au (80 nm:top contact and etch mask)/Cr (5 nm: adhesion layer) byphotolithography. Si windows (3 � 3 mm2) are exposed by thewet-etch of the oxide. The back contacts are fabricated usingCr/Au, or Al following a buffered oxide etch (BOE) for 1 min. TheSWNT thin lms are transferred onto patterned Si wafers byoating on water and then, picked up by the wafers.
Results and discussion
The SWNT lms are fabricated by the superacid sliding methodin the following steps. Briey, the SWNTs are stirred in chlor-osulfonic acid, which dissolves the SWNTs by reversiblyprotonating their walls. The Coulombic repulsion between theSWNTs facilitates disaggregation by counteracting the strongintertube van der Waals attractions.34 Several drops of thissuperacid dispersion are sandwiched between two glass slideswhere they are spread and sheared (Fig. 1a). The lm thicknessis controlled by the dispersion amount and the force applied tospread the solution into a thin lm. As-made lms can then becoagulated using deionized water, or slowly dried in anhydrousair.
We show that SWNT lms prepared by the superacid slidingmethod are consistently more transparent than the Mayer rodSWNT lms of the same sheet resistance (Fig. 1b). To assess theoptoelectronic performance of SWNT network lms, we use agure-of-merit based on the ratio of the direct currentconductivity, sdc, to the optical conductivity, sac (typically at 550nm).35 A higher value of sdc/sac indicates a better photovoltaiclm35 and the superacid SWNT lms exhibit a sdc/sac value of11.2, while that for the Mayer rod SWNT lms is 6.48. This hugeimprovement is attributed in part due to the formation of abetter SWNT dispersion from superacid solution. Davis et al.34
presented Cryo-transmission electron microscopy of SWNTs in
This journal is ª The Royal Society of Chemistry 2013
Fig. 1 (a) Illustration of the superacid slide casting method. (b) Comparison of the optoelectronic properties of SWNT thin films. (c) Scanning Electronic Microscope(SEM) characterizations of a SWNT film at top surface. The scale bar is 1 mm. (d) Atom Force Microscope (AFM) of a SWNT film at bottom surface with RMS roughness of2.78 nm.
Paper Energy & Environmental Science
Dow
nloa
ded
by Y
ale
Uni
vers
ity L
ibra
ry o
n 15
/04/
2013
03:
52:4
3.
Publ
ishe
d on
20
Dec
embe
r 20
12 o
n ht
tp://
pubs
.rsc
.org
| do
i:10.
1039
/C2E
E23
716D
View Article Online
chlorosulphonic acid with individual SWNTs forming anisotropic concentrated phase. In our superacid casted lms, weobserved a small bundle size of �4 nm (Fig. 1c). The improve-ment is also due to the avoidance of sonication (required for theMayer Rod) in the dispersion process. The superacid methodgently disperses the SWNTs using magnetic bars and hencepreserves the integrity and length of the carbon nanotubes.
Surprisingly, the SWNT lms prepared by the superacidsliding method present better mechanical robustness/unifor-mity and ease of handling. In both the superacid and the Mayerrod coating process, the SWNTs are rst coated on glass slidesand subsequently transferred onto the Si wafer through a water-oating process. The structural integrity of the SWNT lms onwater facilitates the transfer process, which directly inuencesthe SWNT/Si interfacial quality and solar cell performance. TheMayer rod SWNT lms on water tend to wrinkle and disas-semble (especially for thin lms with high transmittance>�85%); hence, disturbing the structural uniformity during thetransfer step. Meanwhile, the superacid lms are extraordi-narily robust and retain their integrity during transfer evenwhen they are extremely thin (�8 to 10 nm thickness corre-sponding to �95% transparency at 550 nm wavelength). Weattribute this robustness to the formation of highly aligned andwoven networks of SWNTs induced by the shear forces involvedin the sliding steps during the superacid process. We show thatthe SWNT bundles that are thicker than �15 nm are highlyaligned along the shearing direction and form a spinal structurewhile thin bundles smaller than �10 nm randomly binds thealigned spines forming strong and woven networks (Fig. 1c).
