Solar cells from colloidal nanocrystals: Fundamentals, materials, devices, and economics
This article is not included in your organization's subscription. However, you may be able to access this article under your organization's agreement with Elsevier.
aSchool of Chemical Engineering and the Energy Center, Purdue University, West Lafayette IN 47906, United States
bChemical and Biosciences Center, National Renewable Energy Laboratory, Golden CO 80401, United States
Received 13 April 2009;
revised 5 May 2009;
accepted 6 May 2009.
Available online 13 May 2009.
Abstract
Recent advances in colloidal science are having a dramatic impact on the development of next generation low-cost and/or high-efficiency solar cells. Simple and safe solution phase syntheses that yield monodisperse, passivated, non-aggregated semiconductor nanocrystals of high optoelectronic quality have opened the door to several routes to new photovoltaic devices which are currently being explored. In one route, colloidal semiconductor nanocrystal “inks” are used primarily to lower the fabrication cost of the photoabsorbing layer of the solar cell. Nanocrystals are cast onto a substrate to form either an electronically coupled nanocrystal array or are sintered to form a bulk semiconductor layer such that the bandgap of either is optimized for the solar spectrum (1.0–1.6 eV if the photon to carrier quantum yields less than 100%). The sintered devices (and without special efforts, the nanocrystal array devices as well) are limited to power conversion efficiencies less than the Shockley–Queisser limit of 33.7% but may possibly be produced at a fraction of the manufacturing cost of an equivalent process that uses vacuum-based deposition for the absorber layer. However, some quantum confined nanocrystals display an electron-hole pair generation phenomena with greater than 100% quantum yield, called “multiple exciton generation” (MEG) or “carrier multiplication” (CM). These quantum dots are being used to develop solar cells that theoretically may exceed the Shockley–Queisser limit. The optimum bandgap for such photoabsorbers shifts to smaller energy (0.6–1.1 eV), and thus colloidal quantum dots of low bandgap materials such as PbS and PbSe have been the focus of research efforts, although multiple exciton generation has also been observed in several other systems including InAs and Si. This review focuses on the fundamental physics and chemistry of nanocrystal solar cells and on the device development efforts to utilize colloidal nanocrystals as the key component of the absorber layer in next generation solar cells. Development efforts are put into context on a quantitative and up-to-date map of solar cell cost and efficiency to clarify efforts and identify potential opportunities in light of technical limitations and recent advances in existing technology. Key nanocrystal/material selection issues are discussed, and finally, we present four grand challenges that must be addressed along the path to developing low-cost high-efficiency nanocrystal based solar cells.
Fig. 1. The cost of electrical power from photovoltaic systems is shown as a function of the total upfront cost and the module power conversion efficiency. MC, SP, SIII, and SI are the manufacturing cost, average selling price, installed cost for a utility scale system, and installed cost for a residential system, respectively. SIII includes BOS cost for an on-grid system, and SI includes BOS for on- and off-grid operation with battery storage. The $/Wp were converted to ¢/kWh assuming a module lifetime of 20 years, a 5% cost of borrowing, a 1% yearly operating (maintenance) cost, and an average solar insolation of 200 W/m2 (which is about 5 h of full intensity sunlight/day). Costs are based on 2009 data. The designation of 1st, 2nd, or 3rd Generation is based on the manufacturing cost and potential module efficiency.
Fig. 2. Theoretical limit of efficiency for solar cells that utilize MEG but otherwise obey the same physics as the Shockley–Queisser balance calculation. (a) Several possible quantum yield profiles. (b) Resulting theoretical efficiency. Mmax is the theoretical limit considering only energy conservation. M2 is the case where the absorber has 200% quantum yield for photons with an energy greater than 2 Eg. M1 corresponds to the Shockley–Queisser limit. L2 and L3 are defined by a threshold of 2 Eg and 3 Eg, respectively, with a linear increase in QY. Adapted and reproduced with permission from AIP from Hanna and Nozik [14•].
