SUPPLEMENTARY INFORMATION doi: 1.138/nnno.29.451 Aove-ndgp voltges from ferroelectric photovoltic devices S. Y. Yng, 1 J. Seidel 2,3, S. J. Byrnes, 2,3 P. Shfer, 1 C.-H. Yng, 3 M. D. Rossell, 4 P. Yu, 3 Y.-H. Chu, 5 J. F. Scott, 6 J. W. Ager III, 2 L. W. Mrtin, 2 nd R. Rmesh 1,2,3 1 Deprtment of Mterils Science nd Engineering, University of Cliforni, Berkeley, Berkeley, CA 9472, USA 2 Mterils Sciences Division, Lwrence Berkeley Ntionl Lortory, Berkeley, CA 9472, USA 3 4 Deprtment of Physics, University of Cliforni, Berkeley, Berkeley, CA 9472, USA Ntionl Center for Electron Microscopy, Lwrence Berkeley Ntionl Lortory, Berkeley, CA 9472, USA 5 Deprtment of Mterils Science nd Engineering, Ntionl Chio Tung University, HsinChu, Tiwn 31 6 Deprtment of Erth Sciences, University of Cmridge, Cmridge CB2 3EQ, UK Supplementry Mterils X-ry Diffrction Studies. We completed detiled x-ry diffrction studies of the two different BFO films (i.e., 71 nd 19 domin wll smples). These studies confirm the findings of the piezoresponse force microscopy imges nd show tht the films re of very high qulity nd crystllinity. Here we focus on the detils of 71 domin wll smple. Supp. Figure 1 nd show reciprocl spce mps (RSMs) of the 23 nd 23 pseudocuic diffrction pek of BFO, respectively. From these two RSMs we hve determined the structure of the BFO film to e tht of monocliniclly distorted perovskite shown in Supp. Figure 3c. Supp. Figure 1d nd e show RSMs of the 2 diffrction pek done in oth HL nd KL spce, respectively. The RSM in Supp. Figure 1d consists of series of omeg nture nnotechnology www.nture.com/nturennotechnology 1 21 Mcmilln Pulishers Limited. All rights reserved.
supplementry informtion doi: 1.138/nnno.29.451 (rocking curve) scns done long the direction perpendiculr to the length of the domins s shown in Supp. Figure 1f. From this RSM we oserve not only strong primry 2 diffrction pek (lled with n L ), ut lso the presence of two superlttice peks (leled with F ). There re two interesting points from this RSM. First, Kessig fringes re oserved long the L (out-of-plne) direction nd show tht the film is extremely flt nd of high crystlline qulity. Second, there re dditionl superlttice peks in the HL plne. The spcing of these peks gives n estimte of the domin periodicity to e ~19 nm, consistent with PFM mesurements, nd tht the ngle the 71 domin wlls mke with the sustrte surfce is ~3. This is consistent with theoreticl nd experimentl studies of BFO films on DyScO 3 sustrtes [1]. Rottion of the smple nd completion of RSM long the direction prllel to the long xis of the domins (Supp. Figure 1f) results only in superlttice peks long the L direction nd no other peks in the KL plne. Trnsmission electron microscopy studies. Cross-sectionl high-resolution trnsmission electron microscopy imges were completed on oth 71 nd 19 domin wll smples. Supp. Figure 2 shows n tomic resolution imge of 71 domin wll. The domin wll is found to e inclined to the surfce of the sustrte nd lies in 11-type plne. On the other hnd, Supp. Fig. 2 shows n tomic resolution imge of 19 domin wll. Such domin wlls re found to e norml to the sustrte surfce nd lie in 1-type plnes. These findings re consistent with theoreticl predictions for domin wlls in rhomohedrl ferroelectrics [2]. Light intensity studies. We hve studied the evolution of the photo-response of the BFO films s function of incident light intensity. Supp. Figure 3 shows the evolution of VOC for 71 domin wll smple. Aove light powers of ~15 mw (corresponding to power density of ~285 mw/cm 2 ) VOC ws found to sturte to vlue of ~1 mv / domin wll. We reiterte tht ll mcroscopic light I-V mesurements were completed in the 2 nture nnotechnology www.nture.com/nturennotechnology 21 Mcmilln Pulishers Limited. All rights reserved.