We note that the surfaces of the superacid lms aresmoother than our previously designed Mayer rod lms whichrepresent the best in class with an surface root-mean-square(RMS) roughness of 11.2 nm.30 Since the bottom side (in contactwith a glass slide) of the superacid casted lm is interfaced with
This journal is ª The Royal Society of Chemistry 2013
the surface of a Si wafer in SWNT/Si devices, the roughness ofthe bottom surface is more critical for maximizing the p–njunction areas. Remarkably, the slowly dried superacid castedlms have a RMS 2.78 nm (Fig. 1d). To the best of our knowl-edge, these slowly dried superacid lms are the smoothestSWNT percolated thin lms (with or without polymer lling thepores of SWNTs) that have been reported. We show in ESI† theimportance of coagulation process in engineering the smooth-ness of superacid casted lms. The RMS of the top surface forslowly dried lms and water quenched lms are 4.12 nm and9.99 nm, respectively. (Fig. S1c and d†)
We compare the performances of the SWNT/Si solar cellswith the SWNT lms prepared by the Mayer rod (Fig. 2a) andsuperacid slide casting method (Fig. 2b). We apply of 1% HF for1 min onto SWNT/Si windows to improve the performances ofall the devices owing to the removal of native oxides andH-passivated Si surfaces (Fig. S2†).36 These devices are called “asmade” devices.
We further treated the as made cells with a nitric acid (HNO3,0.5 mol L�1) solution to further p-dope the SWNTs.37 Enhancedp-doping of SWNT is expected to result in a larger Voc byincreasing the built-in electric eld at the SWNT/Si interfaceand improves the separation efficiency of the photo-generatedcarriers. Every device has three states in this study: state one,pretreatment (as made); state two, with wet nitric acid; statethree: nitric acid treated and dried. Superacid SWNT/Si devicesconsistently behave better at all three states. In Table 1, wesummarize the performance of Mayer rod SWNT/Si and super-acid SWNT/Si devices when nitric acid dries (state three). It isnoteworthy that superacid method yields better performance inall aspects of the devices, including lower ideality factor, higherVoc, Jsc, Fill Factor (FF) and power conversion efficiency (PCE).While this improved performance partially results from thedifference in optoelectronic performance (Fig. 1b), it is also
Fig. 2 (a) and (b): J–V characteristics of SWNT/Si solar cells prepared by (a) Mayer rod and (b) superacid methods under 1 sun at different stages in the nitric acidtreatment study. Contact angle measurements of acid droplets on SWNT films prepared by (a inset) Mayer rod (b inset) superacid methods. (c) Surface profile of asuperacid SWNT film (top) and a Mayer rod SWNT film (bottom). (d) Depth-dependence of the percentage of projected SWNT area in contact with Si wafer. (e) and (f)Visualization of SWNTs in contact with Si and not in contact with Si for (e) a Mayer rod SWNT film and (f) a superacid SWNT film. The scale bar is 1 mm.
