full papers www.MaterialsViews.com

Electron Injection

A Layer-by-Layer ZnO Nanoparticle–PbS Quantum Dot Self-Assembly Platform for Ultrafast Interfacial Electron Injection Mohamed Eita, Anwar Usman, Ala’a O. El-Ballouli, Erkki Alarousu, Osman M. Bakr, and Omar F. Mohammed*

Absorbent layers of semiconductor quantum dots (QDs) are now used as material platforms for low-cost, high-performance solar cells. The semiconductor metal oxide nanoparticles as an acceptor layer have become an integral part of the next generation solar cell. To achieve sufficient electron transfer and subsequently high conversion efficiency in these solar cells, however, energy-level alignment and interfacial contact between the donor and the acceptor units are needed. Here, the layer-by-layer (LbL) technique is used to assemble ZnO nanoparticles (NPs), providing adequate PbS QD uptake to achieve greater interfacial contact compared with traditional sputtering methods. Electron injection at the PbS QD and ZnO NP interface is investigated using broadband transient absorption spectroscopy with 120 femtosecond temporal resolution. The results indicate that electron injection from photoexcited PbS QDs to ZnO NPs occurs on a time scale of a few hundred femtoseconds. This observation is supported by the interfacial electronic-energy alignment between the donor and acceptor moieties. Finally, due to the combination of large interfacial contact and ultrafast electron injection, this proposed platform of assembled thin films holds promise for a variety of solar cell architectures and other settings that principally rely on interfacial contact, such as photocatalysis.

1. Introduction Efficient, renewable, and economically feasible energy is needed to reduce greenhouse gas emissions.[1] This need may be at least partially met by the development of thinfilm solar cell technologies.[2–5] Irrespective of the architecture of a solar cell, such as dye-sensitized,[6–10] polymer,[11–14] hybrid,[15–19] or thin-film pn-junctions,[3] the essential step

Dr. M. Eita, Dr. A. Usman, A. O. El-Ballouli, Dr. E. Alarousu, Prof. O. M. Bakr, Prof. O. F. Mohammed Solar and Photovoltaics Engineering Research Center Division of Physical Sciences and Engineering King Abdullah University of Science and Technology Thuwal 23955–6900, Kingdom of Saudi Arabia E-mail: [email protected] DOI: 10.1002/smll.201400939

112

wileyonlinelibrary.com

in photocurrent generation is electron transfer, which takes place across the donor-acceptor interface or the electron transport layer (ETL).[20] The major challenge in improving solar cell efficiency is to maximize the interfacial contact between the donor and the acceptor moieties. This may be achieved by increasing the porosity of the acceptor layer not only to accommodate more absorber material, but also to make large interfacial contacts with the acceptor layer. Various methods have been used to prepare heterojunction solar cells, including physical deposition using high vacuum techniques such as sputtering or thermal evaporation, solution-based processes such as spin-coating, and chemical deposition by direct synthesis on device substrates.[3,16,21] Among them, solution processing methods, which can provide various surface structures, are attracting much attention.[15,21] However, none of these methods has resulted in a designable and controllable porosity that can significantly increase the interfacial contact between the

© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

small 2015, 11, No. 1, 112–118

www.MaterialsViews.com

donor and the acceptor units and thus enhance solar cell efficiency. The optimal interfacial contact achieved so far is a nanowire architecture of the acceptor layer.[22–24] In spite of the increased efficiency of this architecture, the surface area of the acceptor itself is not large enough and most of the donors surrounding the nanowires do not make sufficient interfacial contact with the acceptor. Moreover, the thickness of this particular architecture is on the micrometer scale, in contrast to the general need of making thin solar cells that reduce the cost and the processing time. Recently, thin layers of ZnO nanoparticles (NPs) that were prepared by spin-coating without sintering exhibited high efficiency[25] due to the enhancement of the surface area.[26] In addition, ZnO substrates with large band gaps (3.4 eV) are attractive for the use with semiconductor quantum dots (QDs),[27] because they allow selective excitation of the QDs without interference from the substrate. Here, we specifically introduce a layer-by-layer (LbL) technique[28–30] as a promising methodology to increase the interfacial contact between PbS QDs and ZnO NPs for sufficient electron injection across the interface. We confirmed the increased interfacial contact by both atomic force microscopy (AFM) and cross-sectional scanning electron microscopy (SEM) experiments, which demonstrated clearly how the pores in the thin layers of ZnO NPs were filled with PbS QDs, providing unprecedented interfacial contact for electron injection processes. Advantageously, the ZnO NPs can be easily deposited as thin films from aqueous solutions and the porous structures of the layers can be controlled by adjusting their thickness and the size of the pores to be filled by the colloidal PbS QD electron donors. We utilized transient absorption (TA) spectroscopy[31,32] with broadband capability and 120 fs temporal resolution to explore the electron injection process across the PbS QD–ZnO NP interface. The transient absorption data indicated that electron injection from photoexcited PbS QDs to ZnO NPs occurs on a time scale of a few hundred fs.

