Inorganic p-Type Semiconductors: Their Applications and Progress in Dye-Sensitized Solar Cells and Perovskite Solar Cells

Abstract

Considering the increasing global demand for energy and the harmful ecological impact of conventional energy sources, it is obvious that development of clean and renewable energy is a necessity. Since the Sun is our only external energy source, harnessing its energy, which is clean, non-hazardous and infinite, satisfies the main objectives of all alternative energy strategies. With attractive features, i.e., good performance, low-cost potential, simple processibility, a wide range of applications from portable power generation to power-windows, photoelectrochemical solar cells like dye-sensitized solar cells (DSCs) represent one of the promising methods for future large-scale power production directly from sunlight. While the sensitization of n-type semiconductors (n-SC) has been intensively studied, the use of p-type semiconductor (p-SC), e.g., the sensitization of wide bandgap p-SC and hole transport materials with p-SC have also been attracting great attention. Recently, it has been proved that the p-type inorganic semiconductor as a charge selective material or a charge transport material in organometallic lead halide perovskite solar cells (PSCs) shows a significant impact on solar cell performance. Therefore the study of p-type semiconductors is important to rationally design efficient DSCs and PSCs. In this review, recent published works on p-type DSCs and PSCs incorporated with an inorganic p-type semiconductor and our perspectives on this topic are discussed.

1. Introduction

Since the seminal paper of O’Regan and Grätzel in 1991 [1] dye-sensitized solar cells (DSCs) have become attractive for harvesting solar energy [2,3,4,5,6]. Recent developments have reported power conversion efficiencies (PCEs, η) exceeding 12%, achieved via optimization with materials such as panchromatic dyes and redox couples exhibiting higher redox potentials [7,8,9,10,11]. In DSCs, the dye absorbs light to form an exciton, then the charge generation is performed at the semiconductor-dye interface, and the semiconductor and the electrolyte serve as the charge transporting material. The typical DSC architecture based on a redox electrolyte is broadly composed five components: (1) a mechanical support coated with a transparent conductive oxide (TCO); (2) the semiconductor film, usually TiO₂; (3) a sensitizer (S) adsorbed on the surface of the semiconductor; (4) an electrolyte containing a redox mediator (M/M⁻); (5) a counter electrode capable of regenerating the redox mediator. The sensitizer S is excited by absorption of a photon (Equation (1)). Then the excited sensitizer S* injects an electron into the conduction band of the semiconductor (Equation (2)). The injected electron flows through the n-SC network to the back contact and then through the external load to the counter electrode where it reduces the redox mediator (Equation (3)), which in turn regenerates the oxidized sensitizer S⁺ (Equation (4)). This completes the circuit. Under illumination, the device constitutes a regenerative and stable photovoltaic energy conversion system. The overall efficiency of the device depends on optimization and compatibility of each of the constituents. Losses occur mainly through the recombination of the injected electrons either with the oxidized sensitizer (Equation (5)) or with the oxidized redox couple (M) at the n-SC surface (Equation (6)):

$$S_{\left(adsorbed\right)}+h\upsilon \to S_{\left(adsorbed\right)}^{*}$$

(1)

$$S_{\left(adsorbed\right)}^{*}\to S_{\left(adsorbed\right)}^{+}+e_{\left(injected\right)}^{-}$$

(2)

$$M+e_{\left(CE\right)}^{-}\to M_{\left(CE\right)}^{-}$$

(3)

$$S_{\left(adsorbed\right)}^{+}+M^{-}\to S_{\left(adsorbed\right)}+M$$

(4)

$$S_{\left(adsorbed\right)}^{+}+e_{\left(injected\right)}^{-}\to S_{\left(adsorbed\right)}$$

(5)

$$M+e_{\left(injected\right)}^{-}\to M^{-}$$

(6)

