2.1.

Introduction

The issue that has received the most attention in recent InAs/GaAs QDSC research is the reduction of the $V_{\mathrm{oc}}$ when compared to the GaAs control cell.^26–31 Experimentally confirmed short circuit currents $J_{\mathrm{sc}}$ for both GaAs- and InAs QD-based solar cells are almost identical (note: Some InAs /GaAs QDSC showed slightly higher $J_{\mathrm{sc}}$ than GaAs SC, but the increase amount is fairly small and usually $<0.5\mathrm{mA}/{\mathrm{cm}}^2$).^15,32–35 Under such a situation, the lower $V_{\mathrm{oc}}$ observed in QDSC was attributed mainly to the larger dark recombination current when compared to GaAs control cell. To understand the dark recombination current in QDSC, modeling of the carrier dynamics is indispensable.^6,36–38 A simplified two-voltage region model is proposed by us, as shown in in Figs. 1(a)–1(c).^39,40

Fig. 1

Sketched band diagrams for GaAs SC and InAs/GaAs QDSC under (a) short-circuit condition, (b) low forward bias, and (c) high forward bias conditions.

For a p-i-n type QDSC under low forward bias voltage [see Fig. 1(b)], the voltage is mainly applied across the space charge region (SCR). For the recombination behavior within this region, the dark current for GaAs and InAs/GaAs QDSC is described using a two-diode model as follows:

GaAs SC:

InAs/GaAs QDSC:

When the p-i-n solar cell is under high forward bias [see Fig. 1(c)], the SCR region becomes negligible and the voltage drop is applied equally across the whole device; the dark currents for GaAs SC and InAs/GaAs QDSC are described as follows:

GaAs SC:

InAs/GaAs QDSC:

The reduction of $V_{\mathrm{oc}}$ in InAs/GaAs QDSC has recently become a central issue for further efficiency improvement. In this review, we have limited our focus on the works that have mainly dealt with interplay between carrier dynamics and device performance. Meanwhile, to ensure timeliness and currency, the works reviewed here were sorted and scrutinized from the literatures published in the past 2 to 3 years. The model described in this section will act as a supportive guide to better interpret the results summarized from the reviewed papers.

2.2.

Carrier Escape Nature and Electric Field Effect

The photoelectrons generated by InAs QD light absorption usually, at first, have to escape from the potential formed between InAs QD and GaAs buffer. Understanding the carrier escape mechanism is crucial to further improving the device performance.^42,43 Sellers et al.⁴² have evaluated the carrier escape mechanism in InAs/GaAs QDSC by using photocurrent measurements under sub-bandgap illumination. The 1.17-eV photon source and 0.8-eV photo source were chosen, and their fluencies were varied to reveal the fundamental carrier escaping principle. The inset Fig. 2(a) depicts the three competing mechanisms for carrier extraction from intermediate states in QD: tunneling (pathway 1), thermal (pathway 2), and optically driven (pathway 3).^37,44 When the sample was solely under illumination of a 1.17-eV photon source, photocurrent was found to increase linearly with laser fluence over the range of fluences studied, as shown in Fig. 2(b). The linear increase suggests that thermal and tunneling escape mechanisms dominate any carrier escape driven optically by 1.17-eV photons. The effect of adding a 0.8-eV photon source was presented in Fig. 2(c). Here, ${\mathrm{\Delta }}_{\mathrm{opt}}$ was used to describe the increased photocurrent due to the addition of 0.8-eV light. Again, they found that the ${\mathrm{\Delta }}_{\mathrm{opt}}$ value increased linearly with increasing 0.80-eV fluence under a fixed fluence value of 1.17-eV photon source. If the photocurrent was limited solely by the generation of carriers within the QDs, ${\mathrm{\Delta }}_{\mathrm{opt}}$ should become saturated for high fluences of 0.8-eV photons.⁴² The absence of saturation under these conditions suggests that there is a continuous supply of carriers that can be optically excited by 0.8-eV photons. The authors have proposed a carrier “retrapping model” to explain the experimental results.⁴² Based on this model, ${\mathrm{\Delta }}_{\mathrm{opt}}$ depends linearly on the fluence of 0.8-eV photons because the addition of an optical escape pathway reduces the mean time carriers spent in traps in either the QD or wetting layer and, therefore, increases the conductivity (therefore, the photocurrent) even for a fixed number of carriers.³⁰ It is worthwhile to mention here that in a recent work, Asahi et al.⁴⁵ have reported a saturation of two-step photocurrents. However, unlike the work reported here involving the two-step photoexcitation from valence band to intermediate states and from intermediate states to conduction band, Asahi et al.⁴⁵ observed the saturation by tuning “two-step” photoexcitation from valance band to conduction band and from intermediate states to conduction band.

Fig. 2

(a) Schematic of the InAs/GaAs p-i-n junction solar cell device band gap structure. (a) The carrier escape pathways following InAs QD excitation (pathway 1, tunneling escape; pathway 2, thermal escape; and pathway 3, optically driven escape) are shown. (b) Photocurrent as a function of 1.17 eV illumination fluence with and without the addition of $600\mathrm{mW}/{\mathrm{cm}}^2$ 0.80 eV light. (c) Change in photocurrent with the addition of 0.80 eV light to 1.17 eV illumination. Reprinted with permission from Ref. 42. © 2014 AIP Publishing LLC.

Related to the dynamic nature of carrier escape and separation mechanism, it has been shown that the excitonic dynamics of electrons and holes could be responsible for the nonadditive behavior of the photocurrent contributed by the QDs in a QDSC.³⁷ This nonadditive characteristic was observed when the total photocurrent of the cell was much lower than the sum of the photocurrents contributed by the barrier and the QDs separately, as reported for undoped InAs/GaAs solar cells.⁴⁶ Cedola et al.⁴⁶ have investigated the excitonic and nonexcitonic dynamic nature of the carrier escape of the InAs/GaAs QDSC by numerical simulation. The energy band model is illustrated in Fig. 3(a). By assuming identical or separate time constants for the intersubband carrier transfer processes in the ground and excited states, a device-level model developed in combining drift-diffusion equations for the barrier and rate equations for the QD kinetics was successfully applied.^47,48 Figure 3(b) shows their simulated results. The nonexcitonic cell clearly shows a nearly linear behavior in the short-circuit condition ($J_{\text{Non}-\text{excitonic}}\approx J_{\text{Non}-\text{excitonic},\text{filtered}}+J_{\text{Reference cell}}$). On the other hand, the excitonic cell exhibits a strong nonlinear operation $(J_{\text{Excitonic}}\ne J_{\text{Excitonic},\text{filtered}}+J_{\text{Reference cell}}$).⁴⁹ The linear behavior of the nonexcitonic QDSC indicates that practically all the extra carriers contributed by the QDs are being collected at the cell contacts.^40,42 On the contrary, the nonadditive characteristic shown by the QD current in the excitonic QDSC suggest that all the carriers photogenerated at the QDs are recombining. Further analysis has been performed by visualizing the processes of carrier interchange between states at each QD layer. The results revealed that the increment of the recombination is mainly due to the increased hole population in the QDs while playing a central role in the response of the excitonic device.

