Supplementary Figure S1 TEM images of a synthesis batch of PbS and Bi-doped PbS QDs (Bi/Pb=3.2%) and corresponding size distribution histograms (100 QDs population in each sample) yielding average diameters of 3.5 and 3.6nm respectively. Scale bars are 10nm. 1
Supplementary Figure S2 TEM imaging on an additional to the one shown in Supplementary Figure S1 batch of pure and doped quantum dots with nominal bismuth doping densities ranging from 1.1 to 4.5% as indicated within images. This was done in order to further verify the similarity in the size and morphology between pure and doped quantum dots. The corresponding from these speciments size histograms are further shown in Supplementary Figure S3. 2
Supplementary Figure S3 Size histograms from TEM micrographs of PbS and Bi-doped PbS QDs (as shown in Supplementary Figure S2) with varying nominal doping concentrations as indicated within images (100 particle population from each sample). Size and distribution refers to average value and standard deviation of measured sizes. 3
Supplementary Figure S4 The similarity between the PbS and Bi:PbS, Bi/Pb=3.2% samples shown in Supplementary Figures S2, was further confirmed by comparing similar quality micrographs acquired with scanning HAADF-STEM imaging technique, which provides images with strong Z contrast. This figure includes HAADF-STEM images (scale bars are 5nm) of PbS and Bi:PbS, Bi/Pb=3.2% quantum dots and related size distribution histograms (200 QDs populations). Size and distribution refers to average value standard deviation of measured sizes. 4
Supplementary Figure S5 Crystallinity of individual doped and pure quantum dots as observed, using high resolution TEM (HRTEM) imaging. HRTEM of PbS and doped Bi:PbS (Bi/Pb=3.2%) quantum dots. 5
Supplementary Figure S6 Analyzed XPS Pb4f, S2s, O1s spectra of PbS and Bi:PbS QD films. 6
Supplementary Figure S7 Analyzed XPS C1s, Bi4f,andS2p spectra of PbS and Bi:PbS QD films. 7
Supplementary Figure S8 XRD pattern and zoomed areas of interest of PbS and Bi:PbS QDs with varying doping concentrations, indexed according to simulated reflections (green lines) for a PbS rock salt structure (a=5.93 Å). Peaks in zoomed areas are fitted with Gaussian functions. 8
Supplementary Figure S9 a) logi-v characteristics of Schottky type solar cell architecture of 200nm pure PbS QDs in between glass/ito and varying top electrodes, b) same as for a but for Bi:PbS with Bi/Pb=3.2 %. c) Schematic representation of the Bi:PbS/MoO 3 and NiO/Bi:PbS Schottky devices d) IV curves of ITO/Bi:PbS QD/ MoO 3 and ITO/NiO/Bi:PbS QD /Ag Schottky solar cells, e) EQE spectral response of cells shown in d, indicating that the photoresponce stops at approx. 1200nm light wavelength (approx. 1eV) f) Cap-V characteristics of Schottky architectures bases on NiO and Bi:PbS of varying Bi/Pb ratios, and measured majority (electron) carrier densities and depletion widths. 9
Supplementary Figure S10 Current-Voltage (logi-v) characteristics at light (AM1.5) and dark of ITO/NiO (5nm)/ Bi:PbS (200nm)/Ag with varying Bi/Pb precursor ratio 10
Supplementary Figure S11 Spectral External Quantum Efficiency (EQE) measurement of an ITO/ PbS QD (200nm)/ LiF(1nm)/ Al Schottky device. The graph indicates that the photoresponce stops at 1200nm light wavelength (approx. 1eV). 11
Supplementary Figure S12 Current voltage (logi-v) characteristics at dark and light for ITO/200-250nm Bi:PbS/ LiF (1nm)-(100nm)Al or MoO 3 Schottky junctions for Bi/Pb=0.1% and 0.5%. It has been found that for doping densities Bi/Pb=0.1%, 0.5%, Bi:PbS quantum dot films can form Schottky junctions with both Al (LiF 1nm/100nm Al) and MoO 3. For the case of Bi/Pb=0.1% the junction with MoO 3 is weak (Voc=0.1V) and that indicates that the material is probably p-type. Mott-Schottky analysis of the junction with Al give a hole carrier density of 4.7*10 16 cm -3 which is lower compared to PbS. For the case of Bi/Pb=0.5% both junctions with Al and MoO 3 exhibit strong rectifying behavior. Due to this behavior, the type of charge carriers that could be studied by Mott-Schottky analysis of such junctions is not clear. We call this as the transition region between p- and n-type like regions for smaller (<<0.5%) and larger (>>1%) doping concentrations. 12