A solid-state hole transport material (HTM) has been widely used to replace the liquid electrolyte in DSC, which is to say that a hole transports through a p-SC as shown in Figure 1a. It is noteworthy that replacement of a solid-state HTM aims to avoid concerns about solvent leakage and corrosion from the typical iodide/triiodide redox couple in liquid electrolyte based DSC. The operating principle complies with the n-type sensitization. Therefore the sensitizer S is excited by absorption of a photon (Equation (1)). Then the excited sensitizer S* injects an electron into the conduction band of the n-SC while the hole transports through HTM and arrives at the metallic cathode. 2,2′,7,7′-Tetrakis(N,N-di-p-methoxyphenylamine)-9,9′-spirobifluorene (spiro-OMeTAD) has been the most successful p-type organic conductor employed. Its work function is about 4.9 eV and the hole mobility, 10⁻⁴ cm²/(V·s) [12]. Reported first in 1998 [13], the conversion yields have increased over the last decades, i.e., from a fraction of a percentage to over 7% [14,15,16,17,18,19]. The origin of its good efficiency comes from the inherent properties of spiro-OMeTAD: amorphous, soluble, and small molecular size. However, its high cost, low hole mobility and lack of long term stability call for replacement with new hole transport materials including inorganic materials, e.g., NiO, CuI, CuSCN, CsSnI₃ and so on.

Most of the researches on DSCs have been devoted to architectures based on n-type semiconductors (n-SC) [20,21,22,23,24]. It is however possible to use a p-type semiconductor as photocathode. The working principle is very similar to an n-type DSC, the difference being that the excited state of the sensitizer is now reductively quenched by the semiconductor, which is the sensitizer excited state injects holes into the valence band of the p-type semiconductor [25,26,27]. While the general design of dyes consists of electron-donor (D) and electron-acceptor (A) linked with π conjugation (D-π-A), the main difference between n-type dye and p-type dye is position of the acceptor: the anchoring group is attached on the electron-acceptor part for n-type dyes while the group is attached on the electron-donor part for p-type dyes. Under illumination, the sensitizer S is excited by absorption of a photon (Equation (1)) like n-type DSC. Then the excited sensitizer S* injects a hole into the valence band of the semiconductor (Equation (7)). The injected hole flows through the semiconductor network to the back contact and then through the external load to the counter electrode where it oxidises the redox mediator (Equation (8)), which in turn regenerates the reduced sensitizer S⁻ (Equation (9), see Figure 1b):

$$S_{\left(adsorbed\right)}^{*}\to S_{\left(adsorbed\right)}^{-}+h_{\left(injected\right)}^{+}$$

(7)

$$M^{-}+h_{\left(CE\right)}^{+}\to M_{\left(CE\right)}$$

(8)

$$S_{\left(adsorbed\right)}^{-}+M\to S_{\left(adsorbed\right)}+M^{-}$$

(9)

