1. Introduction

Semiconductor nanomaterials based on colloidal quantum dots (CQDs) or nanocrystals (NCs) have attracted much attention in the fields of optoelectronics [1–9], especially for solar cells [10–12], because of size-dependent optical bandgap (E_g) [11,13], high absorption coefficient [10], tunable carrier mobility and relatively easy fabrication [14]. Thanks to the efforts on surface chemical modification of CQDs [15,16], doping [12,17], interfacial modulation of carrier transport layers [18], and other device structure optimizations [19], the power conversion efficiency (PCE) of CQDs solar cells has increased to 13.3% in recent [20]. Among those proposed strategies, the main idea is the enhancement of carrier diffusion length in CQDs solar cells, so that the separated charge carriers can be efficiently extracted by electrodes [13].

In order to achieve the same goal, organic hosts such as polymer have been mixed with CQDs or NCs to form hybrid solar cells, in which the polymer network is thought to be helpful for carrier extraction. The polymer/NC hybrid solar cells have achieved PCEs exceeding 11% to date [21]. New challenges are introduced at the organic/inorganic interfaces, which would decrease the conductivity in polymer/NC composites [22]. During the development of polymer/NC solar cells, core-shell nanocrystals prepared by aqueous method have solved this problem to some extent [23,24]. It is found that aqueous-processed CdTe NCs would be partly coated by CdS shells after the heat annealing treatment, which results in an efficient self-charge separation occurred in such a giant NC size (∼20 nm) [25]. This shows that even without the assistance of polymer networks, those bulk heterostructures based on nanocrystal solids with a core-shell structure also have great potential for high-performance solar cells [26,27]. For example, Yang et al. have constructed CdTe-CdS bulk heterojunctions by annealing CdTe NCs, and approached an extremely high internal quantum efficiency (IQE) of ∼100% (especially, in the range of 350-550 nm, the IQE values even exceed 100%) [27]. However, underlying photophysical mechanisms for aqueous-processed giant CdTe-CdS NC solids, which are beyond the exciton Bohr radius, are rarely reported. For example, the carrier multiplication (CM) effect or multiple-exciton generation, which could provide an attractive possibility to bypass the “phonon bottleneck” in CQD-based solar cell systems [28–32], is not fully explored in aqueous-processed CdTe/CdS NC solids.

Here, we investigate the CM effect in aqueous-processed giant CdTe-CdS NC solids where the diameter of CdTe core is ∼20 nm [25,27], larger than its bulk exciton Bohr radius (about 7.5 nm). In contrast to the multiple exciton behaviors in the reference films consisting of large CdTe quantum dots (the diameter is estimated to be ∼11 nm) synthesized by an oil-phase approach [33], there is an efficient carrier multiplication by Auger processes in the aqueous-processed giant CdTe-CdS NC solids. In addition, the efficient carrier extractions are further demonstrated when using TiO₂ and MoO₃ as the carrier transport layers for aqueous-processed giant CdTe-CdS NC depleted bulk heterostructures. Our findings reveal the new photophysical mechanisms for the high-efficient all-solid nanocrystal photovoltaics based on giant CdTe-CdS NC solids synthesized by aqueous-solution approaches.

2. Experimental section

2.1 Preparation of samples

Aqueous-processed giant CdTe-CdS NC solids on quartz are generated after the annealed treatment at 300°C in a glove box for spin-coated films consisting of mercaptoethylamine (MA)-capped CdTe QDs [27]. For aqueous-processed giant CdTe-CdS NC depleted bulk heterostructures, the ITO/TiO₂ substrates are used as electron transport layer, and then the active layer of MA-capped CdTe QDs is spin-coated on ITO/TiO₂ substrates. After the annealed treatment at 300°C in a glove box, the hole transport layer (MoO₃ film) is evaporated on top of the grown CdTe-CdS NC solids. Oil-phase CdTe QDs are synthesized by the typical hot-injection method with modifications, and then undergo a ligand exchange treatment [33].

2.2 Femtosecond transient absorption setup

In the femtosecond time-resolved transient absorption (TA) system, we used a mode-locked Ti: sapphire laser/amplifier system (Solstice, Spectra-Physics) to generate 800 nm laser pulses with a width of 100 fs and a repetition rate of 250 Hz. It was divided into two parts: the stronger one was used to pump the samples; the other one passed through a cuvette of 2-mm-thick water, which produced a broadband probe light. The TA data were gathered by a fiber-coupled spectrometer. The dispersion of TA data was corrected by a chirp program. All measurements were performed at room temperature.