1,0x10-8 ITO/ 200nm Bi:PbS (Bi/Pb=4.5%) / MoO 3 /Au / Ag 6x10 16 Cap (F) 8,0x10-9 6,0x10-9 4,0x10-9 1/C 2 (F -2 ) 5x10 16 4x10 16 3x10 16 2,0x10-9 0,0-0,5 0,0 0,5 Voltage (V) 2x10 16 1x10 16-0,4-0,2 0,0 0,2 0,4 0,6 Voltage (V) Supplementary Figure S13 C-V and 1/C 2 -V characteristics of an ITO/Bi:PbS/MoO 3 /Au/Ag Schottky-type device 13
3,6x10 16 1/C 2 (F -2 ) 3,4x10 16 3,2x10 16-1,0-0,8-0,6-0,4-0,2 0,0 0,2 0,4 Voltage (V) Supplementary Figure S14 Current voltage (logi-v) at light (AM1.5) and dark, and 1/C 2 -V (at dark) characteristics of an ITO/ 200nm PbS CQD /LiF (1nm)/ Al (160nm) Schottky junction device From Mott-Schottky analysis of the junction we find that the hole density of the PbS QDs is N h = 6.27*10 16 cm -3, which is very close to typical known values for EDT treated PbS QD films. 14
Supplementary Figure S15 Figures of merit of several bilayer homojunction solar cell devices/pixels. Each point represents a single cell/substrate. Dashed red lines separate points associated with different QD synthesis batches. The average power conversion efficiency obtained from these data is 2.2 ±0.4% and the maximum achieved is 3%. Power conversion efficiency variations are mainly associated with variations in the short circuit current (11.9 ±1.9mA/cm 2 ) and fill factor (47 ±4%), while 0.38-0.42V open circuit voltages are routinely observed. 15
Normal Inverted Capacitance (nf) 10 8 6 4 3L Bi+PbS/6L PbS 4L Bi+PbS/6L PbS 6L Bi+PbS/6L PbS Capacitance (nf) 15 12 9 6 3 4L PbS/5L Bi+PbS 6L PbS/5L Bi+PbS 8L PbS/5L Bi+PbS 2-0.5 0.0 0.5 1.0 Voltage (V) 0-0.5 0.0 0.5 1.0 Voltage (V) Supplementary Figure S16 Capacitance-voltage (C-V) characteristics at dark of Normal and Inverted homojunctions with varying PbS and Bi:PbS QD film thicknesses (number of spincasted layers). We see that for less than 6 layers of PbS, PbS is fully depleted (capacitance is dropping with reverse bias) but for more than 6 layers PbS is not fully depleted (capacitance is not dropping with reverse bias). Thus we estimated that the depletion width W p in the p-type PbS layer of the homojunction is 150nm, which also agrees with the results of supplementary table S5 and EQE spectra of Supplementary Figure S17. From Mott-Schottky analysis of ITO/PbS/LiF/Al we have determined that N h =6.27*10 16 cm -3. We determine the electron concentration N e using the following relationship for p-n junctions 41 : W p =[2εN e V bi /(qn h (N e +N h ))] 1/2, where the dielectric constant ε is assumed to be constant across the homojunction, W p =150nm is the depletion width within the p-layer, and V bi =0.44eV. We calculate that N e =1.4 *10 17 cm -3 which is in good agreement with the measured carrier densities of Bi:PbS from the Schottky-type devices with NiO and MoO 3. We similarly calculate the depletion width W n in the n-type Bi:PbS layer to be 68nm. This value is close to the thickness of approx. 50nm of 2 spin casted layers (2L) of Bi:PbS for which the efficiency of the homojunctions is maximum and drops for thicker layers, as indicated by Supplementary Table S4 and Supplementary Figure S17. 16
50 40 Normal 3L Bi doped PbS/6L PbS 4L Bi doped PbS/6L PbS 6L Bi doped PbS/6L PbS 60 50 Inverted 4L PbS/ 5L Bi doped PbS 6L PbS/ 5L Bi doped PbS 8L PbS/ 5L Bi doped PbS EQE (%) 30 20 EQE (%) 40 30 20 10 10 0 400 600 800 1000 1200 Wavelenght (nm) 0 400 600 800 1000 1200 Wavelenght (nm) Supplementary Figure S17 EQE spectral photo-response of Normal and Inverted homojunctions with varying PbS and Bi:PbS QD film thicknesses (number of spin-casted layers). For the normal homojunction we see that as Bi:PbS number of layers increases to more than 2 (50nm), EQE drops. For the inverted homojunction we see that 6 layers (150nm) of PbS QDs is the optimum thickness for best solar cell efficiency. 17