Fig. 3

(a) Energy band model of the InAs/GaAs system, as considered in this work. (b) $J-V$ characteristics of nonexcitonic and excitonic QDSCs, simulated under full (solid lines) and filtered ($\lambda >870\mathrm{nm}$, dashed lines) AM1.5G solar spectrum illumination. The additive feature of the QD contribution can be inferred by observing the reference cell and QDSCs currents at short circuit. Reprinted with permission from Ref. 46. © 2016 IOP Publishing.

The effect of the internal electric field on carrier escape and separation was investigated quantitatively by Kasamatsu et al.⁵⁰ They have experimentally fabricated InAs/GaAs QDSC with different built-in electric fields by controlling the thickness of intrinsic buffer layers surrounding the InAs/GaAs QDs. The internal electric fields applied to the QDs were estimated as 46 and $193\mathrm{kV}/\mathrm{cm}$ when the total intrinsic layer thicknesses were 299.3 and 69.8 nm, respectively. The authors calibrated the electric field effect by monitoring the carrier dynamic behavior using time-resolved photoluminescence (PL) measurements.⁵⁰ Figure 4 displays the detection wavelength dependence of the PL decay profile measured at 3K. The sample with $46\mathrm{kV}/\mathrm{cm}$ shows a double exponential decay. The rapid initial decay is slightly shorter than 2 ns, and the following slow decay becomes longer than 2 ns. For an extremely large internal electric field as $193\mathrm{kV}/\mathrm{cm}$, the decay is very fast and can be described by double exponential components of s1 and s2 ($\mathrm{s}1<\mathrm{s}2$).⁵¹ A strong electric field of $193\mathrm{kV}/\mathrm{cm}$ causes tunneling-assisted electron escape that occurs easily. The electron escape rapidly reduces the PL intensity with a time constant s1 of 0.46 ns. This short lifetime in the high electric field will prevent photoexcitation of electrons in the intermediate band (IB) in QDs.^51–53 Meanwhile, as mentioned in Sec. 2.1, the dark saturation current $J_0$ is treated as radiative recombination based on the detailed balance principle. For the sample with a high internal electric field, the existence of additional tunneling recombination path will increase the dark saturation current $J_0$, and the $V_{\mathrm{oc}}$ of this SC eventually reduces according to Eqs. (2) and (4).^37,50

Fig. 4

Detection wavelength dependence of PL decay profile measured at 3 K for (a) a reference sample and SCs at (b) 46 and (c) $193\mathrm{kV}/\mathrm{cm}$. The excitation laser wavelength was 800 nm, which provides excitation above the GaAs barrier. The excitation powers for (a), (b), and (c) were 100, 100, and $250\mu \mathrm{W}$, respectively. Reprinted with permission from Ref. 50. © 2014 AIP Publishing LLC.

In addition to the internal electric field, the external electric field has also been reported to exert influence on the carrier dynamics. Shiokawa et al.⁵⁴ have reported an inplane ultrahigh-density InAs (2.2 mL) QDs with density of $3\sim 5\times {10}^{11}{\mathrm{cm}}^{-2}$ grown on the GaAsSb/GaAs(001) by molecular beam epitaxy. By applying voltage ranging from $+1.0\mathrm{V}$ to $-1.0\mathrm{V}$, the PL decay time showed strong bias dependence, as shown in Fig. 5(a). In particular, the long decay time of 10 ns was obtained at $+1.0\mathrm{V}$.⁵⁴ By applying forward bias, it favors the formation of in-plane coupling among the QDs. This is again confirmed in Fig. 5(b) by the PL peak energy shift under different bias voltages. For the forward bias condition, the PL peak shifted toward the low energy side because the enhanced coupling among QDs caused the electron to relax at much lower ground energy states in the QDs with large size.^22,23,55 Meanwhile, the reverse bias was also found to shift the PL peak energy to long wavelength side. This was partially attributed to the strong Stark effect and partially to the fast electron escape rate, which greatly suppressed the high energy interband transition in the QDs.

Fig. 5

(a) Relationships between PL decay time spectra and bias voltage. Ultrahigh density InAs QDs with 2.2 mL was inserted into GaAs pn-junction. (b) PL spectra of ultrahigh density InAs QDs with 2.2 mL as a function of bias voltage. Reprinted with permission from Ref. 54. © 2013 IEEE Publishing.

2.3.

Doping Effect on InAs/GaAs Quantum Dot Solar Cell Performance

Typically, QDs grown in the intrinsic region are subjected to high electric field, and carriers are quickly swept away from the nanostructures as soon as they are excited into the bulk continuum.^56–60 Doping QD in the intrinsic region will affect the band profile and, therefore, the electric field in the intrinsic region. Accordingly, the change in electric field will also affect carrier dynamics as well as the solar cell device performance. Polly et al.⁶¹ have reported on both the theoretical and experimental results of the effect of delta-doping in the InAs/GaAs QD on the solar cell performance. Figure 6(a) shows the simulated results where the doping of $4{\mathrm{e}}^{-}/\mathrm{QD}$ (${\mathrm{e}}^{-}=\text{electron}$) has dramatically flattened the i-region and has shifted the intrinsic region to the emitter. A consequence of this shift in the electric field can be seen in Fig. 6(b), which shows the reduced SRH recombination rate at 1.0 V applied bias. They have also found that doping the hole ($9{\mathrm{h}}^{+}/\mathrm{QD}$) pushed the intrinsic region toward base layer while reducing the SRH recombination similarly to the electron doping.⁴⁷ As shown in Fig. 6(c), the experimental results were consistent with the simulated results. The $8{\mathrm{e}}^{-}/\mathrm{QD}$ doping resulted in the highest $V_{\mathrm{oc}}$ of 0.932 V, which can be considered a direct effect from the decreased SRH recombination due to the shift of intrinsic region.^29,40 However, the doping-induced band flattening has weakened the carrier collection due to the decreased electric field in the QD region. This negative effect was directly linked to the reduction of $J_{\mathrm{sc}}$ for all doped samples. It is interesting to note that the experimental results from Polly et al. are not consistent with the theoretical simulation results by Yoshida et al.⁶² The discrepancy is probably because in the simulation an ideal IBSC operation modeling was applied including the electron occupancy in IB, which is usually more difficult to realize by the current doping technique. In other words, the doping in the IB is not able to form the Fermi-level, as shown in the IB region of Fig. 4(b), in the work of Yoshida et al.⁶² Therefore, besides the region near the top emitter and the bottom base layer, the occupancy rate in other regions of IB is also not close to the optimal value of $1/2$ used in the work of Yoshida et al.

Fig. 6

(a) APSYS modeled band diagram for $0{\mathrm{e}}^{-}/\mathrm{QD}$ (black) and $4{\mathrm{e}}^{-}/\mathrm{QD}$ (blue) structures in the dark at 0 V applied bias. (b) APSYS modeled SRH rates at 1.0-V applied bias across the i-region of varying delta-doping levels. (c) Measured 1-Sun AM0 $J-V$ characteristics of devices. Reprinted with permission from Ref. 61. © 2014 IEEE Publishing.