Table 1 Device performance of Mayer rod SWNT/Si and superacid SWNT/Sidevices after nitric acid dries
attributed to the improved smoothness of superacid SWNTlms as compared to that of Mayer rod SWNT lms. Surfaceroughness controls the effective interfacial area between SWNTnetwork and Si, which determines the distance minoritycarriers need to travel along the Si surfaces to be dissociatedand get collected. Smaller distance is of vital importance tofacilitate carrier dissociation and to minimize surface recom-bination so that lower ideality factor and higher FF can beachieved. Data in Fig. 2c and d are extracted from AFM images
This journal is ª The Royal Society of Chemistry 2013
of Mayer rod and superacid SWNT lms. Fig. 2c demonstrates amuch smoother surface from superacid than Mayer rod madeSWNT lms. It also denes the concepts of “baseline” and“depth” used in Fig. 2d. Using the tallest point of the originalAFM image as a baseline, Fig. 2d shows the ratio the projectedSWNT network area on Si to the Si geometry area as a functionof depth. This ratio represents the percentage of SWNT incontact with Si if the SWNT network of the corresponding depthwere attened. It is obvious that the bigger the percentagewithin a small depth, the larger the effective SWNT/Si interfacialareas are. The smoothness of superacid SWNT lms renders agreatly enhanced contact between SWNT and Si (Fig. 2d). Fig. 2eand f visualize the effective in contact area and not in contactarea assuming that the highest 16% of the SWNTs plus a 1.5 nm(SWNT diameter) can be in contact with the Si. A depth of 37.88nm and 117.46 nm are used for the superacid and theMayer rodSWNT lms respectively. The carriers generated at Si surfaceneed to travel on average 14.6 nm and 55.5 nm to reach thenearest SWNT/Si interface for the superacid coated and theMayer rod coated SWNT thin lms, respectively. When a depthof 37.88 nm is used for Mayer rod SWNT devices (Fig. S3†), anaverage distance of 2.2 mm is needed. The actual depth dependson the exibility and deformability of SWNTs thin lms. Theseestimates indicate that, statistically, excited carries need totravel for a shorter distance along the Si surface to reach SWNT/Si interfaces in superacid devices than that in Mayer roddevices.
Upon nitric acid treatment, the following three phenomenaare also observed (Fig. 2a and b): (1) for both lms, the devicestreated with HNO3 (both state two and three) show improvedperformances compared to pretreatment devices (state one); (2)the efficiency when nitric acid solution is present (hwet; statetwo) is higher than efficiency when nitric acid lm is added anddried (hdry; state three) with hwet/hdry ¼ 1.32 for superacid lms.Yet for Mayer rod lms it is the reverse (hwet/hdry ¼ 0.55). Theefficiency increase for superacid devices are mainly due toimproved Jsc, while the efficiency decrease for the Mayer rod is a
Fig. 3 (a) Variation of Jsc with multiple cycles of with/without of water droplet on thSWNT/Si solar cell under 0.1 sun of solar simulation before (top curve) and after (botof a hydrophobic glass.
This journal is ª The Royal Society of Chemistry 2013
result of a reduced FF and Voc; (3) the Jwetsc (state two) is consis-tently larger than the Jdrysc (state three) with Jwetsc /Jdrysc of 1.33 and1.08 for superacid and Mayer rod SWNT lms, respectively.
Jia et al. attributed the larger Jsc in the wet state over the drystate to a photoelectrochemical reaction.21 Studies by D.Michalak and N. Lewis32,33 demonstrate that the aqueous acidicelectrolyte can induce an inversion layer, reducing the surfacerecombination velocity. The observed increase in solar cellperformance is more likely due to this effect. Besides, theintermediate index of refraction of acidic electrolyte relative toair and Si leads to reduction in light reection, which alsocontributes to a larger Jsc (Fig. S4†). In addition, the SWNTfabrication dependence with nitric acid treatment implies thatcertain surface properties of SWNTs controlled by the SWNTpreparation method changes the way the SWNTs interface withthe nitric acid solution and determines the solar cellcharacteristics.
We suggest that the wetting nature of SWNT thin lmscorrelates with these observations and present the contactangle measurements of HNO3 droplets on the SWNT lms.Large contact angles of 98� (Fig. 2b inset) are observed fromthe SWNT lms prepared by the superacid methods, revealingthe hydrophobic nature of these samples. Meanwhile, 37� isshown for the Mayer rod prepared SWNT lms (Fig. 2a inset).We hypothesize that the enhanced efficiencies of SWNT/Sidevices in their wet states come from a “light concentrationeffect” owing to the hemispherical droplets that efficientlyincouple the incident light.38 Since the superacid SWNTs arehydrophobic, the sharper curvature of the droplet leads to astronger light concentrating effect and higher Jsc enhance-ment. In contrast, a at droplet on a hydrophilic lm (via theMayer rod method) results in a lower concentration effect andhence a smaller Jsc enhancement. In addition, hydrophiliclms tend to delaminate and oat on nitric acid solution,deteriorating the carrier dissociation at the SWNT/Si interfaceand leading to a lower FF and Voc as shown with Mayer rodmade lms (Fig. 2a).