Supplementary Figure S18 Current voltage (logi-v) characteristics at dark and light of homojunctions made with Bi:PbS of varying Bi/Pb ratio, employing PbS QDs doped postsynthetically with Bi. The characteristics indicate that post-synthetic doping of PbS with Bi is also effective for creating homojunction solar cells. 18
Supplementary Figure S19 Raman spectra of PbS and Bi:PbS quantum dot films and Bi 2 S 3 nanocrystalline films. This measurement shows that no Bi 2 S 3 phase can be observed for either doping method 19
Supplemetary Table S1. Overall stoichiometry from XPS peak area analysis as presented in Supplementary Figures S6 and S7 Peak C1s Pb4f Bi4f S2s O1s % composition of Bi:PbS (precursor Bi/Pb=3.2%) % composition of PbS 34.54 31.43 25.33 28.25 1.37 0.00 28.13 29.59 10.62 10.72 20
Supplementary Table S2 Stoichiometry from XPS peak area analysis as presented in Supplementary Figures S6 and S7, considering only Pb, Bi, S. According to it, the Bi/Pb ratio of the doped sample is 5.4% while for the same material the ratio as measured from ICP-OES is 4.1%, thus the two techniques are in good agreement. Also according to it, the (Pb+Bi)/S ratio is 0.95 for both reference and doped QD samples Element Pb Bi S % composition of Bi:PbS (precursor Bi/Pb=3.2%) % composition of PbS 46.20 48.84 2.51 0.00 51.30 51.16. 21
Supplementary Table S3 Calculated d hkl plane spacing, and calculated lattice constant a from spacing of respective (hkl) planes, using the XRD measurements presented in Supplementary Figure S8. Results indicate increasing lattice constant with increasing doping density as further discussed in Supplementary Note S1 Sample d 111 (Å) lattice constant from d 111 (Å) d 220 (Å) lattice constant from d 220 (Å) PbS 3.435017 5.949623 2.108554 5.963892 Bi:PBS, Bi/Pb=2.2% 3.44779 5.971747 2.112783 5.975852 Bi:PBS, Bi/Pb=4.5% 3.455432 5.984985 2.116555 5.986522 22
Supplementary Table S4 Normal homojunction figures of merit with varying number of Bi:PbS spin-casted layers (and thus thickness). We see that for the normal architecture efficiency is maximized for two layers of Bi:PbS (50nm) 23
Supplementary Table S5 Inverted homojunction figures of merit with varying number of PbS spin-casted layers. We see that 6 layers (150nm) of PbS QDs is the optimum thickness for best solar cell efficiency 24
Supplementary Note 1 XRD study The overall spectra as seen in supplementary Figure S8, indicate that the all three materials have the PbS rock salt structure (reference lattice constant a=5.93å). The center of the (111) and (220) peaks is determined by fitting Gaussian lines on the peaks as shown. We calculate the relevant planes spacing d hkl by using Bragg s Law λ=2*d hkl *sin(θ) (S1) (where λ= 1.540598Å the provided X-ray wavelength of the instrument, and angle units are radians). We calculate the lattice constant a using the following relationship a=(h 2 +k 2 +l 2 ) 1/2 *d hkl (S2) The resulting spacings of the (111) and (220) planes and the lattice constants from both sets of planes are shown in Supplementary Table S3. We find a systematic increase of the lattice constant by increasing doping density, and the difference between the pure and the heaviest doped QDs is 0.02-0.03Å depending on the type of planes used to calculate it. The measured lattice constants are larger than the 5.93Å value for bulk PbS and this has been previously reported by Jiang Tang et al. 36 for such small quantum dots. In that same work an increase of the lattice constant by 0.01Å was reported for size change in the quantum dot by diameter up to 0.5nm for small (from 3.2 to 2.7nm diameter) dots, and a 0.02Å lattice constant difference was reported for a QD size difference of 3.5nm (for 6.2nm and 2.7nm quantum dot diameters). The TEM analysis shows that we do not have such large differences in diameter between pure and doped QDs, thus we attribute the systematic lattice constant change that we observe to the incorporation of Bi in the PbS structure. 25
