Heterostructure Silicon Solar Cells with Enhanced Power Conversion Efficiency Based on Si x /Ni3+ Self-Doped NiO x Passivating Contact.

Developing efficient crystalline silicon/wide-band gap metal-oxide thin-film heterostructure junction-based crystalline silicon (c-Si) solar cells has been an attractive alternative to the silicon thin film-based counterparts. Herein, nickel oxide thin films are introduced as the hole-selective layer for c-Si solar cells and prepared using the reactive sputtering technique with the target of metallic nickel. An optimal Ni3+ self-doped NiO x film is obtained by tuning the reactive oxygen atmosphere to construct the optimized c-Si/NiO x heterostructure band alignment. A thin SiO x interlayer was further introduced to reduce the defect of the c-Si/NiO x interface with the UV-ozone (UVO) treatment. The constructed p-type c-Si/SiO x /NiO x /Ag solar cell exhibits an increase in the open voltage from 586 to 611 mV and achieves a 19.2% conversion efficiency.

Introduction

The successful development of advanced passivating-contact technology has boosted the power conversion efficiency (PCE) of the crystalline silicon (c-Si) solar cells by 26%.[1−3] A high-quality passivating contact could not only effectively reduce the carrier recombination near the silicon surface by passivating the silicon surface defects but also extract the carriers more effectively, which offers effective holes and electron separation and also a suitable low contact resistivity (ρc). To date, the most successful passivating-contact structure is based on intrinsic amorphous silicon (a-Si)/doped a-Si films, on which a PCE of 25.11% has been realized.[4] However, the intrinsic parasitic absorption in doped a-Si layers[5] can cause significant current losses for the solar cells. In addition, the toxic and flammable gases applied in the a-Si deposition process pose safety hazards.

In contrast to silicon-based thin films, transition metal oxides (TMOs) not only exhibit a wider work function adjustment space, breaking the confirmation of the c-Si band edge, but also possess more suitable optical characteristics for both sides of the device.[6,7] These TMOs can be deposited by methods such as thermal evaporation,[6,8] magnetron sputtering,[9] and atomic layer deposition.[10,11] Among these methods, the magnetron sputtering method is great for mass production because of its low cost, good compatibility with the scale-up production, and its universality for both flat and textured substrates.[12] Recently, hole-selective contacts using MoO[6,8] and VO[13,14] with a high work function have demonstrated their success in improving cell performance, achieving the high PCE of 23.5%[6] and 21.6%,[13] respectively. NiO is more abundant and stable and exhibits a wide band gap and p-type conductivity, which has been widely studied and applied as hole transport layer materials in perovskite solar cells[15,16] and dye-sensitized solar cells.[17] The large conduction band offset (ΔEc) and the small valence band offset (ΔEV) from c-Si enable NiO to extract the hole while blocking the electrons. However, the best PCE of c-Si solar cells with NiO passivating contact only has reached 15.2%,[18] which is far behind the reported results using hole-selective contact materials like MoO[6] and VO.[13] The limited energy band alignment and defects at the c-Si/NiO interface are to be a roadblock over 20%.

In this work, a detailed understanding of the oxygen partial pressure to the effect of the property of NiO films is provided. The Ni3+ self-doped NiO film is optimized to make the band alignment more suitable for c-Si/NiO contact, via adjusting the oxygen partial pressure during magnetron sputtering. Furthermore, it was found that an interfacial SiO thin film around a few nanometers formed spontaneously between the silicon and metal oxides during the TMO deposition could be beneficial to passivate the silicon surface and construct an effective passivating contact.[8,13,19−21] Herein, a thin SiO film on the rear surface of silicon solar cells was passively formed by low-temperature UV–ozone (UVO) oxidation treatment. With the combination of the better band alignment and improved interface quality, a high PCE of 19.2% is achieved for the heterostructure silicon solar cell with p-type c-Si/NiO (UVO)/Ag back contact. In addition, we also demonstrated magnetron sputtering as a scalable process route for the preparation of highly efficient selective transport layers.

Experiment Details

Film Deposition

NiO film was prepared by radio frequency (RF) reactive magnetron sputtering at room temperature. Before the sputtering deposition, the chamber was pumped down to a base pressure of 6 × 10–4 Pa. Then, a presputtering process was followed for 10 min to clean the target surface and remove any possible contamination. The working pressure was 0.2 Pa, and the sputtering power was 150 W. Four NiO film samples with a constant thickness of 30 nm were fabricated by varying the O2/Ar flow rate ratios (FO2/Ar) viz 3/36, 5/36, 10/36, and 15/36 on glass substrates for the measurements of their optical, electrical, and chemical characteristics. According to the value of the O2 flow rate, these NiO film samples were denoted as NiO-3, NiO-5, NiO-10, and NiO-15, respectively.