e surface of a hydrophobic SWNT/Si solar cell. (b) J–V characteristics of a Mayer rodtom curve) incorporating a light concentric effect from water droplet place on top
To verify the proposed light concentration effect and isolateother possible contributions, we applied a droplet of deionizedwater onto the SWNT/Si window of a cell with a hydrophobicSWNT network, and characterized the performance before/aerthe water droplet dries. We show that the Jsc consistentlyincreases with the addition of a water droplet and recovers tothe original low values when it dries, in multiple cycles betweenthe wet and dry states (Fig. 3a). By optimizing the size (coverageof the p–n junction area) of the water droplet, we report that theJsc can be increased up to �1.9 times over the dry state withhydrophobic cells. An image of water droplet in a hemisphericalshape on an operating cell is shown in Fig. S5.† As a controlexperiment, we placed a water droplet onto a hydrophobic-coated glass placed on the top of a hydrophilic Mayer rod cellunder illumination and show a �2.8 times increase in Jsc (at 0.1sun of solar simulation (10 mW cm�2)) compared to without thewater droplet (Fig. 3b). A maximum power generation of 0.87mW cm�2 is obtained demonstrating an efficiency of 8.7%before applying a water droplet. In the presence of the waterdroplet, a maximum power generation as high as 2.7 mW cm�2
(dened using SWNT/Si device area) is obtained which corre-sponds to a �3.1 times increase in power output. This resultsuggests that the light concentrating effect from the nitric aciddue to the hydrophobicity of the SWNTs is the major
Fig. 4 (a)–(d) Solar intensity-dependent variation of device parameters in superaciover pre-Au doping in the upper inset. Lower inset shows TEM micrographs of a SW
884 | Energy Environ. Sci., 2013, 6, 879–887
contributor to the observed enhanced Jsc instead of the photo-electrochemical reaction.
We further investigate the intrinsic p-doping effect of SWNTsusing gold salt; gold salt is known to be reduced on the surfaceof CNTs, thus, providing a p-doping effect.39 We treated thesuperacid cells with various concentrations of AuCl3 salt innitromethane, and compare their performance in the dry states,ruling out the light concentration effect. We observe 1.2–1.3times Jsc increases with the addition of AuCl3 solution (2 mM to10 mM) in all the tested range of solar intensity (0.1 sun to 1sun) (Fig. 4a). This is partially due to a slight (3%) increase intransmittance due to the depletion of the second singularity(lled electronic states),40 which results in the loss of the cor-responding electronic transition and loss of the associated S2absorption intensity (Fig. S6†). Transmission electron micros-copy (TEM) characterization on the side of the SWNT lm(super-acid sliding) in contact with the Si wafer aer 10 mMgold salt treatment (Fig. 4a bottom inset) reveals a large numberof nanoparticles ranging in size from �1 to 2 nm to �100 nmwith single-crystalline structures (lattice fringes in Fig. 4abottom inset) as also conrmed by energy-dispersive X-rayspectroscopy elemental mapping (Fig S7†). The presence of gold(Au) nanoparticles suggests that optical scattering from thesenanoparticles is also likely to participate in enhancing the light
d SWNT/Si solar cells with varying AuCl3 concentration. (a) Jsc, and normalized JscNT film prepared with a AuCl3 of 10 mM (b) Voc. (c) FF. (d) Efficiency.
This journal is ª The Royal Society of Chemistry 2013
absorption.41 We therefore believe the observed enhancementof Jsc is most likely a combination of both effects. We found thatthe Jsc decreases at 10 mM (compared to 5 mM) of AuCl3 andobserved a huge degradation in device performance whendoped with 20 mM AuCl3. We attribute this observation to a“shorting” problem as an excess amount of Au nanoparticlesform an intimate contact with the Si. The local Si–Au–SWNTjunctions could disrupt the p–n junction areas.