With respect to the doping effect on QDSC, Li et al.⁶³ have recently reported interesting results regarding electron-doping in QD devices, especially its potential impact on the electron capture potential, which affects (1) carrier collection efficiency and (2) below bandgap photon absorption via transitions to quantum confined states. Figure 7(a) shows the current–voltage characteristics of the fabricated solar cells with different electron doping concentrations. The $0\mathrm{e}/\text{dot}$, $2\mathrm{e}/\text{dot}$ to $4\mathrm{e}/\text{dot}$ devices exhibit a $J_{\mathrm{sc}}$ of 10.1, 11.8, and $12.6\mathrm{mA}/{\mathrm{cm}}^2$, respectively. The authors have interpreted the results as a consequence of so-called “charging” effect. As depicted in Fig. 7(b), when above-bandgap photogenerated electrons move through the QDs layer, the repulsive force exerted by the negatively charged QDs can alter the electron trajectory in such a way as to reduce the probability of electron trapping.⁶⁴ The Coulomb potential exerted by the negatively charged QDs is a competing process that acts on the mobile electrons and competes with the QDs trapping potential. When electrons are captured, the trapping effects are progressively deactivated.⁶⁵ For the effect (2), the authors found that the $0\mathrm{e}/\text{dot}$ device had a much higher external quantum efficiency (EQE) value measured at the QD transition energy (1.1 eV) as compared to the doped QD devices. This was considered a result of the available number of unoccupied confined electron states being reduced with further doping. According to the Fermi’s Golden rule, the total transition rate and the absorption in QDs decrease correspondingly, as schematically illustrated in Fig. 7(c).^37,38 These conclusions were further verified by the PL results shown in Fig. 7(d) in which the intensity of QDs emission reduces as the doping increases.

Fig. 7

(a) Current–voltage characteristics of undoped, $2\mathrm{e}/\mathrm{dot}$, and $4\mathrm{e}/\mathrm{dot}$ device. (b) Electrons transport through one QD that affected both QD capturing potential and negatively charged excitons repulsive Coulomb potential. (c) Illustration of the effect of band bending on quantum dot states absorption. (d) PL of 0e, 2e, and $4\mathrm{e}/\mathrm{dot}$ devices. With further electron-doping, the PL signal at QD ground state (centered at 1.08 eV) reduces while PL signal at wetting layer state (centered at 1.27 eV) enhances. Reprinted with permission from Ref. 63. © 2015 AIP Publishing LLC.

2.4.

Effect of Quantum Dot Location on Carrier Transportation and Recombination

QD position in the intrinsic region also showed great influence on carrier dynamics and the device performance.^41,66–68 Driscoll et al.⁶⁹ performed simulations on three positions for InAs QD located in the intrinsic region: (1) near the n-doped base, (2) exact center of i-region, and (3) near the p-doped emitter.⁶⁹ From the simulation, they found that $J_{\mathrm{sc}}$ was nearly identical for the three positions as all were located in a high electric field region and carrier collection remained efficient for all three conditions. However, the model predicts an additional loss of 20 mV in the $V_{\mathrm{oc}}$ for the QD located in the exact center of the intrinsic region versus those located near the doped edges.⁶⁹ The reduction of the $V_{\mathrm{oc}}$ was mainly attributed to SRH recombination, which showed the maximum value when the electron and hole density were similar at the place such as the center of the intrinsic region.⁶⁶ They have also fabricated three samples experimentally with the three QD locations mentioned above to verify the simulated results.⁶⁹ Figure 8(a) summarized the EQE results for the three samples along with the baseline GaAs cell. It is clear that the bulk spectral response from the QD devices are similar to that of the baseline cell, indicating that neither the introduction of the QDs nor their position has had any adverse effects on the carrier collection efficiency.⁶⁹ However, dramatic variation in the $V_{\mathrm{oc}}$ was observed, as shown in Fig. 8(b). The emitter-shifted cell exhibited a substantial decrease in $V_{\mathrm{oc}}$ down to 0.863 V, as compared to the base-shifted and centered cells with promising voltages of 0.957 and 0.945 V, respectively. The reduction in the emitter-shifted cell was contradictory to the previous theoretical simulation results.⁴⁹ A new simulation model was developed by taking the unintentional background n-type doping into consideration, and the reduction of $V_{\mathrm{oc}}$ in the emitter-shifted sample was well reproduced and became consistent with the experimental results.

Fig. 8

(a) Measured EQE versus wavelength. (b) Measured current–voltage characteristics for baseline and QD-enhanced pin-GaAs solar cells. The inset highlights the sub-bandgap photoresponse due to absorption by the QD superlattice. Reprinted with permission from Ref. 69. © 2014 AIP Publishing LLC.

Kechiantz⁷⁰ have reported a novel QDSC structure in which a stack of strain-compensated GaSb/GaAs type-II QDs were embedded in the p-doped emitter region, thus spatially far separated from the depletion region. The original motivation for this theoretical work was to find an ultimate solution to the suppression of the reduction of $V_{\mathrm{oc}}$ in QD while leaving the advantages of QD intact so that it can still act effectively as additional light absorption source to increase $J_{\mathrm{sc}}$, as well as the building block for the realizing IBSC with the conservation of $V_{\mathrm{oc}}$.^71,72 The calculation predicts that the concentration from 1-sun to 500-sun increases the efficiency from 30% to 50%, respectively, without the degradation of the $V_{\mathrm{oc}}$.

2.5.

Other Featured Carrier Dynamics in InAs/GaAs Quantum Dot Solar Cell

2.5.1.

Interdot transportation through Urbach tail

Li et al.⁷³ have proposed an “extended Urbach tail” model, shown in Fig. 9(a), to explain the interdot carrier transport and below-bandgap photon absorption. They have modeled the Urbach tail with two parameters $a_0$ and $E_1$, where $E_1$ is the characteristic width of the absorption edge and $a_0$ is a scaling factor.⁷⁴ As shown in Fig. 9(b), a fit to the linear dependence up to the edge of the QD transition energy estimated the value for $a_0$ and $E_1$ for bulk and QDSC and found 55 meV for QD compared with 15 meV for the bulk GaAs layer.⁷⁴ This indicates that an exponentially increasing continuum density of states occurred in the surrounding GaAs matrix, which facilitates the energy relaxation of excited electrons within the QD conduction band potential. The $V_{\mathrm{oc}}$ measured for the QDSC is 0.77 V while the bulk GaAs control cell is 0.94 V. The deduced conclusion agrees well with the dark $I-V$ curve fitting analysis results, where they have found that the observed change of lower $V_{\mathrm{oc}}$ is due to the change of the SRH recombination coefficient.³⁷

Fig. 9

(a) Schematics of proposed interdot carrier transfer mechanism through continuum density of states due to Urbach tail absorption in the GaAs matrix intrinsic layer and (b) EQE of GaAs/InAs QD sample with no antireflection coating. The Urbach tail contributions for both devices are shown by the dotted lines. Reprinted with permission from Ref. 73. © 2013 AIP Publishing LLC.

2.5.3.