Supplementary Note 2 Schottky diodes studies We first sought to identify potential differences that doping introduces in the ability of the dots to form Ohmic- or Schottky-type contacts with a range of electrode materials. We fabricated glass/ito/ QDs films/ M planar devices where QDs were Bi-doped or pure PbS, and the evaporated electrode contact M was one of Ag, Al (LiF 1nm/ 160nm Al), or MoO 3. While as observed from the IV characteristics at dark and under A.M.1.5 illumination given in Supplementary Figure S9.a, and in accordance with the literature 30, 37-39, pure p-type PbS quantum dots with Fermi level E F =4.8 to 4.9eV 38 form near-ohmic contact with the high work function (φ) electrode material MoO 3 (φ MoO3 =5eV 30 ) and Schottky-type contact with the lower φ metals Ag and Al (φ Ag =4.3eV 37, φ Al =4.2eV 39 ), Bi-doped PbS QDs with Bi/Pb>2.2% have an opposite behavior as seen in Supplementary Figure S9.b. They form near-ohmic contacts with Ag and Al, and Schottky-type contact with MoO 3. In the device with MoO 3 as contact, the electrons flow towards the ITO and away from the QD/ MoO 3 interface. This is another indication that bismuth acts as an electron donor and thus raises the Fermi level of PbS. We demonstrated this further in another device architecture employing NiO as another high work function contact (φ NiO =5eV 40 ), this device architecture is: glass/ito/ NiO x /Bi:PbS/Ag where the sputtered nickel oxide layer thickness is only 5nm and the QD film thickness is 200nm. The architectures of the Schottky-type devices based on Bi:PbS are schematically described in Supplementary Figure S9.c. The current-voltage (IV) (Supplementary Figure S9.d) and external quantum efficiency (EQE) characteristics of these devices (Supplementary Figure S9.e) also demonstrate the ability of the doped particles to be utilized in optoelectronic devices with spectral photoresponce extending to the near infrared (NIR). Nickel oxide has been used in excitonic photovoltaics as a hole transporting material with E F =5 to 5.2 ev and we hypothesized that it will form a better Schottky junction with bismuth doped PbS dots. Indeed, the device s open circuit voltage (V OC ) is superior (0.34V) compared to the Bi:PbS/MoO 3 junction (0.16V) under AM1.5 illumination. Comparison of the EQE spectra of the two architectures (supplementary Figure S9.e) also indicates that the depletion width within the doped PbS should be short because when the junctions interface is away from the optical window (ITO) the EQE in the UV drops while EQE in the NIR, which is absorbed further inside 26
the device, increases. Mott-Schottky analysis of capacitance voltage characteristics of both architectures at dark after illumination allow us to calculate the majority carrier density N and depletion width W for the QD layer, the NiO based architecture gives N=1.14*10 17 cm -3 and W=72nm and the MoO 3 based architecture gives N=1x10 17 cm -3, W=85nm for the case of nominal Bi/Pb=4.5% doped quantum dots. Using the NiO/Bi:PbS architecture we also demonstrate that the calculated majority carrier density N and depletion width W changes as a function of Bi doping density as measured from the Cap-V characteristics shown in Figure S9.f, based on Mott-Schottky analysis. For those diodes, the I-V characteristics at light (AM1.5) and dark are shown in Supplementary Figure S10. 27
Supplementary Note 3 Calculating Carrier density N and depletion Width W from Schottky junctions based on Bi:PbS We plot the Capacitance-Voltage (at dark) data as 1/C 2 V, as shown in Supplementary Figure S13. We then apply linear fitting in the linear-looking region between maximum capacitance (minimum 1/C 2, close to V=0) towards V=0. We use our linear fitting parameters to calculate N, and V bi from the Mott-Schottky equation: (S3) where A is the area of our device, q is the elementary charge, ε is the dielectric constant of the semiconductor, N is the majority carrier density in the semiconductor, k B is Boltzman s constant, T is temperature, and V bi is built in potential of the junction. For the calculations we assume the following parameters: ε= 20* ε o, after previous published work 36, where ε o = 8.854 10 12 F/m is the vacuum permittivity A= 4 mm 2 k B T/q<<V bi Thus we find N=1*10 17 cm -3, V bi = 0.4V (a very similar V bi can also be estimated as the bias at the intersection of light and dark IV curves) We calculate the depletion region W from the equation: (S4) at 0V applied bias to be W=80nm We have done a similar analysis for the case of the NiO based Schottky-type structure ITO/ 5nm NiO / 160nm Bi:PbS / Ag (140nm) with varying Bi/Pb ratio of which the C-V characteristics are shown on Supplementary Figure S9.f. For those diodes, the I-V characteristics at light (AM1.5) and dark are shown in Supplementary Figure S10. 28