doi: 1.138/nnno.29.451 supplementry informtion sturtion regime. Additionlly, s is expected, the JSC is found to increse with incresing light intensity (Supp. Figure 3). Clcultions of J SC. The expected mgnitude of JSC cn e predicted on simple sis. Assume the domin width is d, the electrode length is, the seprtion etween the two electrodes is s, nd the smple thickness is t. Let N e the numer of ove-ndgp photons incident per unit re, which cn e computed y integrting over the xenon lmp spectrum, tking into ccount the mesured totl power nd light spot size. Let δ e the distnce from the center of domin wll where photon cn e sored nd still contriute to the current, nd P e the proility tht such n sored photon does in fct contriute to the current. (Even in the strong domin-wll field, n electron-hole pir my nevertheless recomine with finite proility.) Given the 2nm domin-wll thickness, the hopping trnsport of the mteril, nd the strongly-ound, loclized nture of the excitons, δ is expected to e on the order of few nm. The numer of photons per second contriuting to the current is N s(2δ/d)p. For n electron to pss through the externl circuit, it must e excited y s/d photons, one per domin. Therefore, the expected mesured photocurrent, nd in-plne photocurrent density, re respectively: I SC = e (N s (2δ / d) P) / ( s / d ) = 2eN Pδ, J SC = I SC / ( t) = 2eNPδ/t where e is the elementry chrge. As n exmple, consider the device depicted in Fig. 2, with thickness 1nm nd 285 mw/cm 2 totl illumintion intensity. Assuming the ndgp E g = 2.67eV, δ = 2nm, nd P =.25, the clculted current density is 1 1-4 A/cm 2, in greement with the mesured vlue (Fig. 2). This shows tht the mesured currents re consistent with the proposed model, given resonle ssumptions out device prmeters. nture nnotechnology www.nture.com/nturennotechnology 3 21 Mcmilln Pulishers Limited. All rights reserved.
supplementry informtion doi: 1.138/nnno.29.451 Locl mesurements of V OC. Mcroscopic mesurements for rrys of 19 domin wlls were constrined y the presence of two in-plne vrints of 19 domins (drk nd light stripe regions in Fig. 1d). These two vrints hve opposite norml (out-of-plne) components of the polriztion vector nd opposite in-plne net polriztion direct s well. Thus, on mcroscopic scle the contriutions to V OC of these res cncel. However, on microscopic scle the contriution of single vrint cn e mesured using nnoscle conductive AFM (c-afm) under light illumintion. For given distnce (d) locl I-V mesurements were performed y pplying is etween the Pt contct nd the conductive tip (nitrogen-doped dimond coted Si) nd the V OC ws extrcted. For 19 domin wlls n oscilltory ehvior with period of out 5 m ws oserved. This cn e understood y looking t the specific locl domin pttern of the smple. An symmetric distriution of the two 19 domin vrints leds to opposite potentil drops t the domin wlls in these two res, prtilly cnceling the contriutions to V OC depending on the distnce from the electrode (d). From the initil slope vlue of ~12 mv/ m for V oc /d cn e extrcted for 19 domin wlls (Supp. Fig. 4). Our prior theoreticl work showed tht the mgnitude of the potentil step is higher in the cse of 19 domin wlls (15 mv compred to 2 mv for 71 domin wlls). From microscopic mesurements (detils of mesurements re given in the Methods section) potentil drop of ~6.5 mv per 19 domin wll is oserved, s compred to potentil drop of ~1.8 mv per 71 domin wll (Supp. Figure 4). These vlues re considerly less thn the vlue mesured from mcroscopic mesurements; however, this discrepncy is explined y the fct tht to complete these mesurements in the scnning proe setup, the ngle of incidence of the incoming light ws limited to much shllower ngles (i.e., lower intensity not t sturtion s shown in Supp. Fig. 3) nd tht the contct etween scnning proe tip nd the film is electriclly different from the contct etween the Pt nd BFO. Regrdless, 4 nture nnotechnology www.nture.com/nturennotechnology 21 Mcmilln Pulishers Limited. All rights reserved.
