Modeling, Synthesis, and Characterization of Thin Film Copper Oxide for Solar Cells

Transcription

1 Modeling, Synthesis, and Charaterization of Thin Film Copper Oxide for Solar Cells Davis S. Darvish Harry A. Atwater California Institute of Tehnology, Pasadena, CA, USA ABSTRACT The modeling, growth, and haraterization of Copper Oxide thin films for solar ell appliations are reported. CU20 has several attrative properties whih inlude its diret band gap (Eg=2.17 ev) for use in photoeletrolysis of water and use in tandem multi-juntion ells. Detailed balane alulations predit effiienies on the order of 2000 while CU20 ells have yet to even pass 200 effiieny. The devie physis model reveals that defets, partiularly at the heterojuntion interfae, are the main reason for lowered effiienies. Epitaxial CU20 (100) thin films on MgO are fabriated using RF Oxygen plasma MBE. The films are quite smooth and showed mobilites in the range of m 2N*se and arrier onentrations in the range of 10 14_1017. Finally, the epitaxial growth of CU20 on a MgO template is demonstrated. harge arrier separation. For the purposes of this paper we use n-zno as the heterojuntion partner in the devies we model and desribe. There are many reports on CU20 solar ells prepared by various tehniques inluding eletrodeposition, thermal oxidation of sheet metal, and sputtering deposition. [3,5,6] However, these ells have only reahed energy effiienies that are a fration of the Shokley-Queisser theoretial value. Despite the effort of many researhers, p-n heterojuntions have yet to demonstrate good performane. Additionally the ontrol of thin film growth and properties has not been well investigated. The lak of high quality material has resulted only in a reord effiieny of 2%. [7] We investigate the growth of MBE CU20 in order to better understand and ontrol material properties of our thin film, in the hopes of ultimately inreasing the effiienies of films fabriated in the future. INTRODUCTION Copper Oxide (CU20) was the first semiondutor material disovered, but was soon overtaken by the fast development of silion. Nearly 90 years after its disovery, interest in this material has renewed for use in thin film photovoltais, as there has been muh sientifi progress in the development and growth of thin films. Previous work onduted on thin film photovoltais and heterojuntions suh as CIGS and CdTe has guided our investigations of CU20. Copper Oxide is a non toxi semiondutor that has a diret band gap of 2.17eV, whih is ideal for use in multijuntion ells or for photo-eletrolysis of water.[1] It also has long minority arrier diffusion length (- 101Jm) [2]. Most importantly it is omposed of both earth abundant and inexpensive materials whih makes the terawatt salability of quite feasible espeially if photovoltais will playa large role in the transformation of energy from fossil fuels to solar ells.[3] Copper Oxide is intrinsially a p-type semiondutor predominately due to opper vaanies, and nearly all efforts to form homojuntions by n-doping of CU20 have failed. An exeption is a reent report [4] in whih very preliminary work was reported and no photovoltai properties were observed. For that reason photovoltai devies employing CU20 either use Shottky barriers or semiondutor heterojuntions as a mean for Detailed Balane MODELING To realize the potential of CU20 as both a single juntion and multijuntion solar ell material, it is important to explore the detailed balane thermodynami effiieny model of single, double and triple juntion solar ells. The standard AM 1.5 solar spetrum is used to determine the thermodynami effiieny of CU20 at 300K under 1 sun onentration. The effiieny of a solar ell is alulated by diving the extrated power from the ell by the integrated power of the AM 1.5 solar spetrum on the ell. ICV). V 11=- p J(V) is the urrent density generated by the ell as a funtion of operating voltage V. This model also makes several basi assumptions. These assumptions are that all photons greater than Eg are absorbed by the ell and reate eletron-hole pairs, all reombination ours radiatively and they are non-thermal, and all absorbed photons equal the number of photons reemitted through radiative reombination plus eletron-hole pairs extrated from the ell. The model also takes into aount the ritial (1) $ IEEE