Device Fabrication

Silicon solar cells were fabricated based on 170 μm p-type CZ silicon wafers, with a resistivity of 2 Ω·cm. The standard 156 mm × 156 mm p-Si wafer was cut into small pieces with an area of 31.2 mm × 31.2 mm. The front surface emitter was prepared by thermal phosphorus diffusion, using a liquid source of phosphorous oxychloride (POCl3), resulting in an emitter with a depth of 250 nm and a doping concentration of 4 × 1021 cm–3. A double-layered SiN:H film was deposited on the emitter as the antireflection coating layer. The rear surface of each sample was dipped in dilute HF (5% V%) for 60 s and then rinsed with deionized water. Immediately, the samples were transferred to a UVO surface cleaner, in which only the bare rear silicon surface was exposed to the UVO at room temperature. Then, 4 nm thickness of NiO films were deposited on the UVO-treated rear surface of the silicon solar cell. Ag metal back electrodes were prepared by thermal evaporation without heating the substrate holder. No additional annealing treatment was introduced during and after the preparation of the solar cells.

Film and Device Characterizations

X-ray diffraction (XRD) characterization is carried out using Cu-Kα radiation in a Bruker D8 Advance X-ray diffractometer operated at 40 kV and 40 mA. Scans are performed between the 2θ range of 10° to 90° with 0.1° diffracted beam slits. Ultraviolet photoelectron spectroscopy (UPS) spectra excited by an unfiltered He I (21.20 eV) gas discharge lamp is collected by a SPECS PHOIBOS 100 hemispherical electron energy analyzer with an energy resolution of 100 meV. X-ray photoelectron spectroscopy (XPS) is done using Al-Kα (1486.61 eV) X-rays and the same hemispherical analyzer. The factor analysis of the XPS spectra is performed using the program XPSPEAK ver.4.1. The transmittance of the NiO films deposited on the glass substrate is measured using a Vis spectrometer (V-5600, SHjingmi). The optical band gap was calculated according to the Tauc-plot method. The surface morphologies of the films are studied using scanning electron microscopy (SEM, Gemin SEM 460, ZEISS). The interfacial evolutions of c-Si(p)/NiO/Ag and c-Si(p)/SiO/NiO/Ag stack are observed using a high-resolution transmission electron microscopy (HRTEM, TECHAIG2S-TWIN) instrument with the assistance of a focused ion beam (FIB) cross-sectioning. The measurement of the contact resistance relies on the well-established transfer length method (TLM).[22] The nickel oxide films with a fixed thickness of 4 nm under different FO2/Ar were sputtered on a polished 1 Ω·cm c-Si (p) wafer. A silver (Ag) electrode for I–V measurements was deposited by thermal evaporation. The current density–voltage (J–V) characteristics of the solar cells with c-Si(p)/NiO/Ag and c-Si(p)/SiO/NiO/Ag stack are measured under standard sun conditions (100 mW cm–2, AM1.5G spectrum, 25 °C) using a solar simulator (Newport Oriel sol3A) that is calibrated with a certified Fraunhofer CalLab reference cell. The spectral responses of the silicon solar cells were measured by the external quantum efficiency (EQE) measurement system (C9920-12, Hamamatus). The heterojunction contact energy band alignment was further simulated by AFORS-HET software from Helmholtz–Zentrum Berlin (HZB).

Results and Discussion

Because a suitable sputtering atmosphere has been proven to be the prerequisite for the preparation of high-quality TMOs,[13,23] we have prepared a series of NiO films in different O2/Ar conditions with pure Ni metal targets. XRD measurement is carried out to investigate the dependence of crystallinity of NiO films on different O2/Ar conditions, as shown in Figure a. It can be seen that the NiO film shows an amorphous structure when the oxygen flow rate is as low as 3 sccm. As the oxygen flow rate was increased to 5 sccm, a tiny diffraction peak at 43.3° corresponding to NiO (200) orientations (PDF card No.: 00-047-1049) emerges, and no other peaks can be recognized. The preferred orientation of (200) can be explained by the reason that the (200) crystalline plane is the most densely packed plane with minimized surface energy among the planes composed of nickel atoms and oxygen atoms.[24] Upon further increasing the oxygen flow rate, the full width at half maximum (FWHM) of the (200) peak becomes decreased, demonstrating an improved crystallinity. The results indicate that a sufficient supply of oxygen during the sputtering is of importance to enhance the crystallinity of NiO thin films.