Upper limit of V_oc for nearly defect-free InAs/GaAs quantum dot solar cell

One question has recently been raised regarding the realistic upper limit of $V_{\mathrm{oc}}$ for InAs/GaAs QDSC. ^{30,32,34,37,47} Under the ideal case, the device is assumed to be defect-free; therefore, the SRH or NR recombination will be completely eliminated. As discussed in Sec. 2.1, for a InAs/GaAs QDSC to be able to reach the same $V_{\mathrm{oc}}$ as GaAs SC, the dark saturation current $J_{0,i+\mathrm{QD}}$ must be minimized to the value of $J_{0,i-\mathrm{GaAs}}$.³⁸ Jolley et al.⁷⁷ have addressed this issue by performing both experimental and theoretical studies. As shown in Fig. 10(a), a comparison between the InAs/GaAs QDSC and GaAs SC showed that the QDSC has a smaller activation energy of $\sim 0.1\mathrm{eV}$ than GaAs SC, which is regarded as the leading cause for the reduction of $V_{\mathrm{oc}}$.⁷⁸ Figure 10(b) shows the proposed principle in which the QD layer next to the n-layer will have a raised potential, which results in a reduction of the energy required to transport a hole from the p-layer to the QD layers and, therefore, a reduction in dark current activation energy.^79,80 The crucial point to compensate for this 0.1 V reduction in activation energy is that the optical excitation time must be sufficiently short to reduce the carrier occupation in QDs. Given the fast thermal processes, it is expected that intersubband optical excitation time would have to be on the order of 1 ns or below to have a large impact on the dark current processes; this is, however, hardly achievable.⁸¹ Therefore, 0.1 V difference in $V_{\mathrm{oc}}$ between InAs/GaAs QDSC and GaAs SC can be regarded as the realistic upper limit for the current InAs/GaAs QDSC and is expected to hold even in the complete absence of crystal defects in the device.

Fig. 10

(a) Dark current activation energy as a function of applied bias as calculated from current values for device temperatures within the 280 to 310 K range. Plots of the activation energy plus the applied bias are also shown. (b) Simplified band diagram of the QDSC. The dominant room temperature dark current mechanisms aredepicted. Reprinted with permission from Ref. 77. © 2000 by John Wiley Sons.

It is noteworthy to mention that in very recent research, Varghese et al.⁸² have reported inspiring results by demonstrating a complete voltage recovery in the InAs/GaAs QDSCs by completely suppressing the fast capture of photoelectrons from the GaAs conduction band to the localized states in QDs. The mechanism of this approach reflects exactly the mechanism described in Fig. 10(b), which we proposed for the upper limit of 0.1 V difference.

3.1.

Introduction

Semiconductor QDs have attracted more and more interest recently because of the following unique properties. First, the energy gaps of the QDs and their absorption spectra can be tuned by changing the size of QDs. Second, QDs have large absorption coefficients resulting from the quantum confinement effect. Third, QDs have a potential to generate multiple electron–hole pairs with one single photon absorption [i.e., the multiple exciton generation (MEG)], which would lead to incident-photon-to-current efficiencies of over 100%. Thus, it has been predicted theoretically that the maximum thermodynamic efficiency for photovoltaic devices could be improved to as much as $\sim 44\%$ by employing semiconductor QDs, which is much higher than the Shockley–Queisser limit (33%).^2–5 QDs can be prepared with chemical methods, which are all called CQDs here to make a difference from the QDs prepared with epitaxial growth method, as mentioned in Chapter 2. The CQDs have the advantages of low-temperature fabrication and solution-based processing, which can make the preparation cost very inexpensive.^24,25 Thus, CQDSCs have been expected to be a candidate of cost-effective next generation solar cells. In recent years, two kinds of CQDSCs—QD-sensitized solar cells (QDSSCs) and quantum dot heterojunction solar cells (QDHSCs)—have been developed rapidly.^24,25 Great efforts have been devoted to a fundamental understanding of QDSSCs and QDHSCs; specifically, interface engineering has been carried out systematically, which leads to a significant improvement in the record PCE of these CQDSCs from $<1\%$ to as high as 11.6%.^83–91 The breakthrough in the CQDSCs mostly results from the rapid development of the synthesis technique of high-quality QDs with lower surface trap states and the interface control through various organic and inorganic ligand exchanges on the QDs.^24,25

In this chapter, we will focus on the interface engineering strategies developed in recent years for improving the efficiencies of QDSSCs and QDHSCs and the mechanism for the improvement of the photovoltaic performance.

3.2.

Quantum Dot Sensitized Solar Cells

Figure 11(a) shows the schematic illustration of the working principle and configuration of typical QDSSCs. Similar to dye sensitized solar cell (DSC), QDSSCs consist of a QD sensitized photoelectrode and a counter electrode separated by a liquid electrolyte. In general, wide bandgap metal oxide (${\mathrm{TiO}}_2$, ZnO, and ${\mathrm{SnO}}_2$; in the following ${\mathrm{TiO}}_2$ is used as an example) nanostructures [such as nanoparticulate, nanorods, nanowires (NWs), nanotubes, and inverse opal] are used as the photoelectrodes in QDSSCs. QDs can be deposited on the photoelectrodes by two kinds of in-situ growth methods, i.e., chemical bath deposition (CBD) and successive ionic layer absorption and reaction (SILAR), and by adsorption of presynthesized QDs through direct adsorption or linker-assisted QD deposition method.^25,92–94 The in-situ deposition methods lead to high loading of QDs, but it is difficult to precisely control the QD size distribution, and surface trap state density in the QDs is high. For the presynthesized QD deposition method, the QDs are size-controlled and high-quality with low surface trap states, but normally the QD loading was low (14%) and thus the PCE was as low as $<1\%$.⁹⁵ However, recently, Zhong’s group developed a method to increase the QD loading up to 34%, which leads to the highest record PCE of QDSSCs of 11.6%.⁸⁷ This is comparable to the record PCE of DSC (11.9%).⁹¹ A number of QDs, such as CdS, CdSe, CdTe, PbS, ${\mathrm{Sb}}_2{\mathrm{S}}_3$, ${\mathrm{CuInS}}_2$, CdSeTe, ${\mathrm{CuInSe}}_{1-x}{\mathrm{S}}_x$ and core–shell structure or double layered QDs, such as CdSe/CdTe, ZnTe/CdSe, CdS/CdSe, PbS/CdS, have been applied as sensitizers in QDSSCs.²⁵ Aqueous polysulfide solution is most widely used as the electrolyte for QDSSCs. Some solid-state hole transport materials (HTMs), such as P3HT, Spiro-OMeTAD, and CuSCN, are also used as hole scavenging and transport layers in QDSSCs.^96–98 For the polysulfide electrolyte, copper sulfide (${\mathrm{Cu}}_2\mathrm{S}$) is the most used counter electrode. For the solid-state HTMs, gold and silver are commonly used as electrodes.

Fig. 11

Schematic illustrations of (a) the working principle and configuration of QD-sensitized solar cells and (b) possible charge transfer processes occurring at ${\mathrm{TiO}}_2/\mathrm{QD}/\text{electrolyte}$ interfaces, including injection (Inj), trapping (Trp), and recombination (Rec) of photoexcited carriers.