Supplementary Note 4 Synthesis via post-synthetic treatment, and diode application of Bi:PbS QDs. The reaction took place in a 3-neck flask attached to a standard Schlenk line. The method is described as follows: Bismuth oleate precursor was prepared by mixing the required amount of Bismuth acetate (0.0385g in an example) in 6ml of oleic acid and 20ml of octadecene, at 90 o C under vacuum overnight (16hr). Then the temperature was lowered to 70 o C and the atmosphere was switched to Ar. Subsequently 12ml of QD/toluene solution of appropriate concentration (0.016g/ml in an example) depending on the required Bi/Pb ratio, originally stored in a nitrogen atmosphere glovebox, were injected in the bismuth-precursor, under vigorous stirring. Upon injection the temperature of the solution in the flask drops, then the flask is heated until its temperature reach 70 o C, and the flask is removed from the hotplate and left cool to down naturally. QDs are further cleaned similarly to the method used for PbS synthesis. We have fabricated solar cells based on homojunctions employing such Bi:PbS dots prepared by post-synthetic treatment of PbS with bismuth. The solar cells diode character and photovoltaic figures of merit like power conversion efficiency η under light (light refers to 1sun AM1.5 illumination) shown below in Supplementary Figure S18 illustrate the functional effectiveness of the post-synthetic doping of PbS quantum dots with Bi. Raman spectroscopy was performed on nanocrystalline Bi:PbS prepared by adding bismuth during the PbS synthesis (Bi/Pb=4.5%) or via post synthetic treatment of PbS with Bi (Bi/Pb=25%), in order to further study whether there is formation of Bi 2 S 3 by either route. The spectra were compared with those of reference nanocrystalline PbS and Bi 2 S 3 as shown in Supplementary Figure S19. This measurement shows that no Bi 2 S 3 phase can be observed for either doping method. Considering that in this particular post synthetic treatment of PbS we used a high Bi/Pb ratio, this experiment also indicates that transformation of PbS to Bi 2 S 3 during our post synthetic doping approach is highly unlikely, due to the very different crystal structures of PbS and Bi 2 S 3 as well as the necessity for the anionic sublattice transformation. 29
Supplementary References 36. Tang, J. et al. Quantum Dot Photovoltaics in the Extreme Quantum Confinement Regime: The Surface-Chemical Origins of Exceptional Air- and Light-Stability. ACS NANO. 4, 869-878 (2010) 37. Brown, P. R. et al. Improved Current Extraction from ZnO/PbS Quantum Dot Heterojunction Photovoltaics Using a MoO3 Interfacial Layer. Nano Lett. 11, 2955-2961 (2011) 38. Gao, J., Luther, J. M., Semonin, O. E., Ellingson, R. J., Nozik, A. J., Beard, M. C. Quantum Dot Size Dependent J-V Characteristics in Heterojunction ZnO/PbS Quantum Dot Solar Cells. Nano Lett, 11, 1002-1008 (2011) 39. Johnston, K. W. et al. Schottky-quantum dot photovoltaics for efficient infrared power conversion. Appl. Phys. Lett. 92, 151115 (2008) 40. Olivier, J., Servet, B., Vergnolle, M., Mosca, M., Garry, G. Stability/instability of conductivity and work function changes of ITO thin films, UV-irradiated in air or vacuum: Measurements by the four-probe method and by Kelvin force microscopy. Synth. Metals. 122, 87-89 (2001) 41. Rath, A. K., Bernechea, M. Martinez, L., Konstantatos, G. Solution-Processed Heterojunction Solar Cells Based on p-type PbS Quantum Dots and n-type Bi2S3 Nanocrystals. Adv. Mater. 23, 3712-3717 (2011) 30