Figure 1. (A) Near infrared absorption spectra of PbS QDs on top of ZnO thin films with different numbers of immersion cycle showing the absorption peak of the PbS QDs. The absorption intensity increases with the number of bilayers; (B) UV-Vis absorption spectra of the corresponding ZnO thin layers showing that absorbance increases with the number of bilayers; and (C) thickness of the ZnO thin layers on ITOcoated glass substrates as measured by spectroscopic ellipsometry.

2. Results and Discussion 2.1. Optical Characterization of ZnO NPs and ZnO NP–PbS QD Thin Films By adopting a simple LbL approach, we prepared thin layers of ZnO NPs. As shown in Figure 1A,B, the absorption spectrum of the thin layers of ZnO NPs is in the UV region with a broad shoulder in the visible range, the absorption intensity obviously increased with the number of immersion cycles. Loading PbS QDs onto the ZnO NP thin layers gives an additional band in the near infrared region, which we attributed to the lowest optical transition between the spatially confined electron and hole states of QDs.[33] The cooperative absorption of the thin layers of ZnO NPs and PbS QDs therefore covers a significant portion of the solar spectrum from UV to the near infrared. The absorption intensity of PbS QDs increases with the thickness of the layer of ZnO NPs, implying that the number of PbS QDs incorporated into the layer increases as the thickness, surface roughness, small 2015, 11, No. 1, 112–118

and pore size increase. It is worth mentioning that the alignment of energy levels of PbS QDs and ZnO NPs energetically allows electrons to transfer from the excited PbS QDs to ZnO NPs.[34–36] This would mean that it is reasonable to consider PbS QDs and ZnO NPs as a promising pairing for heterojuction solar cells, similar to TiO2–PbS QD heterojunction cells.[34] The thickness of the ZnO NP layers as a function of the number of immersion cycles is shown in Figure 1C. The thickness, refractive index (which is obtained from the Cauchy model),[37] and the surface roughness of the ZnO NP layers are summarized in Table 1. The thickness of the layers is completely controlled by the number of immersion cycles at an increment of 5.5 nm per immersion cycle. The high porosity and anti-reflective properties of the ZnO/PAA (polyacrylic acid) thin films cause their refractive indices to be much lower than those of bulk ZnO (refractive index = 2.0 at 610 nm),[38] glass substrates, or ITO layers. The correlation of the increase in the surface roughness and the decrease

© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

www.small-journal.com

113

full papers www.MaterialsViews.com Table 1. Thicknesses, refractive indices, and the surfaces roughness of ZnO NP films with different numbers of immersion cycles (bilayers). Bilayer Number

Thickness [nm]

Refractive index

Surface Roughness [nm]

6

40

1.66

16.5

10

61

1.51

18.8

20

112

1.4

25.5

30

161

1.35

31.3

Sputtered

103

1.98

2.8

in the refractive index with the thickness of the ZnO layer suggests that the porous structure is created in a stepwise manner with the thickness of the layer. AFM images and the surface roughness of the ZnO NP layer before and after incorporation of PbS QDs (Figure 2) clearly show the nature of the QDs in the thin layer. As expected, the LbL approach creates pores spanning the entire area of the ZnO thin layer, and the QDs fill in the pore spaces and flatten the surface of the ZnO thin layer. For instance, incorporating PbS QDs reduced the surface roughness of a 61-nm-thick ZnO thin layer from 18.8 nm to 1.5 nm. To gain further insight into the structure of the ZnO NP layer-PbS QD heterojuction, we captured SEM images (Figure 3) of the film’s cross-section. The images clearly show that the ZnO NP-PbS QD heterojunction prepared using the LbL method consists of the ZnO NPs sticking to each other to create a hilly shape, leaving pores of a few hundreds of nm in diameter and a rough surface, with the QDs filling in the pores and creating sheet above the ZnO thin layer that is less than 40 nm thick. This suggests the PbS QDs are adsorbed into the pores rather than forming a separate layer on top of the ZnO layer. Thus, interfacial contact between the ZnO NP and PbS QD layers is curved to follow the hilly surface of the ZnO NP layer and donor-acceptor herojunctions are created without any large-scale defects that could shorten the lifetime of the device. In comparison, cross-sectional SEM images of ZnO NP-PbS QD heterojunctions prepared by sputtering method show a 100-nm-thick ZnO layer with a very smooth surface with the roughness measured as 2.8 nm (Table 1) and PbS QDs deposited using dip-coating to create a 300-nm-thick layer on the other layer, forming a two-layer structure (Figure 3C). Both surfaces of the ZnO NPs layer and the PbS QD layer exhibit rigid and tightly packed structures, producing a planar interfacial contact. In addition, line profiles (Figure 2C) of ZnO layer surfaces indicate that the interfacial contact area of the heterojunction prepared by the LbL approach is larger than that of the interfacial contact of the heterojunction prepared by sputtering. Since the thickness of the ZnO layer plays a role in the final efficiency of solar cells,[25] the controllable and designable LbL films will be useful for such applications. The LbL method could be applied to a variety of semiconductor materials by choosing the optimum condition for a stable dispersion, and by adopting the polymer that together with the semiconductor could form a porous structure. A crucial prerequisite of the success of the LbL method in building a porous structure is to keep the adsorption of