As shown in Fig. 11(a), once electron–hole pairs are generated in the QD sensitizer after optical absorption, photoexcited electrons are injected into the metal oxide electrode and then transported to collection electrode (transparent conducting electrode), while the oxidized QDs are regenerated by electrolyte (hole scavenging medium) and then the oxidized species of redox couple are regenerated at the counter electrode. Figure 11(b) shows the main possible charge transfer processes occurring at ${\mathrm{TiO}}_2/\mathrm{QD}/\text{electrolyte}$ interfaces, including injection (Inj), trapping (Trp), and recombination (Rec) of both photoexcited electrons and holes.⁸³ The possible injection paths include electron injection to ${\mathrm{TiO}}_2$ from LUMO (Inj1) and from electron trapping levels (Inj2) and hole injection to electrolytes from HOMO (Inj3) and from hole trapping levels (Inj4). The possible recombination paths are internal recombination of photoexcited electrons and holes in QDs directly (Rec1) and through trapping levels (Rec2), injected electrons in ${\mathrm{TiO}}_2$ back transfer to QDs (Rec3, at $\mathrm{QD}/{\mathrm{TiO}}_2$ interface) and to electrolyte (Rec4, at ${\mathrm{TiO}}_2/\text{electrolyte}$ interface), recombination of photoexcited electrons in QDs with oxidized species in electrolyte (Rec5, at QD/electrolyte interface).⁸³ All of the charge transfer processes are key factors for the lower energy conversion efficiency of QDSSCs. In particular, charge recombination at each interface will reduce the charge separation efficiency and charge collection efficiency, which results in poor photovoltaic performance, i.e., low short circuit current $J_{\mathrm{sc}}$, low open circuit voltage $V_{\mathrm{oc}}$, and low fill factor. Here, surface trap states in QDs, which do not exist in molecular dyes of DSCs, are crucial factors for larger recombination and poorer photovoltaic performance in QDSSCs.⁹⁴ In the last few years, many strategies, such as (1) surface/interface passivation of QDs and photoelectrodes and (2) use of core–shell QDs or double layered QDs as sensitizers, have been carried out for suppressing the recombination, reducing trapping and improving the injection of electron and holes in the QDSSCs. In the following, an overview of these strategies will be shown.

3.2.1.

Surface passivation

One important strategy for reducing charge recombination, and thus enhancing charge separation and collection efficiency, is surface passivation on QD-adsorbed photoelectrodes using wide band gap semiconducting materials, such as ZnS, MgO, ${\mathrm{SiO}}_2$ and ${\mathrm{Al}}_2{\mathrm{O}}_3$ through post-treatment.^99–104 ZnS deposition with SILAR method was first applied to PbS/CdS cosensitized QDSC,⁹⁹ and it has attracted much attention since it was proven to double the PCE of CdSe QDSSCs by Toyoda’s group.^100,101 Nowadays, ZnS passivation has become a routine surface passivation method for various QDSSCs, such as CdSe/CdS, and Mn-doped ${\mathrm{CuInS}}_2$, and the PCE of the QDSSCs has been improved largely.^105,106 For example, a PCE as high as 8.1% was reported for CuInSe QDSSCs with ZnS surface passivation.¹⁰⁷ The effects of the ZnS surface passivation and the mechanism for improving the photovoltaic performance of QDSSCs were studied systematically and deeply.^108–110 Hachiya et al.¹⁰⁹ studied the influences of ZnS passivation on charge transfer and recombination dynamics of QDSSCs, such as PbS QDSSCs using transient grating (TG) technique. Their results demonstrate that the ZnS passivation can enhance the charge injection efficiency largely because of the reduction of carrier trapping and recombination in PbS QDs after ZnS surface passivation. The ZnS surface passivation was found to play important roles on the stability and the photovoltaic performance of the QDSSCs, as illustrated in Fig. 12: (1) preventing photocorrosion of QDs in electrolyte and improving the stability; (2) preventing reverse electron transfer from ${\mathrm{TiO}}_2$ to electrolyte and increasing charge collection efficiency; and (3) reducing surface trap states of QDs and thus increasing electron injection efficiency.

Fig. 12

Schematic illustration of the effects of surface passivation on QD adsorbed photoelectrodes in QD sensitized solar cells: (1) prevention of photocorrosion; (2) prevention of reverse electron transfer; and (3) reduction of surface trap states.

ZnSe was used as surface passivation material instead of ZnS for some QDSSCs, such as CdS, CdSe, and ${\mathrm{CuInS}}_2$ QDSSCs.^111–114 Tachan et al.¹⁰² applied MgO to QDSSCs as surface passivation layer.¹⁰² MgO was deposited on ${\mathrm{TiO}}_2$ electrode before and after QD adsorption, and it was found that charge recombination can be reduced significantly because of MgO surface passivation. Surface modification of QDSSCs with ${\mathrm{SiO}}_2$ coating resulted in enhanced photovoltaic performance due to suppression of back electron transfer from the photoelectrode to electrolyte.¹⁰³ ${\mathrm{Al}}_2{\mathrm{O}}_3$ is also a useful material for the passivation of QDSSCs and was demonstrated to decrease charge recombination and enhance electron lifetime in solid-state CdS QDSSCs.¹⁰⁴ As discussed earlier for ZnS coating, these wide bandgap semiconductor coating layers act as energy barriers to prevent the photoexcited electrons in QDs and ${\mathrm{TiO}}_2$ from recombination with oxidized species in electrolyte or solid-state HTMs.

Recently, Zhong’s group made a great development in the surface passivation technique, and the PCE of QDSSCs has been enhanced step by step; now, a certified PCE of over 11% has been achieved.^25,86,87 First, they employed a unique double barrier $\mathrm{ZnS}/{\mathrm{SiO}}_2$ coating on the QD adsorbed photoelectrodes.¹¹⁵ As a result, the PCE of ${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs was improved from 6.37% to 8.55% (certified PCE of 8.21%). They found that the double-layer $\mathrm{ZnS}/{\mathrm{SiO}}_2$ surface passivation could significantly reduce electron dissipation at the outermost surface of photoelectrodes and electron back recombination and largely increase the charge collection efficiency and device stability.¹¹⁵ Then, they further improved the double layer passivation method by inserting an amorphous ${\mathrm{TiO}}_2$ ($\mathrm{a}\text{-}{\mathrm{TiO}}_2$) layer between the photoelectrode and $\mathrm{ZnS}/{\mathrm{SiO}}_2$ passivation layer (Fig. 13), by which method the PCE of ${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs was increased to 9.28% (certified PCE of 9.01%).¹¹⁶ They also investigated a series of metal oxyhydroxide coating on photoelectrode in QDSSCs and found that NbCl and ${\mathrm{ZrO}}_2{\mathrm{Cl}}_2$ modification can improve the photovoltaic properties, especially the open-circuit voltage, and a best PCE of 9.73% was obtained for ${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs.¹¹⁷

Fig. 13

An improved surface passivation on photoelectrode with an amorphous ${\mathrm{TiO}}_2$ $(\mathrm{am}\text{-}{\mathrm{TiO}}_2)/\mathrm{ZnS}/{\mathrm{SiO}}_2$ and the PCE of ${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs was improved to 9.28%. Reprinted with permission from Ref. 116. © 2015, American Chemical Society.

Very recently, an impressive PCE of 11.6% was reported for Zn-Cu-In-Se QDSSCs with the $\mathrm{ZnS}/{\mathrm{SiO}}_2$ surface passivation, which is the best certified efficiency of QDSSCs up until now.⁸⁷ Besides the inorganic material passivation, organic agent passivation, such as dimethylamine, ethylenediamine, ethanedithiol (EDT), thioglycolic acid, and formic acid, were also applied to QDSSCs by Barea et al.¹¹⁸ and de la Fuente et al.¹¹⁹ In addition, double passivation with inorganic ZnS and the above organic agents were employed on QDSSCs, and it was found that the inorganic/organic multilayer surface passivation can suppress the recombination more completely and can improve the PCE of QDSSCs further.