We reveal that the Voc increases with increasing AuCl3concentration up to 10 mM and decrease at higher concentra-tions (Fig. 4b). The increase of Voc with Au doping can beexplained by the increase of charge carrier density of SWNTs.The p–n junction solar cells made of thin p-type emitter (SWNTin our case) on n-type Si wafer, Voc is given by
Voc ¼ kT
qln
�DpðNd þ DnÞ
ni2þ 1
�
where k, T, q, ni, Nd represents the Boltzman constant,temperature, unit charge, intrinsic carrier concentration of Si,and n-doping concentration, respectively.42 Since, Dp (Dn)represents the excess minority carriers (holes and electrons,respectively) imposed by doping, the equation predicts a higherVoc with increased p-doping of the SWNT. In our device geom-etry, the SWNTs functions as both a donor and as a transparentconductive electrode. The increase of Voc with increased workfunction of the transparent conductive oxide front contactelectrode was observed in amorphous Si/crystalline-Si hetero-junction solar cells where the thin amorphous Si functions asan emitter.43 The above equation is also expressed as;44
Voc ¼ kT
qln
�Jsc
J0
�
where J0 is the saturation current density. From dark Ln(J)–Vplots, J0 is determined by extrapolating the linear slope ofLn(J)–V to zero bias and J0 is found to decrease with increasingAu concentration; Jundoped0 ¼ 1.31 � 10�5 mA cm�2, J2mM
0 ¼3.68 � 10�6 mA cm�2, J5mM
0 ¼ 2.78 � 10�6 mA cm�2, J10mM0 ¼
2.26 � 10�6 mA cm�2 (Fig. S8†). The decrease of J0 and the
Fig. 5 (a) J–V characteristics showing 11.5% efficiency under 1 sun from an optimizdata from (a) and (b) were obtained from superacid SWNT/Si cells.
This journal is ª The Royal Society of Chemistry 2013
increase of Jsc together also predict the increase of Voc,consistent with our observation.
The ll factor, FF, is also observed to increase withincreasing AuCl3 concentration (Fig. 4c). This is primarily dueto the reduction of SWNT thin lm sheet resistance(Fig. S9†).45–47 This improvement results in an increase of theoverall efficiency (Fig. 4d); 1.7 times increase at 1 sun (from5.3% to 9.13%) and 1.5 times increase at 0.1 sun (from 7.2% to10.8%) with 5 mM Au doping over un-doped cells.
Fig. 5a shows the J–V characteristics of the optimizedsuperacid SWNT/Si solar cells both in dark (inset) and underillumination. We use Al as the back contact (Fig. S10†) andchromium/gold (Cr/Au) as the front contact. The SWNT thinlms are casted by our superacid sliding method and slowlydried in anhydrous air. The lms are transferred on pre-patterned Si wafers that are freshly treated by HF. The cells arethen treated subsequently with a 0.5 mol L�1 nitric acid and a 5mM gold salt solution by spin casting. The inset shows the darkLn(J)–V characteristics which yields an ideality factor of 1.08. At1 sun, Voc of 0.533 V, Jsc of 29.31 mA cm�2 and FF of 73.8% areobtained, resulting in a power conversion efficiency of 11.5%.Another important feature of our optimized solar cell is its longminority carrier lifetime characterized by the reverse recoverytransient method.48 This characterization method relies on thetransient dynamics of carriers electrically injected into thedepletion layer of p–n junctions.49 A p–n junction diode issubject to an abrupt switching of the forward-to-reverse bias,resulting in a transient response from which the carrier lifetimeis extracted. We show a reverse recovery transient characteristic(Fig. 5b) from the SWNT/Si solar cell in Fig. 5a. The corre-sponding current response shows three distinct phases ofconstant forward current (If), constant reverse current (Ir) and aconstant storage time (ts) followed by slow decay/saturation ofthe reverse current. The minority carrier lifetime is given by s ¼ts/ln(1 + If/Ir), and is determined to be �35 ms in dark, which iscomparable (5.9 to 19.3 ms) to the lifetimes of high-quality, polycrystalline p–n junction Si solar cells grown/processed at hightemperatures.50 This long carrier lifetime combined with the
ed cell. (b) Reveres recovery transient characteristic from an optimized cell. All the
small ideality factor close to unity suggests that the observeddark J–V characteristics were not severely affected by carrierrecombination at the interfaces of SWNT/Si. These ndingssuggests that the photovoltaic efficiency may be furtherimproved by enhancing the absorption/trapping efficiency ofthe incoming light by patterning Si and increasing the interfa-cial area of SWNT/Si, which is currently under investigation.