3. Results and discussion

3.1 TA experiments of aqueous-processed giant CdTe-CdS NC solids

Both the steady-state absorption spectrum and the broadband TA spectra of CdTe-CdS bulk heterojunction solids (Fig. 1) confirm its E_g at 1.512 eV (820 nm). On account of the large size, this E_g of giant CdTe-CdS NC solids is close to the bulk CdTe of 1.475 eV [34]. So, the pump pulses at 800 nm correspond to the band-edge excitation, and the pump pulses at 400 nm (3.1 eV, larger than 2E_g) could induce the CM effect. The initial exciton population in our TA experiments is estimated by the equation < N(0)> = j_pσ₀ [35], where j_p is the pump fluence, and σ₀ is the absorption cross section obtained from the literature [36].

 

Fig. 1. (a) Steady-state absorption spectrum of CdTe-CdS NC solids (inset is the TEM image of aqueous-processed CdTe-CdS NC solids; scale bar is 20 nm). (b) Typical TA spectra under 800 nm excitation (<N(0)> = 1.12) and 400 nm excitation (<N(0)> = 3.34) for CdTe-CdS NC solids (probed at 3 ps, and normalized at the ground state bleaching peak of CdS state). The vertical dash lines indicate the positions of E_g (820 nm) and 2E_g (410 nm) of CdTe core in CdTe-CdS NC solids.

Download Full Size | PDF

According to the typical TA spectra of CdTe-CdS NC solids (Fig. 1(b)), the high-energy state around 510 nm is assigned to the ground state bleaching signal of CdS shells [25,27]. Under 800 nm excitation, only the CdTe core is populated, but the self-electron separation is expected to occur in the conduction bands [37], leading to the CdS^- state. When it is pumped by the 400 nm light, both the CdTe core and CdS shell are directly excited, inducing the CdS_1S state (contributed by the sum of electrons and holes). Assuming the oscillation strength for the CdS state would be one third of that of CdTe state due to the potential type-II band alignment for CdTe-CdS heterostructures and the low fraction of coverage for CdS shells in CdTe-CdS NC solids, the band offset between the conduction band edge of CdS and CdTe could be estimated from -13 meV to -38 meV (Supporting Information Note 1). So, the electrons in the conduction band are delocalized readily in the type-II CdTe-CdS NC solids. In contrast to layered CdS/CdTe thin films, which are type-II structures with large band offsets like bulks at the interface [38,39], the small band offset for CdTe-CdS NC solids could be modified by the interface stains [40].

3.2 Global analysis for TA data

In order to shed lights on the initial carrier population in CdTe-CdS NC solids which could involve with the self-electron separation and CM processes, we use the global fitting to treat the TA data within the first 8 ps [41]. It shows that under the resonance excitation of CdTe core at 800 nm (<N(0)> = 1.12), representing a relatively simple case, there are two lifetime components (the fast one is 2.40 ps and the slow one is 23.0 ps) for CdTe-CdS NC solids (Fig. 2(a)). In the lifetime component curve resulted from the global fitting approach, the positive part presents the arising processes, and the negative part corresponds to the decay processes, which share the same lifetime at different probe wavelengths. This global fitting approach could reveal the possible relation between different electronic states. Thus, under 800 nm excitation (the state-fill effect could be ignored for CdTe core in the case of band-edge excitation), the signals at CdTe_1S state and CdS^- state are generated synchronously in CdTe-CdS NC solids. The positive signals around 820 nm in 2.40-ps lifetime component even imply the CM phenomenon occurs due to the Auger processes for the CdTe_1S state under 800 nm excitation with a relatively high pump fluence (<N(0)> = 1.12), which disappears in the low pump fluence cases (Fig. S1). The absolute amplitude of the negative part around 730 nm is comparable with that of the positive part around 510 nm in the 2.40-ps lifetime component, implying it is concerned with the formation of CdS^- state. Namely, the self-electron transfer from CdTe core to CdS shell in CdTe-CdS NC solids could involve with the photogenerated electrons at the energy levels which are higher ∼0.2 eV (the energy difference between ∼730 nm and ∼820 nm) than the conduction band edge of CdTe core. Besides, the negative parts in the 23.0-ps lifetime component reflect that the decay of CdTe_1S state and CdS^- state are relatively slow under 800 nm excitation.

 

Fig. 2. Global fitting for the TA data within the first 8 ps in aqueous-processed giant CdTe-CdS NC solids under (a) 800 nm excitation (<N(0)> = 1.12) and (b) 400 nm excitation (<N(0)> = 3.34). A comparison between the time-resolved dynamics of (c) CdTe state and (d) CdS state in aqueous-processed giant CdTe-CdS NC solids under 800 nm excitation and 400 nm excitations. (e) Global fitting for the TA data within the first 8 ps in large oil-phase CdTe QD films under 400 nm excitation and (f) corresponding exciton dynamics (<N(0)> = 0.89). The red solid lines are the fitted curved by global analysis of TA data.