By applying various surface passivation on the QDSSCs as mentioned above, the surface trap states of the QDs were reduced greatly and the reverse electron transfer from the QD-sensitized photoelectrodes to electrolyte was prevented effectively. Thus, the PCEs increased dramatically, as summarized in Table 1. On the other hand, the passivation coating also works as a barrier for the photoexcited holes in the QDs. Therefore, the thickness of the passivation coating is also very crucial and the holes excited in the QDs cannot transfer effectively to hole transfer materials, such as the electrolyte if the thickness of the passivation coating is too large. It has been confirmed that there are optimized thicknesses or coating times for the surface passivation of the QDSSCs.¹⁰⁹

Table 1

Some recent advances in QDSSCs discussed in the paper.

Type of QDSC	Device parameter	PCE	References
CdSe QDSSCs	ZnS passivation	2.1%	100, 101
CdS/CdSe QDSSCs	ZnS passivation	4.92%	105
Mn-doped ${\mathrm{CuInS}}_2$	ZnS passivation	5.38%	106
CuInSe QDSSCs	ZnS passivation	8.1%	107
${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs	$\mathrm{ZnS}/{\mathrm{SiO}}_2$ passivation	8.55%	115
${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs	$\mathrm{a}\text{-}{\mathrm{TiO}}_2/\mathrm{ZnS}/{\mathrm{SiO}}_2$ passivation	9.28%	116
${\mathrm{CdSe}}_x{\mathrm{Te}}_{1-x}$ QDSSCs	NbCl and ${\mathrm{ZrO}}_2{\mathrm{Cl}}_2$ modification	9.73%	117
Zn-Cu-In-Se QDSSCs	$\mathrm{ZnS}/{\mathrm{SiO}}_2$ passivation	11.6%	87
PbS/CdS QDSSCs	ZnS passivation	4.2%	124
PbS/CdS QDSSCs	$\mathrm{ZnS}/{\mathrm{SiO}}_2$ passivation	7.19%	128
CuInS2/ZnS QDSSCs	ZnS passivation	7.04%	129
CdTe/CdSe QDSSCs	ZnS passivation	6.76%	130
ZnTe/CdSe QDSSCs	ZnS passivation	6.82%	131
CdTe/CdS/CdS QDSSCs	CdS passivation	6.32%	132
CdS/CdSe QDSSCs	ZnS passivation	5.32%	134

3.2.2.

Double-layered quantum dots and core/shell structured quantum dots sensitization

Double-layered QDSSCs, such as CdS/CdSe and PbS/CdS, have been demonstrated to show better photovoltaic performances than the single-layer QDSSCs, such as CdSe and PbS QDSSCs.^{105,120–125} CdS/CdSe QDSSCs was first reported by Niitsoo et al.¹²⁰ Then, Lee and Lo¹²¹ clarified that the band position alignment occurred and CdSe band position shifted upward when CdS was deposited between ${\mathrm{TiO}}_2$ and CdSe, which results in smooth electron transfer from the QDs to ${\mathrm{TiO}}_2$. Zhang et al.¹⁰⁵ reported a PCE of 4.9% by optimizing the ${\mathrm{TiO}}_2$ photoelectrode nanostructure. Recently, Osada et al.¹²³ confirmed that the PCE was enhanced by 70% if CdS was adsorbed on ${\mathrm{TiO}}_2$ before CdSe adsorption, but the PCE was lowered by 50% if CdS was adsorbed after CdSe adsorption.¹²³ They studied the mechanism by measuring the charge dynamics using transient absorption and TG methods, and they clarified that the photoexcited electrons could be injected to ${\mathrm{TiO}}_2$ smoothly due to the cascade band edge alignment of $\mathrm{CdSe}/\mathrm{CdS}/{\mathrm{TiO}}_2$. PbS/CdS double-layered QDSSCs together with the control PbS QDSSCs were studied by several groups.^124,125 It was observed clearly that the outer layer of CdS could improve the photocurrent and, thus, the PCE of the QDSSCs largely.¹²⁴ It was clarified that the electron trapping and recombination could be suppressed effectively, which resulted in a significant increase in the electron injection efficiency by the CdS passivation on PbS-sensitized photoelectrode.¹²⁵

On the other hand, presynthesized core/shell structured QDs, where an epitaxial surface layer of a different material is grown around the core QD, have been proven to be useful sensitizers for the inhibition of charge recombination in QDSSCs. The core–shell structured QDs are divided into different types: type I, type II, and inverted type I, according to the relative energy band positions of the core and shell semiconductor materials, as shown in Fig. 14.²⁵ For type I core/shell structured QDs, the shell material has higher CB edge and lower VB edge than those of core material, as shown in Fig. 14(a). The shell reduces the trap state density of the core surface remarkably and works as an energy barrier for photoexcited electrons at the QD/electrolyte interface. Thus, the electron back transfers from the core to electrolyte and from ${\mathrm{TiO}}_2$ to the trap states of QDs, as well as recombination with the holes in the core, can be largely suppressed. However, the shell thickness has to be thin enough ($<1\mathrm{nm}$) to ensure that the photoexcited electrons and holes can be injected from the core to the photoelectrode and electrolyte, respectively, by tunneling through the shell.¹²⁶ Lai et al.¹²⁷ reported direct deposition of PbS/CdS core/shell QDs on mesoporous ${\mathrm{TiO}}_2$ electrode, and a PCE of 1.28% was obtained. Very recently, Jiao et al.¹²⁸ studied the shell thickness dependence of photovoltaic performance of PbS/CdS core/shell QDSSCs. The QDs were deposited on the ${\mathrm{TiO}}_2$ mesoporous electrode using the capping ligand-induced self-assembly method, and $\mathrm{ZnS}/{\mathrm{SiO}}_2$ passivation was carried out. They found that photoexcited carrier lifetimes in the QDs, electron injection efficiency from the core to ${\mathrm{TiO}}_2$, and charge recombination depended greatly on the shell thickness. A maximum PCE as high as 7.19% was achieved when the shell thickness was 0.33 nm.¹²⁸ Another example is type I ${\mathrm{CuInS}}_2$ (CIS)/ZnS core/shell QDSSCs reported by Pan et al.¹²⁹ The ZnS thin layer significantly reduced the surface trap state density of CIS core and charge recombination. As a result, a certified PCE of 7.04% was achieved for an optimized thickness of ZnS shell.

Fig. 14

Schematic illustration of the structures of core/shell QDs: (a) type I, (b) type II, and (c) inverted type I.