Conclusions
We present improved SWNT/Si solar cells by applying advancedSWNT lm preparation methods and post-fabrication treat-ments. We show the superior performance of solar cells preparedby the superacid sliding method, which benets from the supe-rior optoelectronic properties, morphological uniformity androbustness of the SWNT lms. By optimizing the SWNTmaterialproperties as well as the SWNT/Si and Si/metal interfaces, wegreatly improved the Jsc, Voc, and FF, and achieved a maximumefficiency of 11.5% at 1 sun. We also identied that the CNTwettability signicantly affects solar cell characteristics in theirwet state and unveiled that the increased short circuit currentand improved power conversion efficiency in the presence ofnitric acid/water droplets are from a light concentrating effect.We envision this study to be useful for developing transformativehybrid solar cells based on CNTs where novel functionalizationswould be possible via simple, low-cost processes.
Acknowledgements
This work was supported by the SOLAR program of the NationalScience Foundation under DMR-0934520. Support is alsoacknowledged from the NSF-CAREER award (CBET-0954985) andNASA (CT Space Grant Consortium). Support is also acknowl-edged by ‘Nanostructures for Electrical Energy Storage’, anEnergy Frontier Research Center funded by the USDepartment ofEnergy, Office of Science, Office of Basic Energy Sciences, underAward number DESC0001160 (Y. J. and M. A. R.). The authorsacknowledge Dr Nilay Hazari for helpful discussions, for allowingus to use his facilities as well as his help with superacid prepa-ration and processing. Southwest Nanotechnologies areacknowledged for their kind supply of single walled carbonnanotubes. David Kohn is acknowledged for helpful discussionof results and experiment design. Facilities use was supported byYINQE and NSF MRSEC DMR 1119826.
Notes and references
1 T. Durkop, S. A. Getty, E. Cobas and M. S. Fuhrer, Nano Lett.,2004, 4, 35–39.
2 A. Javey, H. Kim, M. Brink, Q. Wang, A. Ural, J. Guo,P. McIntyre, P. McEuen, M. Lundstrom and H. J. Dai, Nat.Mater., 2002, 1, 241–246.
3 A. Javey, J. Guo, Q. Wang, M. Lundstrom and H. J. Dai,Nature, 2003, 424, 654–657.
4 Z. Yao, C. L. Kane and C. Dekker, Phys. Rev. Lett., 2000, 84,2941–2944.
886 | Energy Environ. Sci., 2013, 6, 879–887
5 C. Z. Loebick, R. Podila, J. Reppert, J. Chudow, F. Ren,G. L. Haller, A. M. Rao and L. D. Pfefferle, J. Am. Chem.Soc., 2010, 132, 11125–11131.
6 H. P. Liu, D. Nishide, T. Tanaka and H. Kataura, Nat.Commun., 2011, 2, 309–317.
7 F. Leonard, in Micro & Nano Technologies; Variation: Micro &Nano Technologies, William Andrew, 2009.
8 F. S. Gittleson, D. J. Kohn, X. Li and A. D. Taylor, ACS Nano,2012, 6, 3703–3711.
9 L. H. Zhang, Y. Jia, S. S. Wang, Z. Li, C. Y. Ji, J. Q. Wei,H. W. Zhu, K. L. Wang, D. H. Wu, E. Z. Shi, Y. Fang andA. Y. Cao, Nano Lett., 2010, 10, 3583–3589.