The emerging solid-state organometallic lead halide perovskites have recently attracted tremendous attention because of their promising power conversion efficiencies which have been boosted to over 20% in recent years [28]. As evolved from solid-state DSCs, the common perovskite solar cells (PSCs) display device architecture that is composed of a compact titania dioxide layer, a mesoporous metal oxide scaffold infiltrated by the perovskite and covered by a perovskite capping layer, and a p-type organic semiconductor film contacted by silver or gold. Since the first successful application of spiro-OMeTAD in solid-state perovskite solar cells, spiro-OMeTAD has been extensively employed for fabrication of high-efficiency perovskite solar cells [29]. Recently, many HTMs have been demonstrated to replace spiro-OMeTAD aiming for low-cost and stability. The p-type contact materials for PSCs can be categorized into organic and inorganic semiconductors [30,31]. Organic HTMs, including small molecular and polymeric HTMs, are low-cost alternatives capable of low-temperature device processing. Small molecular HTMs are easy to purify, suitable for forming crystalline thin films, and feasible to match the band gaps of various perovskites by modifying their structures. On the other hand, polymeric HTMs present good stability and processibility, and are applied to fabricate PSCs with decent efficiency. The most widely used organic HTM, poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS), for the inverted p-i-n planar heterojunction perovskite solar cells has been reported to achieve a recorded PCE of 18.1% [32]. However, PEDOT:PSS causes device degradation due to its acidity and hydrophilic properties. Compared with organic HTMs, inorganic p-type semiconductors (i.e., NiO, CuI, CuSCN, CuO and Cu₂O) as HTMs appear to be an ideal choice due to their low cost, ease of synthesis, high mobility, high transparency in the visible region and good chemical stability. Accordingly, PSCs employing inorganic HTMs have been demonstrated to exhibit decent stability. Furthermore, the use of inorganic charge transport layer of electron transport material (ETM) and HTM for PSCs has recently been demonstrated to protect the perovskite photoactive layer from exposure to ambient environments, thus enhancing the resistance to degradation of perovskite and achieving highly stable perovskite-based solar cells [33,34,35,36]. The scope of this perspective review is to present the recent developments in DSCs and PSCs based on inorganic p-type semiconductors, including NiO, CuSCN, CuI, Cu₂O and CuO.

2. p-Type Hole Transport Materials for DSCs

An appropriate HTM in solid-state dye-sensitized solar cells (SSDSCs) must satisfy several requirements: (i) the upper edge of the valence band of the p-type semiconductors must be located above the highest occupied molecular orbital (HOMO) ground state level of the dye; (ii) its hole mobility should be sufficiently high; (iii) it must be able to fill pores of mesoporous n-SC; (iv) it should be transparent in the range of dye absorption mostly, e.g., the visible light; and (iv) it should be stable during operation [13,37,38,39,40,41]. Copper-based inorganic semiconducting hole transport materials, such as CuI, CuSCN, CuO and Cu₂O, have shown promises for their use in dye-sensitized and quantum dot-sensitized solar cells. These materials can be deposited with solution process with good pore-filling. Meanwhile, these wide-band-gap semiconductors have high conductivity, suitable energy level and good transparency.

2.1. CuSCN

Copper thiocyanate (CuSCN) serves as a universal hole-transport interlayer material for numerous optoelectronic applications, including transparent thin-film transistors, organic, dye-sensitized, and perovskite solar cells, and organic light-emitting diodes. CuSCN combines intrinsic hole-transport (p-type) characteristics with a large bandgap (>3.5 eV) to show good optical transparency throughout the visible to near infrared spectrum and is chemically robust owing to its polymeric structure [42,43]. CuSCN is readily available from commercial sources while it is inexpensive and fabricated by a variety of methods such as dip-coating, and spray-coating or doctor-blading, chemical bath deposition, and electrochemical deposition. Furthermore, deposition of CuSCN using low-temperature with solution processes makes it compatible with flexible substrates. As compared to the widely utilized HTM spiro-OMeTAD, CuSCN exhibits a superior optical transparency, higher hole mobility (0.01–0.1 cm²/(V·s) vs. Spiro-OMeTAD: 1.6 × 10⁻³ cm²/(V·s)) and good chemical stability [44].