In type II core/shell QDs, both conduction and valence bands of the core material are either higher or lower than those of the shell material [Fig. 14(b)]. If the valence and conduction bands of the shell are both lower, the photoexcited electrons are confined in the shell and the holes are confined in the core. Thus, the electrons and holes are spatially separated from each other, which results in an exciplex state absorption (a red-shift of the absorption edge) and longer lifetime of the photoexcited carriers. Up until now, several kinds of type II core/shell QDs, such as ZnSe/CdS, ZnTe/ZnSe, CdTe/CdSe and ZnTe/CdSe, have been applied as sensitizers.^25,130–132 Recently, CdTe/CdSe core/shell QDSSCs were demonstrated to show larger electron injection efficiency and minimized charge recombination rate, resulting in a PCE of 6.76%.¹³⁰ ZnTe/CdSe core/shell QDSSCs showed a faster electron injection rate compared to that of CdTe/CdSe QDs because of the larger conduction band offset, and a certified PCE of 6.82% was achieved.¹³¹ A unique CdTe/CdS (epitaxial)/CdS (nonepitaxial) core/shell/quasi-shell QDSSCs were reported by Sahasrabudhe and Bhattacharyya.¹³² They observed that the epitaxial shell passivates the core surface traps and the nonepitaxial quasishell passivates the ${\mathrm{TiO}}_2$ surface states, as shown in Fig. 15. As a result, a PCE of 6.32% was obtained, which was higher than that of the CdTe/CdS core/shell QDSSCs.

Fig. 15

Schematic illustration of the band diagram of the core/shell/quasishell QDSSCs: (a) CdTe core, (b) CdTe/CdS core/shell, and (c) CdTe/CdS/CdS core/shell/quasishell QDSSCs, respectively. Reprinted with permission from Ref. 132. © 2015, American Chemical Society.

For inverted type-I core/shell structured QDs, the shell material has a narrower band gap compared to the core material and is epitaxially grown around the core [Fig. 14(c)]. Then, a cascade band edge alignment is produced, so photoexcited electrons and holes will transfer to the shell from the core and they will transfer to ${\mathrm{TiO}}_2$ and electrolyte easily. An inverted type I CdS/CdSe core/shell QD-sensitized solar cell was reported and a PCE of 5.32% was obtained.¹³³

Besides the above mentioned type I, type II, and inverted type I core/shell QDs, there is a kind of gradient alloyed QDs in which the element transition is not abrupt in the QDs.¹³⁴ Kim et al.¹³³ reported a simple hot injection method to synthesize the colloidal ${\mathrm{CuInTe}}_{2-x}{\mathrm{Se}}_x$ gradient alloyed QDs with the highest selenium content at the surface of the applications to solar cells. The solution-processed solar cell based on the gradient alloyed ${\mathrm{Cu}}_{0.23}{\mathrm{In}}_{0.36}{\mathrm{Te}}_{0.19}{\mathrm{Se}}_{0.22}$ QDs was prepared, and a PCE of 3.1% was achieved. It was found that the gradient alloyed ${\mathrm{Cu}}_{0.23}{\mathrm{In}}_{0.36}{\mathrm{Te}}_{0.19}{\mathrm{Se}}_{0.22}$ QDs exhibited greatly improved stability over the ${\mathrm{CuIn}}_{1.5}{\mathrm{Te}}_{2.5}$ QDs.

3.3.

Quantum Dot Heterojunction Solar Cells

There are two typical architectures of QDHSCs: planar depleted heterojunction and bulk HQDSCs.^24,135 Figure 16 shows examples of the device structures and the working principles of these two kinds of QDHSCs.¹³⁵ Typical planar depleted HSC is basically composed of a transparent conduction electrode (ITO or FTO), a metal oxide semiconductor thin layer with a thickness of a few 10 nm, an absorbing QD film with a thickness from a few 10 nm to a few 100 nm, and a metal electrode such as Au. The QDs are presynthesized and dispersed in solution with surface passivation of oleic acid (OA) and spin coated on the metal oxide substrate. The QD surface ligand exchange from the original long organic ligand OA to a short organic ligand or inorganic molecule is very crucial for improving the carrier mobility and diffusion length of the QD film and minimizing the possible surface trap states induced during the ligand exahnge.^24,136 Nowadays, the picture of the working principle of the planar QDHSCs is considered as follows [Fig. 16(b)].^24,135 The QDs are separated with each other and, thus, the quantum confinement feature of carriers in QDs is kept. But QD–QD spacing is controlled to be small enough through the ligand exchanging, and, thus, carriers can transfer from one QD to other QDs and move within the QD film through hopping or tunneling. Under working condition of the solar cell, the light is incident from the side of metal oxide semiconductor and absorbed by the QD film. It is believed that a depleted region exists at the heterojunction of metal oxide/QDs and a quasineutral region exists near the metal electrode. The photoexcited electrons and holes in the QDs are drifted by the intrinsic electric field existing in the depleted region and diffuse in the quasineutral region to their collection electrodes, respectively. Subsequently, only photoexcited carriers in the depleted region plus the diffusion length can be effectively collected.^24,135 As a result, the thickness of the QD film, and thus the optical absorption, is limited by the sum of the depletion length and the diffusion length. The depletion width depends on the free-carrier density at each side of the junction, especially that of the QD film. The carrier diffusion length is strongly dependent on surface trapping density and the surface passivation ligands of the QDs. For example, the carrier diffusion length is about 100 nm for PbS QD film, and the typical optimum thickness is about 300 nm for planar depleted heterojunction PbS QDSCs. This thickness is not sufficient for absorbing all incident above bandgap solar radiation. One key issue for this type of QDHSCs is to improve the $J_{\mathrm{sc}}$, which is limited by the insufficient thickness of QD film. By contrast, the $J_{\mathrm{sc}}$ can be enhanced remarkably in the bulk HQDSCs [Fig. 16(c)],¹³⁵ in which nanostructured metal oxides, such as pillars¹³⁷ or NW arrays,^{135,138–143} are applied and CQDs are interpenetrated into these nanostructured metal oxide. Such a bulk junction structure allows for the extension of the depletion region as long as micrometers, which results in a great enhancement in optical absorption, charge separation, and collection efficiency simultaneously.^{135,138–143} $J_{\mathrm{sc}}$ higher than $30\mathrm{mA}/{\mathrm{cm}}^2$ has been obtained by this strategy.^138,139 However, the increase of the interfacial area between the metal oxide nanostructure and the QDs results in an increase in interfacial recombination, so the $V_{\mathrm{oc}}$ is lower compared to that of planar HQDSCs.^138–140 The morphology and the thickness of the nanostructures of the metal oxide have been demonstrated to be key factors for photovoltaic performances of the bulk HQDSCs.^138,139

Fig. 16

(a) Schematic and (b) energy band diagram at short-circuit (cross-section along A-A′) of a planar depleted heterojunction QDHSC, and (c) schematic and (d) energy band diagram (cross-section along B-B′) of a bulk heterojunction QDHSC. Reprinted with permission from Ref. 135. © 2013 by John Wiley Sons.