Download Full Size | PDF

Under 400 nm excitation for CdTe-CdS NC solids (Fig. 2(b)), it becomes three lifetime components (0.99 ps, 1.85 ps and 3.00 ps). The lifetime components of 0.99 ps and 3.00 ps show the entire shapes similar to that of 23.0-ps lifetime components under 800 nm excitation, except the 3.00-ps lifetime component exhibits larger signal amplitudes in the probe window. This indicates faster decay trends for both the CdTe_1S and CdS_1S states, probably due to the Auger processes related to the CM mechanism in CdTe core and the hole transfer in CdS shell, compared with that under 800 nm excitation. So, the additional 1.85-ps lifetime component becomes very interesting. The negative part ranging of 400-480 nm in 1.85-ps lifetime component could correspond to the excited states of CdS_1S state under 400 nm excitation. The absolute amplitude of the negative part around 730 nm is ∼2.5 times larger than that of the positive part around 510 nm in the 1.85-ps lifetime component. This implies that the carrier population associated with the electron transfer from the conduction band of CdTe core to CdS shell could increase ∼2.5 times. The large positive signals around 820 nm in 1.85-ps lifetime component indicate an average arising lifetime for the CdTe_1S state under 400 nm excitation, since this lifetime is too long to be considered as the pure state-filling time (the hot electrons generated by absorbing the pump photon energy of 3.1 eV will relax to the band-edge state, and it should take hundreds of femtoseconds, as we observe in the control experiments for large oil-phase CdTe QD films). So, it could be an average effect of multiple transient events, including the state-filling processes, CM processes, and hole transfer processes from CdS to CdTe.

The typical exciton dynamics for the CdTe_1S and CdS states in CdTe-CdS NC solids further confirm the above analysis. Especially for the CdTe_1S state (Fig. 2(c)), it shows an obviously slower arising under the high-energy excitation within the first 3 ps, in comparison with that under 800 nm excitation. For the CdS^- and CdS_1S states (Fig. 2(d)), both decay traces are almost the same within the first 2 ps, and then they present different relaxation behaviors. It implies that the hole transfer processes from CdS to CdTe could be relatively slower than the possible CM processes occurred in the CdTe core. Thus, the dominant event for the large positive signals around 820 nm in the 1.85-ps lifetime component could correspond to the underlying CM processes.

We further used the large oil-phase CdTe QD films to perform the control experiments. The TA spectra and pump-density-dependent exciton dynamics of oil-phase CdTe QD films under 400 nm excitation are presented in Fig. S2. According to its E_g at 1.621 eV (765 nm), we estimate that the diameter of oil-phase CdTe QDs is ∼11 nm [36]. The global fitting results for the oil-phase CdTe QD films at the low pump condition of 400 nm excitation present two lifetime components of 0.49 ps and 2.03 ps, where no CdS signals are identified (Fig. 2(e)). The positive part around 765 nm indicates the reasonable state-filling time is 0.49 ps. Time-resolved dynamics for oil-phase CdTe QD films probed at its high-energy electronic state at 496 nm decay faster than its CdTe_1S state (765 nm) due to the state-filling processes, and both disappear in the first 6 ps because of the efficient exciton-exciton annihilation and defect trapping (Fig. 2(f)). Those not only reflect that the traditional exciton recombination behaviors in oil-phase CdTe QD films are different from the observed phenomena in aqueous-processed CdTe-CdS NC solids, but also highlight the crucial role of CdS shells on the change of exciton relaxation models in giant CdTe-CdS NC solids, which could achieve the CM effect by Auger processes.