Figure 1

Comparison of NiO thin films at various FO2/Ar. (a) XRD diffraction pattern. (b) SEM images. (c) Transmittance spectra and (d) optical band gap of the NiO thin films with different samples.

The surface morphology of NiO films observed by SEM is shown in Figures b and S1. Under low oxygen concentration conditions (NiO-3 and NiO-5), the density of the NiO-5 films became significantly worse, and more voids appeared on the surface. Moreover, as the oxygen increased, some wormlike bulges formed on the surface of the NiO films. The bulged areas had been observed on the surface, which we suggest may be the grains of NiO.[23] To evaluate the grain size, the samples were counted, as shown in Figure S2. It can be seen that the grains size becomes enlarged and the grain boundary becomes more obvious with the increase in oxygen flow rate. This tendency may relate to the improved crystallinity of NiO films, resulting from the increased oxygen flow rate, as shown in Figure a.

Transmittance spectra of the NiO films deposited under different FO2/Ar with the same sputtered time are plotted in Figure c, showing that the transmittance of the NiO samples reduces with increasing oxygen flow rate. The variation of film transmittance due to the sputtering atmosphere may be due to two reasons: on the one hand, there could be a change in the film composition, and, on the other hand, there might be a difference in film thickness because of the change in the sputtering rate. According to the transmittance spectra, we obtained the optical band gap of the films using the Tauc-plot method, as shown in Figure d. The obtained optical band gaps of NiO-3, NiO-5, NiO-10, and NiO-15 samples are 3.55, 3.52, 3.51, and 3.46 eV, respectively, showing a decreasing trend with the increase in the oxygen flow rate. The calculated optical band gap of NiO is consistent with the value from the other literature.[25]

XPS was used to analyze the effect of the FO2/Ar on the chemical components and electronic states of the NiO film. As shown in Figure , the asymmetric Ni 2p3/2 and Ni 2p1/2 doublet cannot be detected within the detection limit, suggesting the absence of the metallic nickel in all the NiO samples.[26] The characteristic multiplet splitting of the main Ni 2p3/2 peak reveals that the NiO is not only composed of Ni2+ ions but also a significant amount of Ni3+ ions. Ni3+ possibly comes from two species of Ni2O3 and NiOOH.[27] However, Ni2O3 is not stable at low temperatures in ambient air[12] and easily binds the atmospheric hydroxyl groups and forms NiOOH during the transfer of the sample.[28] Meanwhile, Ni2O3 species will lead to the improved optical band gap, according to the reported literature.[25,29] Our experimental results show that the band gap decreases with the increase in the Ni3+ ratio. Therefore, it is very likely that the Ni3+ peaks in our work come from the NiOOH species.

Figure 2

Ni2p XPS spectra of samples prepared with different FO2/Ar. (a) NiO-3. (b) NiO-5. (c) NiO-10. (d) NiO-15. All NiO films are deposited on silicon substrates.

In addition, the XPS analysis shows that the area ratio of Ni3+ to Ni2+ is enhanced from 0.26 to around 0.47, 0.52, and 0.56 for the NiO-3, NiO-5, NiO-10, and NiO-15, respectively. The stoichiometry of the corresponding samples calculated from the XPS data show an increasing trend of Ni3+/Ni2+ in the films with the increased FO2/Ar. The ratio of Ni3+/Ni2+ is related to the introduced tailbands,[30] which is responsible for the reduced optical band gap of the films prepared with higher FO2/Ar.[31]

In the silicon solar cell structure, the energy band alignment, as well as the interface between the charge transport layers, and the silicon absorbing layer play critical roles in charge extraction and transportation. To analyze the band alignment between c-Si/NiO, we performed UPS measurements to investigate the effect of FO2/Ar on the work function (WF: ϕ) and the valence band spectra maxima (EVBM) of the NiO thin films, as shown in Figure a,c. The WF and EVBM were obtained from the following formulas.where 21.2 is the value of the energy of the He light source, Ecutoff is the secondary electron cutoff energy, and Eonset is the energy difference between the Fermi energy (Ef) and the valence band energy (EV). Both the NiO-3 and NiO-5 samples have the same WF of 4.53 eV. As the FO2/Ar increases, the WF increased to 4.68 and 4.75 eV for the NiO-10 and NiO-15 samples, respectively. Meanwhile, the Eonset of the samples decreases from 0.79 to 0.63 eV with the increase of FO2/Ar, as shown in Figure d. The result is consistent well with the above XPS results. As the FO2/Ar increased, a larger portion of Ni2+ will be transformed into Ni3+, which will deepen the Fermi level and increase the work function, as shown in Figure b. It will be beneficial to hole transportation, and however, simultaneously change the energy band structure.