10 D. D. Tune, B. S. Flavel, R. Krupke and J. G. Shapter, Adv.Energy Mater., 2012, 2, 1043–1055.
11 M. H. Ham, G. L. C. Paulus, C. Y. Lee, C. Song, K. Kalantar-zadeh, W. Choi, J. H. Han and M. S. Strano, ACS Nano, 2010,4, 6251–6259.
12 G. Li, R. Zhu and Y. Yang, Nat. Photonics, 2012, 6, 153–161.13 A. Tada, Y. F. Geng, Q. S. Wei, K. Hashimoto and K. Tajima,
Nat. Mater., 2011, 10, 450–455.14 R. M. Jain, R. Howden, K. Tvrdy, S. Shimizu, A. J. Hilmer,
T. P. McNicholas, K. K. Gleason and M. S. Strano, Adv.Mater., 2012, 24, 4436–4439.
15 V. C. Tung, J. H. Huang, J. Kim, A. J. Smith, C. W. Chu andJ. X. Huang, Energy Environ. Sci., 2012, 5, 7810–7818.
16 D. J. Bindl, M. Y. Wu, F. C. Prehn and M. S. Arnold, NanoLett., 2011, 11, 455–460.
17 Z. Li, V. P. Kunets, V. Saini, Y. Xu, E. Dervishi, G. J. Salamo,A. R. Biris and A. S. Biris, ACS Nano, 2009, 3, 1407–1414.
18 Z. R. Li, V. Saini, E. Dervishi, V. P. Kunets, J. H. Zhang, Y. Xu,A. R. Biris, G. J. Salamo and A. S. Biris, Appl. Phys. Lett., 2010,96, 033110.
19 Y. Jia, J. Q. Wei, K. L. Wang, A. Y. Cao, Q. K. Shu, X. C. Gui,Y. Q. Zhu, D. M. Zhuang, G. Zhang, B. B. Ma, L. D. Wang,W. J. Liu, Z. C. Wang, J. B. Luo and D. Wu, Adv. Mater.,2008, 20, 4594–4598.
20 P. G. Collins, K. Bradley, M. Ishigami and A. Zettl, Science,2000, 287, 1801–1804.
21 Y. Jia, A. Y. Cao, X. Bai, Z. Li, L. H. Zhang, N. Guo, J. Q. Wei,K. L. Wang, H. W. Zhu, D. H. Wu and P. M. Ajayan, NanoLett., 2011, 11, 1901–1905.
22 Q. K. Shu, J. Q. Wei, K. L. Wang, H. W. Zhu, Z. Li, Y. Jia,X. C. Gui, N. Guo, X. M. Li, C. R. Ma and D. H. Wu, NanoLett., 2009, 9, 4338–4342.
23 Y. Jia, P. X. Li, J. Q. Wei, A. Y. Cao, K. L. Wang, C. L. Li,D. M. Zhuang, H. W. Zhu and D. H. Wu, Mater. Res. Bull.,2010, 45, 1401–1405.
24 X. C. Miao, S. Tongay, M. K. Petterson, K. Berke,A. G. Rinzler, B. R. Appleton and A. F. Hebard, Nano Lett.,2012, 12, 2745–2750.
25 G. Gruner, J. Mater. Chem., 2006, 16, 3533–3539.26 D. Kozawa, K. Hiraoka, Y. Miyauchi, S. Mouri and
K. Matsuda, Appl. Phys. Express, 2012, 5, 042304.27 M. S. Strano, C. A. Dyke, M. L. Usrey, P. W. Barone,
M. J. Allen, H. W. Shan, C. Kittrell, R. H. Hauge, J. M. Tourand R. E. Smalley, Science, 2003, 301, 1519–1522.
This journal is ª The Royal Society of Chemistry 2013
28 Y. H. Lee, H. Z. Geng, D. S. Lee, K. K. Kim, G. H. Han andH. K. Park, Chem. Phys. Lett., 2008, 455, 275–278.
29 P. Vichchulada, M. A. Cauble, E. A. Abdi, E. I. Obi,Q. H. Zhang and M. D. Lay, J. Phys. Chem. C, 2010, 114,12490–12495.