3.3 Population dependence TA behaviors for the CM mechanism

Considering the small oscillation strength for CdS and ground state bleaching signal in TA experiments, the carrier distribution between the CdTe_1S and CdS_1S states under 400 nm excitation is further corrected (Supporting Information Note 2). The corrected exciton population distributions are presented in Table S1. It shows the initial exciton population is equally distributed between the CdTe core and CdS shell at the relatively low pump intensity. Then, the different evolution trends for the initial exciton population of CdTe_1S and CdS_1S states in CdTe-CdS NC solids under high-energy excitation are plotted in Fig. 3(a). The orange line for CdTe_1S state is fitted by an empirical equation, -ΔO.D./α₀ = k₁< N(0)>/(k₂ + <N(0)>), where α₀ is steady-state absorption value in Fig. 1(a) for the given electronic state, k₁ = 0.32 and k₂ = 3.10. The blue line for CdS_1S state is fitted by another empirical equation, -ΔO.D./α₀ = k₃< N(0)>/(k₄ + <N(0)>), where k₃ = 0.83 and k₄ = 13.1. The parameters, k, could be category-dependent, and affected by the environment of samples, such as which are in solution or dried as solids. According to the literature 35, those expressions are used to estimate the average population dynamics < N(t)> for the CdTe_1S and CdS_1S states in aqueous-processed giant CdTe-CdS NC solids. The average population dynamics for the CdTe_1S and CdS_1S states in aqueous-processed giant CdTe-CdS NC solids under 400 nm excitation are presented in Fig. 3(b). For both the CdTe_1S and CdS_1S states, as the corrected < N(0)> increases up to ∼4, the slow arising process is still observed in the first 3 ps of average population dynamics, which could be connected with the CM processes as we discussed in Fig. 2. After the first 10 ps of average population dynamics, the sign of bleaching signals is inverted, which are stronger as the corrected < N(0)> increases. This implies the formation of long-lifetime free charges. Those demonstrate that the result of CM processes in CdTe-CdS NC solids is to generate free carriers with long lifetimes. On the other hand, when the pump fluence further increases, the initial exciton population of CdS_1S state becomes dominant, accompanied with larger band offset in the type-II structure.

 

Fig. 3. (a) Population dependence of the normalized ground-state bleaching signals for CdTe_1S and CdS_1S states in CdTe-CdS NC solids under 400 nm excitation after the correction of population distribution. (b) Average population dynamics for the CdTe_1S state in aqueous-processed giant CdTe-CdS NC solids under 400 nm excitation with different pump fluences. Inset is the average population dynamics for the CdS_1S state.

Download Full Size | PDF

Now, we present the physical pictures of aqueous-processed giant CdTe-CdS NC solids and large oil-soluble CdTe QD films. The giant CdTe-CdS NC solids have a type-II structure with small band offsets of a few tens of meV, and the delocalization of electrons in the conduction band is much easier than that in oil-soluble CdTe QD films, even at the band-edge excitation of 800 nm (red arrow in Fig. 4, left). Under 400 nm excitation (blue arrow), oil-soluble CdTe QD films show efficient exciton-exciton annihilation processes and defect trapping, after the hot electron relaxation (Fig. 4, right) [42–44]. While for CdTe-CdS NC solids, the inherent Auger processes (represented by the two grey dash arrows and the orange double-headed arrow) would lead to the CM formation in a relatively large pump fluence condition (Fig. 4, middle), thanks to the small band offset and the strong delocalization of electrons in the conduction bands of CdS shell and CdTe core. In addition, the hole transfer from CdS shell to CdTe core in CdTe-CdS NC solids under 400 nm excitation is relatively slow, which ensures the sustainable occurrence of CM processes in CdTe core.

 

Fig. 4. Schematic illustration of the CM formation in aqueous-processed CdTe-CdS NC solids under (left) 800 nm and (middle) 400 nm excitation conditions, compared with (right) the exciton relaxation in large oil-soluble CdTe QD films.