Figure 3

UPS spectra of NiO film samples. (a) Work functions measured from the photoemission cutoff in the He I spectra. (b) Work functions with different samples. (c) Valence band spectra depicted on the right-hand side are taken at He II excitation. (d) Ef–Ev with different samples.

To further investigate the c-Si/NiO band alignments, the simulation of different c-Si/NiO structures was carried out by AFORS-HET software. The parameters used in the simulation are listed in Table S1.

The ΔEV value of the c-Si/NiO heterojunction is given by the following formula.where ΔEV is the valence band energy difference between c-Si and NiO, Eg is the optical band gap, and χis the electron affinity energy. ΔEc is given by the following formula.where ΔEc is the conduction band energy difference between c-Si and NiO. Figure a presents the simulated band alignment of the c-Si/NiO heterojunction. The large ΔEc sufficiently blocks the transportation of electrons, while the small ΔEV makes the holes easily transported through the interface, as shown in Figure b. The band alignment exhibits a good carrier selectivity of the c-Si/NiO stack. The band parameters of the different c-Si/NiO heterojunctions are summarized in Table . The NiO-5 sample is found to possess the lowest ΔEV. It suggests that the NiO-5 possesses the minimal barrier to hole transportation and the best hole selectivity. The device performances of each c-Si solar cell are shown in Figure S3, which also demonstrates the best PCE for NiO-5 sample.

Figure 4

(a) Modeling band structure at the interface of c-Si/NiO without surface recombination. (b) The enlarged plot of the valence band at the c-Si/NiO interface corresponds to the red dotted rectangle shown in (a).

Table 1

Energy Band Parameter of the Different c-Si/NiO Heterojunction with Different FO2/Ar

FO2/Ar	Eg (eV)	χ (eV)	ΔEc (eV)	ΔEv (eV)
NiOx-3	3.55	1.74	2.11	0.116
NiOx-5	3.52	1.75	2.17	0.096
NiOx-10	3.51	1.82	2.08	0.156
NiOx-15	3.46	1.87	2.12	0.156

To further reduce the defect of c-Si/NiO interface, a thin passivation SiO film was inserted into the optimal c-Si/NiO-5 interface by the low-temperature UVO oxidation, which was labeled as sample NiO-5 UVO. Figure a shows the contact resistance characteristics of the c-Si/NiO/Ag structure with different NiOx thin films and the corresponding ρc extracted from the I–V curves. The measured resistance, which is uniformly proportional to the slope of I–V curves, increases with the increasing distance of the Ag electrode, as shown in Figure S4. By fitting the trend of resistance versus front Ag contact distance, the ρc for all the stacks is extracted, as shown in Figure b. A low ρc of 6.6 Ω·cm2 for the NiO-5 sample is obtained, demonstrating the minimal barrier for hole transportation, which could be attributed to the optimal band alignment. With the addition of the UVO-induced SiO passivating interlayer, the c-Si/ NiO(UVO)/Ag contact is calculated to possess the lowest ρc of 5.4 Ω·cm2 (Figure b). Similar results on the decrease in ρc with inserted SiO were also found both by Yang et al.[32] and Nayak et al.[18] In our previous study, we also found a similar result when we added the SiO at the interface of silicon and MoO.[8,11] These results may be due to the insertion of SiO film for chemical passivation and field-effect passivation of the NiO film,[13,19,33] which thereby reduces the suspended bonds on the silicon surface and density interface states (Dit),[34] as shown in Figure c,d.

Figure 5

(a) Contact resistance characteristics of the c-Si/NiO/Ag structure with different NiOx thin films measured using the transfer length method. (b) Extracted ρc of the corresponding measured samples, band alignment diagram of c-Si/NiO heterojunction (c) without and (d) with UVO treatment.