114 www.small-journal.com

Figure 2. (a) An AFM height image of a ZnO NPs/PAA bilayer after 10 immersion cycles showing the porous structure of the film; (b) an AFM height image of the corresponding ZnO NP thin layer coated with PbS QDs and (c) the section profile analysis showing the smooth surface of the ZnO NP thin layer after coating with PbS.

both components the lowest possible by adjusting the pH, molecular weight and the concentration of the polyelectrolyte. A higher adsorption of the polyelectrolyte will lead to increasing the adsorption of the semiconductor and to thicker films with very limited porous structure. PAA and ZnO used in this report were left at their native pH of 8.3 and 6.7, respectively, without adjustment. PAA, as an anionic polyelectrolyte, was reported to have the lowest adsorption and to form the thinnest films in the pH region of 5.5–8.5 if combined with a cationic polyelectrolyte having acidic pH values equal or lower than 7.5.[39] The interfacial contact between the donor and acceptor, and hence the resulting

© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

small 2015, 11, No. 1, 112–118

www.MaterialsViews.com

Figure 3. SEM cross-sectional images of (a) a ZnO NP thin layer on top of ITO layer; (b) the PbS QD coating the ZnO NP thin layer and penetrating into the pores; and (c) the sputtered ZnO film coated with a 300 nm thick PbS layer.

solar cell efficiency, are controllable according to a wellestablished thickness dependence method. In addition to the novelty of the current method and its outstanding features, it exceeds the field of solar cells to other fields requiring higher porosity or increased interfacial contact, e.g. catalysis, fuel cells or optoelectronics. Due to the antireflective properties of the film, it will keep its visual transparency and high light transmission if used with a transparent donor and will allow for more absorption of the incident solar energy.

2.2. Electron Injection from PbS QDs into ZnO To verify the electron injection from PbS QDs absorber layer to ZnO NPs at the heterojunction, we used fs TA spectroscopy with broadband capability and 120 fs temporal resolution as schematically shown in Figure 4 (more details about small 2015, 11, No. 1, 112–118

the experimental setup are given elsewhere).[33,40] Briefly, in TA spectroscopy, a 35 fs, 1 kHz, tunable wavelength pump pulse excites the PbS QDs. A tunable infrared or visible probe pulse overlaps in the heterojunctions and the absorbance change of the probe pulse is detected. As the electron is removed from the ground state, leaving an electron hole, the photogenerated exciton generates ground state bleach (GSB), which is probed during various time delays after the pump pulse. As can be seen in Figure 5A, the TA spectra of deposited PbS QDs without a ZnO layer shows the GSB of the 1S exciton band at ∼800 nm and a broad positive band from 700 to 760 nm immediately after excitation with photon energy at 1.2 times the band gap (Eg) of the PbS QDs. It is worth pointing out that the optical excitation at 1.2 Eg was chosen to avoid any contribution from the scattered light in the GSB and excited state absorption (ESA) signals, ensuring high-quality data and accurate dynamics for the electron transfer process. Since the ESA and the GSB are directly associated with the electron population in the conduction band of PbS QDs, we used them as a convenient probe to follow the carrier transfer from PbS QDs to ZnO NPs in real time. As can be seen in Figure 5A, in the absence of the ZnO layer, no change was observed in these exciton-induced bands. In contrast, the TA spectra of the PbS-ZnO heterojunction (Figure 5B) show that a fraction of the excited state absorption undergoes ultrafast decay, whereas that of the GSB remains constant. These notions are supported by the kinetics of the GSB and the positive band as shown in Figure 5C. This fast decay in the presence of ZnO NPs can be attributed to the injection of electrons from PbS QDs to ZnO NPs. The best fit from the data with an exponential decay results in an ultrafast time constant of 160 fs. It should be noted that we focus only on the early time dynamics for two reasons: i) it is the time scale for the electron injection between the donor and acceptor units, and ii) in the longer time scale, the carrier trapping and Auger recombination become competitive deactivation channels, an observation that it beyond the scope of this work. Because the energy levels of photoexcited electrons and holes of PbS QDs are at −3.7 and −5.0 eV[35,41] and the lowest unoccupied molecular orbital (LUMO) of ZnO NPs is −4.3 eV,[36] the estimated change in the Gibbs free energy (ΔG) that is associated with the driving force for the electron transfer[42] from the excited PbS QDs (donating species) to ZnO NPs (accepting species) is about −0.6 eV. Most importantly, based on the energetic alignment exciton, quenching hole transfer is not possible.[35,36] In addition, we also rule out the energy transfer mechanism due to the lack of spectral overlap between the absorption of ZnO NPs and the emission of PbS QDs. Thus, the above-mentioned ultrafast decay (160 fs) can be attributed to an ultrafast transfer of the photogenerated charge carrier from PbS QDs into the LUMO of ZnO NPs and another fraction of the single exciton has a much longer lifetime in the conduction band edge as observed in the deposited PbS QDs without ZnO layers. The time constant of the electron transfer from the excited PbS QDs to ZnO NPs is much faster than that from photoexcited PbS QDs into TiO2 (