Download Full Size | PDF

3.4 Carrier extraction in aqueous-processed giant CdTe-CdS NC depleted bulk heterostructures

Finally, the efficient carrier extractions are demonstrated by using TiO₂ and MoO₃ as the carrier transport layers in the fabricated CdTe-CdS NC heterojunctions [25,27]. The global fitting results indicate that for TiO₂/CdTe-CdS NC heterojunctions where TiO₂ is the electron transport layer, the fast lifetime component of 0.76 ps under 800 nm excitation is smaller than that of 2.40 ps in CdTe-CdS NC solids, but the slow lifetime component of 43.0 ps becomes much larger (Fig. 5(a)). It implies the additional electron transfer to TiO₂ electrode mainly affects the fast lifetime component, which suppresses the recombination between electrons and holes. For MoO₃/CdTe-CdS NC heterojunctions where MoO₃ is the hole transport layer, the global fitting results present that the fast lifetime component of 2.87 ps under 800 nm excitation is similar to that in CdTe-CdS NC solids, but the slow lifetime component of 10.9 ps becomes smaller than that of 23.0 ps in CdTe-CdS NC solids (Fig. 5(b)). It shows the additional hole transfer to MoO₃ electrode mainly affects the slow lifetime component. Thus, the band-edge carrier extraction by the transport layers of TiO₂ and MoO₃ is imbalanced, where the electron extraction efficiency (η_e) by the TiO₂ layer is estimated at ∼68%, and the hole extraction efficiency (η_h) by the MoO₃ layer is estimated at ∼53%, according to the carrier extraction equations, ${\eta _e} = 1\textrm{ - }\tau {(0.76ps)_{\textrm{Ti}{\textrm{O}_\textrm{2}}\textrm{/CdTe - CdS}}}\textrm{/}\tau (2.40ps){}_{\textrm{CdTe - CdS}}$ and ${\eta _h} = 1\textrm{ - }\tau {(10.9ps)_{\textrm{Mo}{\textrm{O}_\textrm{3}}\textrm{/CdTe - CdS}}}\textrm{/}\tau (23.0ps){}_{\textrm{CdTe - CdS}}$. As a result, for TiO₂/CdTe-CdS/MoO₃ NC depleted bulk heterostructures under 800 nm excitation (Fig. 5(c)), the global fitting results show that both the fast and slow lifetime components are larger than those in CdTe-CdS NC solids, due to the largest space separation for both the electrons and holes. In addition, for TiO₂/CdTe-CdS/MoO₃ NC depleted bulk heterostructures under 400 nm excitation (Fig. 5(d)), the global fitting results show that the first two lifetime components of 0.67 ps and 1.75 ps are smaller than those of 0.99 ps and 1.85 ps in pristine CdTe-CdS NC solids, and the lifetime component of 10.4 ps is much larger than that of 3.00 ps in CdTe-CdS NC solids. It implies the efficient hot-carrier extractions in TiO₂/CdTe-CdS/MoO₃ NC depleted bulk heterostructures, due to the slower hot-carrier cooling and band-edge exciton recombination. We could estimate a hot-carrier extraction efficiency (η_hot) of ∼32% by the fastest lifetime components in Fig. 5(d) and Fig. 2(b) (${\eta _{\textrm{hot}}} = 1\textrm{ - }\tau {(0.\textrm{67}ps)_{\textrm{Ti}{\textrm{O}_\textrm{2}}\textrm{/CdTe - CdS/Mo}{\textrm{O}_\textrm{3}}}}\textrm{/}\tau (\textrm{0}\textrm{.99}ps){}_{\textrm{CdTe - CdS}}$). It is consistent with the enhanced IQE value from ∼80% at 800 nm to that of ∼110% at 450 nm in TiO₂/CdTe-CdS/MoO₃ solar cells [27].

 

Fig. 5. Global fitting for the TA data within the first 8 ps in TiO₂/CdTe-CdS NC heterojunctions under (a) 800 nm excitation (<N(0)> = 1.12), where the inset is the schematic of band edge alignment for TiO₂/CdTe-CdS NC heterojunctions. Global fitting for the TA data within the first 8 ps in MoO₃/CdTe-CdS NC heterojunctions under (b) 800 nm excitation (<N(0)> = 1.12), where the inset is the schematic of band edge alignment for MoO₃/CdTe-CdS NC heterojunctions. Global fitting for the TA data within the first 8 ps in TiO₂/CdTe-CdS/MoO₃ NC depleted bulk heterostructures under (c) 800 nm excitation (<N(0)> = 1.12) and (d) 400 nm excitations (<N(0)> = 3.34).

Download Full Size | PDF

4. Conclusion

In summary, we experimentally depict the efficient CM mechanisms through a series of Auger processes in aqueous-processed giant CdTe-CdS NC solids. Its type-II structure with small band offsets and strong delocalization of electrons has been revealed. In addition, the phenomenon of hot-carrier extractions is confirmed in TiO₂/CdTe-CdS/MoO₃ NC depleted bulk heterostructures. Those could explain the reported high IQE values, especially in the high-energy region (350-550 nm) for the all-solid CdTe-CdS NC solar cells. The small band offset in the conduction bands of aqueous-processed giant CdTe-CdS NC solids is crucial for the CM processes. On the other hand, we find that the commonly used transport layers like TiO₂ and MoO₃ can extract charges for aqueous-processed giant CdTe-CdS NC solids, but the carrier extraction efficiencies are imbalanced. The hole trapping states located between the valence band minimum of CdTe and CdS [37] could be more close to the valence band minimum of CdTe core in CdTe-CdS NC solids. This could be one of reasons responsible for the lower hole transfer efficiency in MoO₃/CdTe-CdS NC heterojunctions. Furthermore, the initial carrier extraction efficiencies at the band edge still have great room to upgrade in future, which is one of the main reasons that limit the current performance of aqueous-processed NC solar cells. Hence, searching for new carrier extraction materials and reducing the interface traps between CdTe and CdS could be more important for promoting the device performance of all-solid nanocrystal photovoltaics based on aqueous-processed giant CdTe-CdS nanocrystal solids.