To further investigate the interfacial property of the c-Si/NiO/Ag with and without UVO treatment, HRTEM was constructed. It is found that an ultrathin interlayer of less than 5 nm thickness is formed between the Si and NiO for both the c-Si/NiO/Ag and c-Si/ NiO (UVO)/Ag samples (Figure a,b). XPS was used to characterize the interlayers, as shown in Figure S5. According to the patterns, the O-1s state and small Si–O peaks centered at ∼532.5 eV can be observed, which confirms the existence of a SiO layer for both the c-Si/NiO/Ag and c-Si/ NiO (UVO)/Ag samples. For the c-Si/NiO/Ag sample, the thin SiO interlayer around 2 nm may be naturally formed by the reaction between c-silicon and ambient oxygen. The corresponding elemental maps of the c-Si/NiO/Ag sample are shown in Figure c. It was found that Ag atoms will migrate into the c-Si layer, which will result in the formation of deep energy-level defects in the c-Si and deteriorate the device performance.[2,8,35] In comparison, the thicker SiO interlayer (∼4 nm) is observed in Figure b for the sample of c-Si/NiO (UVO)/Ag. The corresponding elemental maps in Figure d show the little Ag atom migration into the c-Si layer. It means that a thicker 4 nm SiO interlayer could block the Ag atom migration, which will reduce the possible interface defects and the contact resistance.

Figure 6

Morphology, crystal structure, and chemical information of c-Si /NiO/Ag with and without the UVO treatment sample. (a,b) TEM image of c-Si /NiO/Ag without and with UVO treatment for 30 s. (c,d) Corresponding elemental maps of c-Si/NiO/Ag without and with UVO treatment for 30 s.

The p-type c-Si solar cell with different full-area layer stacks at the rear side was then constructed, as sketched in Figure a. The characterization current density–voltage curve of c-Si solar cells under one sun standard illumination is presented in Figure b. The main photovoltaic parameters are summarized in Table . The control cell with Ag rear contact exhibits a low efficiency of 14.9%, which is mainly attributed to the low open-circuit voltage (Voc) (586 mV) and fill factor (FF) (68.3%). The poor performance of Voc and FF results from the high carrier recombination velocity and Schottky barrier at the c-Si/Ag interface.[35] The intercalation of NiO efficiently improves the contact conduction between the absorber and the electrode, enhancing the cell efficiency to 17.1%. Furthermore, the c-Si/NiO (UVO)/Ag sample with a SiOx interlayer shows the best PCE of 19.2%, the highest Voc of 611 mV, a short-circuit current (Jsc) of 37.4 mA/cm2, and the best FF of 79.6%. The SiO layer enhanced the electron blocking and reduced the Dit at the interface, meanwhile blocking the Ag metal atoms from entering the silicon surface to form deep energy-level defects. The external quantum efficiency (EQE) was conducted and shown in Figure c, confirming an enhanced EQE in the c-Si/NiO (UVO)/Ag stack. The integrated photocurrent from the EQE spectra for c-Si/NiO (UVO)/Ag is 37.37 mA/cm2, which is higher than that of 36.47 mA/cm2 for c-Si/NiO/Ag. The integrated photocurrent of the two samples all matches well with the Jsc measured by I–V measurement. Because PCE is mainly limited by Voc and FF due to the insufficient electrical conductivity of NiO and the relatively high contact resistance, our next research work will be focused on surface passivation modification to improve the conductivity of NiO films without affecting the energy-band matching.

Figure 7

(a) Sketch and (b) photovoltaic performance of c-Si /Ag, c-Si /NiO/Ag, and c-Si/NiO (UVO)/Ag. (c) EQE of the NiO-5 without/with UVO treatment.

Table 2

Photovoltaic Parameters of the Inverted Planar P-Type Silicon Solar Cell with a Different Structure

structure	Jsc (mA/cm2)	Voc (mV)	FF (%)	efficiency (%)
c-Si/Ag	37.4	586	68.3	14.9
c-Si/NiOx/Ag	36.4	600	78.1	17.1
c-Si/ NiOx (UVO)/Ag	37.4	611	79.6	19.2

Conclusions

In summary, we demonstrated NiO as the effective hole-selective contact to improve the contact property of c-Si/Ag contact for c-Si solar cells. The Ni3+ doped NiO films were controlled by tuning the oxygen atmosphere during the magnetron sputtering process. As controlling the O2/Ar flow rate ratios of 5/36, the optimized Ni3+ self-doped NiO film was obtained and showed the most suitable band alignment for c-Si/NiO contact, which has been confirmed by AFORS-HET software theoretical simulation. Furthermore, the SiO interlayer was introduced to the c-Si/NiO interface by the UVO treatment, which was found to reduce the interface defects. For the c-Si/NiO (UVO)/Ag structure, the lowest contact resistivity of 5.4 Ω·cm2 is obtained. The c-Si solar cell with full area SiO/NiO/Ag hole-selective contact shows a PCE up to 19.2% because of both the optimal c-Si/NiO band alignment and the reduced interfacial defects. This is the highest PCE data reported on the c-Si solar cells with hole-selective NiO contact to date. The results provide a large industrialization potential to design high-efficiency c-Si solar cells with carrier-selective passivating contact structures.