doi: 1.138/nnno.29.451 supplementry informtion the locl mesurements revel tht the mgnitude of the potentil step t the 19 domin wll is lrger thn tht of the step t 71 domin wll which is consistent with prior theoreticl predictions [3]. We lso included dditionl informtion concerning I-V dt for different electrode spcings, film thicknesses nd contct mterils, s well s time dependence of the oserved effect, nd ferroelectric switching thresholds in our electrode test structures (see suppl. figures 5 to 8). [1] Y.-H. Chu, et l., Adv. Mter., 18, 237 (26). [2] S. K. Streiffer, et l., J. Appl. Phys. 83, 2742 (1998). [3] J. Seidel, et l., Nture Mter. 8, 229 (29) nture nnotechnology www.nture.com/nturennotechnology 5 21 Mcmilln Pulishers Limited. All rights reserved.
supplementry informtion doi: 1.138/nnno.29.451 Supp. Figure 1 X-ry Diffrction 23c 4.78 4.75 4.75 4.72 4.72 4.69 []DSO 4.66 3.14 3.17 q x 3.155 3.2 (Å-1) d 3.23 2c F q (Å-1) 3.152 L 3.149 F 3.146 3.143 -.3 4.66 c 3.14 3.17 3.2 3.23 q x (Å-1) δ 2 x 19 nm 2c 3.152 ~3o 3.149 3.146. f e β = 3.947(3) Å; = 3.951(8) Å; c = 3.989(2) Å β =.73(4)o; δ = [1]DSO q y (Å-1) 3.155 c 23c [11]DSO 4.69 [11]DSO [11]DSO q (Å-1) 4.78 3.143.3 -.3 q y. (Å-1) t = 25 nm [1]DSO.3 []DSO Supp. Figure 2 - TEM 71 19 [1] [1] [1] [1] [1] 6 [1] nture nnotechnology www.nture.com/nturennotechnology 21 Mcmilln Pulishers Limited. All rights reserved.
doi: 1.138/nnno.29.451 Supp. Figure 3 supplementry informtion 1 8 6x1 V OC (V) 6 4 2 Light intensity for mcroscopic mesurements Cur rrent (A) 4x1 2x1 5 1 15 2 25 3 Light Intensity (mw) 5 1 15 2 25 3 Light Intensity (mw) Supp. Figure 4 - Locl proes of potentil steps t domin wlls OC (V) V O 1.2 1..8.6.4.2 DSO d Pt +V V OC (V). 2 4 6 8.4.2 Distnce (µm). 2 4 6 Distnce (µm) nture nnotechnology www.nture.com/nturennotechnology 7 21 Mcmilln Pulishers Limited. All rights reserved.
supplementry informtion doi: 1.138/nnno.29.451 Supp. Figure 5 I-V dt for different electrode spcings nd film thicknesses 8x1 6x1 Current (Isc, A) 2 µm 5 µm 1 µm 2 µm 4x1 2x1-2 5 nm -1 4.x1 2 nm -1 2.x1 1 nm. -15-1 -5 Voltge (V) 8-1 6.x1 5 1 15 2 Electrode distnce (µm) nture nnotechnology www.nture.com/nturennotechnology 21 Mcmilln Pulishers Limited. All rights reserved.
doi: 1.138/nnno.29.451 supplementry informtion Supp. Figure 7 Time dependent I-V dt 6x1 on on on 4x1 2x1 off off off 1 2 3 4 5 Time (min) Voltge (V) -5-1 -15 6x1 4x1 2x1 V oc I sc 1 2 3 4 5 6 Time (min) Supp. Figure 8 Ferroelectric switching thresholds 3x1-9 2x1-9 1x1-9 -1x1-9 -2x1-9 -3x1-9 3x1 2x1 1x1-1x1-2x1-3x1 -.8 -.4..4.8 Voltge (V) -1-5 5 1 Voltge (V) 5V step 1V step 2x1-1 1x1-1 2x1 1x1.1V step -1x1.25V step.5v step 1.V step -2x1 -.6 -.3..3.6 Voltge (V).1V step -1x1-1.25V step.5v step -2x1-1 1.V step -1-5 5 1 Voltge (V) nture nnotechnology www.nture.com/nturennotechnology 9 21 Mcmilln Pulishers Limited. All rights reserved.