In 1996, O’Regan and Schwartz demonstrated SSDSC based on formed by a heterojunction with rhodamine-coated TiO₂ and electrodeposited CuSCN (TiO₂/dye/CuSCN), exhibiting an incident photon to current conversion efficiency (IPCE) of 2.2% (absorbed photon to current conversion efficiency (APCE) of >70%) [45]. Apparently improved photocurrent was demonstrated by Kumara et al. wherein CuSCN dissolved in n-propyl sulfide was deposited on Ru-dye coated TiO₂ film, showing PCE of 1.25% and J_sc of 3.52 mA/cm² [46]. In 2002, O’Regan et al. reported a further improved SSDSC by pore filling control, performing an efficiency of 2.1% with a higher photocurrent of ~8 mA/cm² [47]. The results so far, however, are far from comparable with liquid electrolyte-based DSCs because the recombination of electrons injected to n-SC with the S⁺ or with CuSCN is relatively faster than that of liquid electrolyte-based DSCs [48]. Surface passivation of n-SC was introduced, e.g., Al₂O₃ on TiO₂ (2.1% with improved V_oc) [49] and MgO on SnO₂ (2.82% with 10.18 mA/cm²) [50]. Conductivity enhancement was also addressed as another approach to further improve SSDSCs. Perera et al. demonstrated that an excess of SCN-doped CuSCN improved its conductivity via creation of acceptor levels located 0.9 eV above the valence band edge, resulting in an improved PCE of 2.39% from 0.75% [51]. In the same sense, modifications of the CuSCN crystal structure by introducing triethylamine-coordinated Cu(I) [52] or Cu(II) [53] were demonstrated. It was found that the triethylamine-coordinated Cu(II) led to hole conductivity from 0.01 to 1.42 S/m and its application into SSDSC achieved a PCE of 3.4% [53].

2.2. CuI

Another p-type wide-bandgap (~3.1 eV) semiconductor, copper iodide (CuI), is an ionic solid and serves as a potential candidate of HTM on the basis of its suitable valance band (VB) position, good optical transparency, higher hole mobility (0.5–2 cm²/(V·s)) over CuSCN, and compatibility of solution-deposited process with the perovskite absorber. In 1995, Tennakone et al. first demonstrated an SSDSC based on CuI HTM in conjunction with cyanidin dye [54], which was later improved by replacement with a Ru-bipyridyl complex dye, showing a PCE of 4.5% [55]. There have been several strategies for photovoltaic performance enhancement. For instance, retarding crystallization of CuI that is known to inhibit the pore filling of mesoporous n-SC was proposed to improve pore filling by addition of a small amount of 1-methyl-3-ethylimidazoliumthiocyanate (MEISCN) [56,57] or triethylamine hydrothiocyanate (THT) [58]. The resulting efficiency of 3.75% with impressive stability for about 2 weeks under continuous illumination was demonstrated [58]. Surface passivation with a thin MgO layer on mesoporous TiO₂ was applied [59,60] and its combination with THT treatment achieved an enhanced PCE of 4.7% [60]. Further improvement achieved by interfacial engineering to improve contact between CuI and dye molecules was demonstrated: [55,61,62,63] and the best PCE of 7.4% from SSDSC employing PEDOT:PSS modified with guanidine thiocyanate was reported by Sakamoto et al. in 2012 [64].

2.3. CsSnI₃

CsSnI₃ belongs to a class of semiconducting perovskites of the composition CsSnX₃ (X = Cl, Br, I or mixed halides) and exist in two polymorphs at room temperature: a yellow coloured one-dimensional double-chain structure (γ-CsSnI₃) and a black coloured three-dimensional perovskite structure (B-γ-CsSnI₃). Recently the B-γ-CsSnI₃ showed its potential as a HTM in SSDSC on account of its properties including a high hole mobility of 585 cm²/(V·s) at room temperature, high p-type metal-like conductivity (>200 S/cm²), and the well-matched VB with the HOMO of Ru dye (N719) for efficient charge separation [65,66]. Chung et al. showed that by doping CsSnI₃ with 5% SnF₂ the efficiency could be further improved and combination with fluorine plasma treated TiO₂ electrode and a photonic crystal introduction achieved 8.51% with a shadow mask [65]. Very recently Lee et al. demonstrated Cs₂SnI₆ as a HTM, where Sn is in the 4⁺ oxidation state, which is to say that it is more stable over CsSnI₃. The SSDSC with Ru dye (Z907) showed a PCE of 4.63% and the cell with mixture dyes and introducing a phonic crystal achieved a PCE of 7.8% (It is noted that use of a mask is not described) [67]. Achievement of the high efficiencies has proved the potential of these materials and has opened up the opportunity to further optimize SSDSCs. We summarize the performance of p-type inorganic SC-based SSDSCs with their photovoltaic parameters in

Table 1. Summary of photovoltaic characteristics of dye-sensitized solar cells (DSCs) based on p-type inorganic solar cells (SCs).