The planar depleted HQDSCs with n-type ZnO and p-type PbSe QDs were first reported by Leschkies et al.,¹⁴³ and a PCE of 1.6% was obtained. Then, Luther et al.⁸⁸ reported ZnO/PbS QDHSCs with a certified PCE of 3%. The bulk HQDSCs with ZnO NWs and PbSe QDs were also first demonstrated by Leschkies et al.¹⁴⁴ with a PCE of 2%. The main reason for the poor photovoltaic performance of the QDHSCs, especially the low $V_{\mathrm{oc}}$, is the interfacial recombination. During the last few years, the photovoltaic properties of the two kinds of QDHSCs have been improved rapidly due to the advances of the QD preparation method and device engineering.^24,25 In particular, some strategies of interface engineering have been explored to reduce the recombination occurring at (1) the QD surface and/or QD–QD interface;^145–149 (2) the QD/metal electrode (usually Au) interface;^150–152 and (3) the QD/metal oxide (ZnO or ${\mathrm{TiO}}_2$) interface.^139,153,154

For suppressing the recombination (1), QD surface passivation through organic dithiol ligand¹⁴⁵ and inorganic atomic halide ligand¹⁴⁶ exchanging has been proven to be very useful. In addition, a hybrid organic-inorganic ligand approach could effectively decrease the midgap electronic trap state density of QDs, and an increase in $V_{\mathrm{oc}}$ was achieved.¹⁴⁷ Another approach to reduce the midgap trap state density is to passivate the QD surfaces during the synthesis process. Using this method, Zhang et al.¹⁴⁸ achieved high-performance PbSe QDHSCs with record PCE over 6% by using the Cd and Cl passivation. Recently, the same group prepared air stable PbSe QDHSCs (over 50 days) with a certified PCE of 5.9% by passivating the PbSe QD surfaces with halide anions and residual Zn cations through cation-exchange reaction of ZnSe QDs with ${\mathrm{PbX}}_2$ ($\mathrm{X}=\mathrm{Cl}$, Br, or I) precursors.¹⁴⁹

For suppressing the recombination (2) at the QDs/metal electrode interface (i.e., electron transfer from QDs to metal electrode), a strategy of band alignment at the interface was proposed. Brown et al.¹⁵⁰ have demonstrated that the absolute energy level of QD film is critically dependent on surface chemistry and can be modified through ligand exchange. Bawendi et al. introduced an EDT-capped PbS QD layer between an tetrabutylammonium iodide (TBAI)-treated (i.e., iodine passivated) PbS QD layer and the metal electrode to control the band alignment of the QD layers in ZnO/PbS QDHSCs, as shown in Fig. 17. This band alignment engineering led to a certified PCE of 8.55% and the device showed a long-term air stability.¹⁵¹ Very recently, Cao et al.¹⁵² used the same device structure of ZnO/PbS QDHSCs but replaced TBAI with 1-ethyl-3-methylimidazolium iodide for halide-based ligand exchange, and they achieved a certified PCE of 8.7% and a better long-term photostability was demonstrated.

Fig. 17

Schematic illustration of the proposed band bending at short-circuit conditions in the (a) ZnO/PbS–TBAI and (b) ZnO/PbS–TBAI/PbS–EDT (TBAI = tetrabutylammonium iodide). Reprinted with permission from Ref. 151. © 2014 Nature Publishing Group.

For suppressing the recombination (3) occurring at the QD/metal oxide interface, surface passivation of the metal oxide nanostructure is very important, especially for bulk HQDSCs. Chang et al.¹³⁹ have demonstrated that surface trap state density of ZnO NWs can be reduced greatly by coating a thin ${\mathrm{TiO}}_2$ layer on ZnO NWs in ZnO NW/PbS bulk HSCs. Both transient photovoltage decay and impedance spectroscopy characterizations indicated that the interfacial recombination between PbS and ZnO NWs had been significantly suppressed by the surface passivation strategy. As a result, $V_{\mathrm{oc}}$ and PCE have been improved over 40%, and a PCE as high as 6.13% was achieved (device active area: $16{\mathrm{mm}}^2$).¹³⁹ It is worth noting that from the transient photovoltage decay, the authors observed that over 40% of the photoexcited carriers had a much faster recombination, as shown in Fig. 18 (recombination I). The authors found that the slower recombination process II became much slower after the ${\mathrm{TiO}}_2$ surface passivation on ZnO NWs, but the faster recombination process I was not influenced by the surface passivation; further, recombination process I was believed to be recombination occurring at QD–QD interface and/or QD/Au interface. Therefore, appropriate passivation of these interfaces can improve the photovoltaic performance further.¹³⁹ Jang et al. effectively reduced the interband trap sites of the ZnO nanoparticles by using simple EDT treatment, which resulted in a great suppression of the interfacial recombination in ZnO/PbS QDHSCs. As a result, a certified PCE of 10.14 was achieved.¹⁵³ Kim et al.¹⁵⁴ reported a high performance ZnO/PbS QDHSCs with a certified PCE of 10.7% by depositing robust self-assembled monolayers on ZnO surface to adjust the energy alignment of the interface. This is the highest PCE of QDHSCs at the end of 2015. Very recently, the group has achieved a certified PCE of QDHSCs as high as 11.3%.⁹¹ Some recent advances in QDHSCs discussed in the paper is summarized in Table 2.

Fig. 18

(a) Normalized open-circuit photovoltage decay curves for bulk heterojunction ZnO/PbS QDHSCs with and without the ${\mathrm{TiO}}_2$ coating on the ZnO-NWs, showing two dominant decay processes, I and II. (b) The effective carrier lifetime calculated from the voltage decay curves.¹³⁹

Table 2

Some recent advances in QDHSCs discussed in the paper.

To push the PCE of QDSCs to be as high as over 40%, as expected theoretically, fundamental studies on MEG in QDs and its realization in practical QDSC devices are very important and have attracted much interest. The MEG has been observed optically in several kinds of QDs, such as PbS, PbSe, PbTe, CdSe, InAs, Si, carbon nanotube, and ${\mathrm{Cd}}_x{\mathrm{Hg}}_{1-x}\mathrm{Te}$.^155–160 It was found that the generation of multiple-exciton state by MEG in the QDs, such as PbS and PbSe, occurred as fast as a few 100 fs and the relaxation to a single-exciton states on a time scale of a few 10 to a few 100 ps through an Auger like process.^155,159 The extraction of more than one electron per absorbed photon as electrical current in CQDHSCs has been reported with the internal and EQE exceeding 100%, such as PbS, PbSe, and ${\mathrm{CuInSe}}_2$ QDHSCs.^161–163

For improving the efficiencies of both QDSCs and QDHSCs, photoexcited electron and hole transfer from the QDs are essential, especially for designing the devices of MEG-type QDSCs. Dependence of electron transfer from photoexcited QDs to molecular acceptors on the QD sizes has been studied systematically by Zhu et al.¹⁶⁴ They found that the photoexcited electron transfer rate increased as the QD size decreased, i.e., the driving force increased. The unusual driving force dependence was explained by an Auger-assisted electron transfer model in which the electron transfer could be coupled to the excitation of the hole, circumventing the unfavorable Franck–Condon overlap in the Marcus inverted regime.¹⁶⁴ Olshansky¹⁶⁵ investigated the relationship between driving force and rate for hole transfer from photoexcited QDs systematically by using CdSe/CdS core/shell QDs and different molecular hole acceptors. The experimental relationship between the rate and energetic driving force for hole transfer was not well explained by the standard two-state Marcus model because the inverted Marcus region was not observed. An Auger-assisted photoexcited hole transfer mechanism was proposed in which experimental results were modeled successfully.¹⁶⁵ These relationships between interfacial charge transfer and energetic driving force are very useful for designing CQDSCs to maximize interfacial charge transfer and minimize energetic losses associated with the driving force.