Fig. 3. Electronically coupled NCs that reside within a built-in electric field (E) suitable for device operation. The NC states (assuming the independent particle approximation) are labeled with two quantum numbers, n (the principal quantum number) and L (the angular momentum, S or P) for electrons 1Se, 1Pe, etc. and holes 1Sh, 1Ph, etc. Large potential barriers, with height barrier and width δ, impede exciton delocalization (depicted here as wavefunction overlap) while maintaining quantum confinement effects. For efficient devices competing rates (Г for carrier cooling, MEG, carrier trapping, and charge transfer) must be balanced so as to maximize carrier mobility and MEG while decreasing recombination and trapping pathways. As a result, the MEG efficiency may also be dependent on the surface chemistry and inter-NC coupling. These rates also depend upon the electric field present under operating voltage. While not depicted here, similar pathways for carrier relaxation, transport, and trapping also occur within the valence band (hole states).
Fig. 4. Schottky-junction NC solar cell. (a) Schematic of the device, ITO/Glass/NC-film/metal electrode. The NC-film is lightly p-doped and forms a Schottky-junction with the metal electrode. (b) Cross sectional SEM of a device with 250 nm NC film, scale bar = 100 nm. (c) typical I–V curve for NC device with 4 nm PbSe NC film treated with 1,2 ethanedithiol. (d) and (e) IQE analysis devices incorporating either a 60 nm or 125 nm NC film. (f) Equilibrium band diagram for the Schottky device. Light enters the ITO side (field-free region for thicker devices). The depletion width is 150 nm. Adapted and reproduced with permission from ACS from Luther et al., Nano. Lett., 8, 3488 [49••] and Law et al., Nano. Lett., 8, 3904 [76].
Fig. 5. Hybrid semiconductor nanocrystal and amorphous silicon PV device. (a) Band alignment of the ITO, PbS NC film, a-Si and Al back electrode. Photogenerated electrons and holes are collected at the Al and ITO electrodes. (b) EQE of PbS/a-Si device overlaid with the NC optical density. Inset shows the EQE of devices with and without the a-Si. (c) photoconductivity of the a-Si, treated and untreated PbS film. (d) The IV characteristics of the device. Reproduced with permission from ACS from Sun et al. Nano Lett, 9, 1235 [77••].
Fig. 6. Heterojunction photovoltaic device using printed colloidal quantum dots. (a) EQE and the percent absorbed by the NCs for three different devices each with a different sized NCs. The onset of photoconductivity followed the effective bandgap of the NCs showing the NCs retain their quantum confinement within the devices. Part (b) displays the light and dark I–V curves for one device with and without the NCs. The operation of the device is shown in the inset. Photogenerated electrons are blocked by the TPD layer and inject into the ITO electrode, while the holes are transferred to the TPD and diffuse to the PEDOT electrode. Reproduced with permission from ACS from Arango et al. Nano Lett, 9, 860 [70].
Fig. 7. Copper (I) sulfide nanocrystals and resulting solar cells on flexible substrates. (Top image) TEM image of Cu2S nanocrystals. (Bottom Image): I–V curves under illumination before and after bending the flexible NC solar cell. The dark I–V curve is shown in black. Inset: Photograph of the cell being bent. Adapted and reproduced with permission from ACS from Wu et al., Nano Lett. 8, 2551–2555 [24•].
Fig. 8. CuInSe2 solar cells fabricated from a nanocrystal ink. FESEM images of (a) CuInSe2 nanocrystal thin film after casting and (b) after sintering in a Se/Ar atmosphere. (c) A photograph of the completed solar cells after chemical bath deposition of CdS, sputtering of i-ZnO and ITO, and evaporation of metal contacts. (d) The I–V characteristic of the finished device. Reproduced with permission from ACS from Guo et al. Nano Lett. 8, 2982–2987 [23•].
barrier and width δ, impede exciton delocalization (depicted here as wavefunction overlap) while maintaining quantum confinement effects. For efficient devices competing rates (Г for carrier cooling, MEG, carrier trapping, and charge transfer) must be balanced so as to maximize carrier mobility and MEG while decreasing recombination and trapping pathways. As a result, the MEG efficiency may also be dependent on the surface chemistry and inter-NC coupling. These rates also depend upon the electric field present under operating voltage. While not depicted here, similar pathways for carrier relaxation, transport, and trapping also occur within the valence band (hole states).
250 nm NC film, scale bar = 100 nm. (c) typical I–V curve for NC device with 
Corresponding authors. School of Chemical Engineering and the Energy Center, Purdue University, West Lafayette IN 47906, United States. 