2 angle for emission to a medium of different refrative index. Using the geometry of a thin single-heterojuntion on refletive bak surfae ontat the effiieny is determined to be 18.74%. It is important to note that the distribution of power in the solar spetrum is broad, and it annot be effiiently harnessed using a single band gap ell. Beause the dominant soures of loss are photons with energies either greater than or less than the band gap, multijuntion ells are used to more effiiently absorb the broad solar spetrum. Using CU20 as the top ell in both two and three juntion ells, it is determined that the optimal lower ell band gap in a 2-juntion ell was 1.58 ev resulting in an overall effiieny of 34.21%. In a 3juntion ell the optimal bad gaps for the lower ells are determined to be 1.69 ev and 1.35 ev with an overall effiieny of 45.76%. It is important to note that all the modeled ells are urrent mathed and running them in parallel would offer larger effiien ies. The numbers alulated above are for ideal systems with ideal band gaps, but it is important to look at urrent material systems being produed to see if any of these ells will gain from being paired with CU20. Using the same assumptions in the single juntion model above, effiienies of CU20Si and CU20GaAs dual juntion solar ells are determined to be 27.11% and 30.08% respetively. effiieny drops as the onentration of these defets are inreased. These simulations assume Lambert-Beer's law of absorption of light and are modeled as losely as possible to materials that have urrently been fabriated to date. Optial and eletroni of films fabriated in lab are utilized in the simulation. The fabriation of these films will be desribed in the following setion. Figure 2a shows the band diagram of a ell under AM 1.5 illumination. The simulated ell has a ZnO layer that is 200nm thik and a CU20 that is 1IJmthik. The band offsets are determined by the eletron affinity of the two heterojuntion materials. The CU20 layer has an intrinsi arrier onentration of 5x10 and the ZnO layer is almost degenerately doped Zn-ZnO. Figure 2 b, shows ell performane under AM 1.5 illumination. The slightly low short iruit urrents an be attributed to defet and interfae reombination. The high Vo's are enouraging as many heterojuntions fabriated previously have Vo's a fration (20%) of these values. The fill fators are strongly dependent on the series resistane of the ells and an greatly vary depending on the doping and mobility of thin film layers as well as ontat resistane whih is not taken into aount in these simulations. The external quantum effiieny alulated for the arrier onentration of 5x10 is as expeted. The higher values both in Vo and QE an be attributed to modeling a higher quality interfae that is possible to be fabriated using MBE. Devie Physis Model ( p-type The devie physis model of a CU20lZnO (Fig 1) heterojuntion ell will also allow one to gain a better understating of the band struture, and to model ell performane under AM 1.5 illumination. Afors-Het (v 2.2) [8], a heterojuntion devie physis program developed for a-si, is used in modeling the solar ell and alulating numerial results. ) n-type depletion regio n u,o=2.17 rtl. -- l\ =.2 -,.--- Quasi- ermi Levels=1.4 L a) Fig. 1) Shemat i of CU20lZnO solar ell. As expeted, the ell is most effiient with no traps or interfae defets between the n-znop-cu20, and the $ IEEE

3 0,010 0,008 N < E 0,006 r ===::::::::::::: 'iii (I) o 0,004 :!!:! :; o 0,002 b) - - Carrier [SE18] - - Carrier [SE17] - - Carrier [SE] - - CarrierfSE1S1 0,000 L- L - _L - _IU...JLL- ' 0,0 0,5 1.5 Voltage [VI External Quantum Effiieny EXPERIMENT IV for Different Carrier Cone. r , From the insight that was gained from modeling, Moleular Beam Epitaxy (MBE) seemed to be the best method to fabriate the solar ells, as it provides the greatest ontrol over ritial growth onditions suh as temperature, flux, base pressure, and interfae sharpness. We used ubi Magnesium Oxide (MgO (100), a=4.22a) as our substrate with a low lattie mismath of 1.1% between CU20 (a=4.27a) and the substrate. We used a opper effusion ell operating through a temperature oc0c range T=1 050 T=1 080 and oxygen partial pressure By varying the oxygen partial pressure and opper effusion rate, we were able to hange properties of the film inluding doping. The optimal onditions were 0C determined to be T=1 060 for the Knudsen Copper effusion ell, with a substrate temperature of T=650oC. The thin films were grown in the presene of a RF oxygen plasma (P=300W) at 10-6torr. Several different post deposition annealing steps were explored to reate the highest quality film. In-situ haraterization of our film was done with Refletive High Energy Eletron Diffration (RHEED). Further analysis was done via x-ray diffration and EDS to onfirm the material grown and rystallinity, as well as Hall measurements to obtain the eletrial properties of our film. 0,5 W RESULTS 0,0 ) Wavelength [nm] Fig. 2.a) Band diagram of ZnOCu20 heterojuntion under AM1.5 iiiumination.b) IV Curves of ells with different arrier onentration for CU20 layer under AM 1.5 illumination. ) EQE of ell with arrier onentration of 5x10 under AM 1.5 illumination. Clean MgO substrate CU20 is one of two CuO stoihiometries. Beause of this it was very important to ontrol the growth of the film ; espeially the flux of Cu and O. An ative Oxygen plasma allowed the film to grow at a muh lower pressure by making the more reative atomi oxygen available instead of moleular oxygen. As mentioned previously, MgO was used as the growth substrate. Both the substrate and CU20 have a ubi rystal struture and losely mathed lattie parameters. We observed that ube on ube epitaxial CU20 was grown on the MgO substrate. In-situ RHEED was used to onfirm the epitaxial growth, whih an be seen in Figure 3. RHEED 30 nm growth 65 nm growth Fig. 3) In-situ RHEED images of epitaxial CU20 on MgO bulk substrate with diffration spots indexed $ IEEE