30 X. Li, F. Gittleson, M. Carmo, R. C. Sekol and A. D. Taylor,ACS Nano, 2012, 6, 1347–1356.
31 P. Wadhwa, G. Seol, M. K. Petterson, J. Guo and A. G. Rinzler,Nano Lett., 2011, 11, 2419–2423.
32 D. J. Michalak and N. S. Lewis, Appl. Phys. Lett., 2002, 80,4458–4460.
33 D. J. Michalak, F. Gstrein and N. S. Lewis, J. Phys. Chem. C,2008, 112, 5911–5921.
34 V. A. Davis, A. N. G. Parra-Vasquez, M. J. Green, P. K. Rai,N. Behabtu, V. Prieto, R. D. Booker, J. Schmidt,E. Kesselman, W. Zhou, H. Fan, W. W. Adams, R. H. Hauge,J. E. Fischer, Y. Cohen, Y. Talmon, R. E. Smalley andM. Pasquali, Nat. Nanotechnol., 2009, 4, 830–834.
35 L. Hu, D. S. Hecht and G. Gruner, Nano Lett., 2004, 4, 2513–2517.
36 E. Yablonovitch, D. L. Allara, C. C. Chang, T. Gmitter andT. B. Bright, Phys. Rev. Lett., 1986, 57, 249–252.
37 W. Zhou, J. Vavro, N. M. Nemes, J. E. Fischer, F. Borondics,K. Kamaras and D. B. Tanner, Phys. Rev. B: Condens. MatterMater. Phys., 2005, 71, 205423.
38 J. H. Karp, E. J. Tremblay and J. E. Ford, Opt. Express, 2010,18, 1122–1133.
This journal is ª The Royal Society of Chemistry 2013
39 H. C. Choi, M. Shim, S. Bangsaruntip and H. J. Dai, J. Am.Chem. Soc., 2002, 124, 9058–9059.
40 Z. C. Wu, Z. H. Chen, X. Du, J. M. Logan, J. Sippel,M. Nikolou, K. Kamaras, J. R. Reynolds, D. B. Tanner,A. F. Hebard and A. G. Rinzler, Science, 2004, 305, 1273–1276.
41 F. J. Beck, A. Polman and K. R. Catchpole, J. Appl. Phys., 2009,105, 114310.
42 R. A. Sinton and A. Cuevas, Appl. Phys. Lett., 1996, 69, 2510–2512.
43 E. Centurioni and D. Iencinella, IEEE Electron Device Lett.,2003, 24, 177–179.
44 S. M. Sze and K. K. Ng, Physics of Semiconductor Devices,Wiley-Interscience, 2007.
45 Y. Zhang, X. D. Dang, C. Kim and T. Q. Nguyen, Adv. EnergyMater., 2011, 1, 610–617.
46 R. Mauer, I. A. Howard and F. Laquai, J. Phys. Chem. Lett.,2010, 1, 3500–3505.
47 S. Jeong, E. C. Garnett, S. Wang, Z. Yu, S. Fan,M. L. Brongersma, M. D. McGehee and Y. Cui, Nano Lett.,2012, 12, 2971–2976.
48 Y. W. Jung, A. Vacic, D. E. Perea, S. T. Picraux andM. A. Reed,Adv. Mater., 2011, 23, 4306–4311.
49 H. J. Kuno, IEEE Trans. Electron Devices, 1964, 11, 8–14.
50 M. S. Mason, C. E. Richardson, H. A. Atwater andR. K. Ahrenkiel, Thin Solid Films, 2006, 501, 288–290.