4 osillations were observed, indiating that the thin film was growing in a layer-by-iayer growth regime, typ ially seen if growth of the film an be well ontrolled and grown slowly (approximately.2 Alse). In addition, the streaky nature of the RHEED image indiated that the film is very smooth. X-ray diffration was onduted on the thin film samples post growth to onfirm epitaxy as well as X-ray diffration of CU20 on MgO to determine the stoihiometery of the film. n-k of Copper Oxide n --k >< x-ray diffration ,-- Epi Cu20 on MgO - - Clean MgO Sub - - Poly Cu20 on MgO r=--:-:::-:=-=----=-=-::----, Fig. 5) Measured n-k optial data for CU20 using spetrosopi ellipsometery. iii Wavelength L ' ' J 80 2 Theta (degrees) Fig. 4) x-ray diffration of CU20 on MgO Sub. Energy Dispersive X-ray Spetrosopy further onfirmed the omposition of the film and did not indiate impurities in the film. Hall mobility measurements showed mobilites 2N*se in the range of m and arr ier 14_1017 onentrations in the range of 10, whih is dependent on substrate temperature and oxygen plasma partial pressure. A very smooth film and the ability of insitu passivation of our interfae will hopefully provide the quality interfae needed to ah ieve muh higher ell effiienies. Optial ellipsometery was onduted on the films to determine index of refration and absorption (see Fig 5). The data measured was subsequently used on other samples to verify film thiknesses and quality post growth. Very thin template layers on the order of nm ofmgo (100) were also grown using Ion Beam Assisted Deposition (IBAD) on heap and amorphous substrates. IBAD e-beam MgO was deposited on top of Silion Nitride and other heap substrates, thus eliminating the need of using ostly MgO substrates. In addition, beause the template is thin, the resulting film that is grown on top will be less strained and onsequently will be of higher quality. As disussed earlier in the modeling setion of this paper, the advantage of CU20 solar ells may be used in multijuntion tandem ells. IBAD MgO an signifiantly lower the ost of the overall ell, and makes it onvenient to integrate with existing ells as several ommerial ells on the market today use SbN emitter layers thus mak ing growth of our CU20 ell on top of existing ells fairly easy. Figure 6 shows RHEED images of IBAD MgO grown on an amorphous Si02 layer. Subsequent epitaxial deposition of CU20 was observed. onm MgO - SiO, Sub 15nm i\lgo - SiO, Sub 15nm Cu,O; 15 nm i\lgo - SiO, Sub Fig. 6) CU20llBAD MgO grown on Si $ IEEE

5 CONCLUSION A thermodynami detailed balane model was used to obtain ultimate effiienies of both single and multijuntion solar ells made with a CU20 IZnO heterojuntion. The most effiient band gaps were also determined for multijuntion ells under AM 1.5 illumination. The devie physis model of the CU20 ell explored the effets of material quality, surfae, and interfae quality as well as identifying the target eletrial properties the films should have. Current solar ell tehnologies on the market were also onsidered for use in CU20 multijuntion tandem ells. These tandem ell ombination detailed balane thermodynami effiienies were also alulated. MBE growth of epitaxial CU20 was demonstrated on (100) MgO. Strutural and eletrial qualities of the film were haraterized using RHEED, x ray diffration, EDS, and Hall mobility measurements. Further haraterization of material quality via PL lifetime and TEM are underway and will be reported in the future in order to help haraterize quality of our material and juntions. [5] J. Katayama et. ai., "Performane of CU20 IZnO solar ell prepared by two-step eletrodeposition", Journal of Applied Eletrohemistry 34, 2004, pp 687 [6] Nobuko Naka, et. ai., "Thin Films of Single-Crystal Cuprous Oxide Grown from the Melt", Japanese Journal of Applied Physis, 44, No. 7A, 2005, pp [7] Alberto Mittiga, et. ai., "Heterojuntion solar ell with 200 effiieny based on a CU20 substrate", Appl. Phys. Lett, 88, 2006,pp 3502 [8] R. Stangl, M. Kriegel, M. Shmidt, "AFORS-HET, Version 2.2" ACKNOWLEDGEMENTS We aknowledge finanial support from U.S. Department of Energy under grant DE-FG36-08G018006, and the Calteh Center for Sustainable Energy Researh. We thank Dr. Matthew Diken, Greg Kimball, and Carrie Hofmann for engaging disussions and assistane. REFRENCES [1] Minami, Tadatsugu et ai., "Effet of ZnO film deposition methods on the photovoltai properties of ZnO CU20 heterojuntion devies", Thin Solid Films. 494, 2006, pp47 [2] Olsen, L. C., Quarterly Progress Report, 1 Nov Jan Joint Center for Graduate Study, Rihland, WA [3] A. Parreta, "Polyrystalline n-znop- CU20 heterojuntions grown by RF-magnetron sputtering", Phys. Stat. Sol. (a) 155,1996, pp 399 [4] L. Wang, M. Tao, "Eletrohemially deposited p-n homojuntion uprous oxide solar ell", Eletrohem. Solid-State Lett. 10,2007, pp $ IEEE