Type 1	Architecture 2	Voc (V)	Jsc (mA/cm2)	FF (%)	η (%) 5	Ref.
CuSCN
HTM	FTO/mp-TiO2/N3/CuSCN/graphite	0.616	3.52	0.576	1.25	[46]
HTM	FTO/cp-TiO2/mp-TiO2/N3/CuSCN/graphite	0.60	7.8	0.44	2.1	[47]
HTM	FTO/cp-TiO2/mp-TiO2/Al2O3/N3/CuSCN/graphite	0.69	5.1	0.59	2.1	[49]
HTM	FTO/cp-SnO2/mp-SnO2/MgO/D149/CuSCN/graphite	0.544	10.18	0.51	2.82	[50]
HTM	FTO/mp-TiO2/N3/SCN doped CuSCN/graphite	0.512	9.09	0.514	2.39	[51]
HTM	FTO/mp-TiO2/N719/THT + CuSCN/Pt	0.578	10.52	0.556	3.39	[53]
CuI
HTM	FTO/cp-TiO2/mp-TiO2/ZnO/N3/MEISCN + CuI/Au	0.59	6.84	0.57	3.8	[57] 3
HTM	FTO/cp-TiO2/mp-TiO2/MgO/N3/ CuI/Au	0.51	8.74	0.54	2.9	[59] 4
HTM	FTO/ mp-TiO2/MgO/N3/CuI/Au	0.62	13	0.58	4.7	[60]
HTM	FTO/cp-TiO2/mp-TiO2/N3/CuI/PEDOT:PSS (carbon black: 200 wt%)	0.739	14.5	0.69	7.4	[64]
CsSnI3
HTM	FTO/mp-TiO2/N719/CsSnI3/ZnO PC/Pt	0.723	15.9	0.739	8.51	[65]
Cs2SnI6
HTM	FTO/cp-TiO2/mp-TiO2/Z907/Cs2SnI6/Pt	0.571	13.2	0.613	4.63	[67]
HTM	FTO/cp-TiO2/mp-TiO2/multi-dyes/Cs2SnI6/Pt with 3D PC	0.618	18.6	0.680	7.80
NiO
p-DSC	FTO/mp-NiO/dye 3/I3− & I−/Pt	0.218	5.35	0.35	0.41	[27]
p-DSC	FTO/2 steps-NiO/P1/I3− & I−/Pt	0.084	5.48	0.34	0.15	[70]
p-DSC	FTO/NiO-μBs/PMI-6T-TPA/I3− & I−/Pt	0.208	6.36	0.34	0.46	[74]
p-DSC	FTO/NiO-NR/PMI-6T-TPA/I3− & I−/Pt	0.292	3.30	0.41	0.40	[75]
p-DSC	FTO/RDS-NiO /P1/I3− & I−/Pt	0.125	2.84	0.34	0.12	[76]
Cu(I) delafossite
p-DSC	FTO/CuAlO2/PMI-6T-TPA /I3− & I−/Pt	0.333	0.33	0.40	0.041	[87]
p-DSC	FTO/CuAlO2/O2/I3− & I−/Pt	0.103	0.954	0.375	0.037	[89]
p-DSC	FTO/CuGaO2/P1/Co3+ & Co2+/Pt	0.357	0.165	0.308	0.018	[88]
p-DSC	FTO/CuGaO2/P1/T− & T2/CoS	0.199	2.05	0.445	0.182	[90]
p-DSC	FTO/CuGaO2:1%Mg/PMI-NDI/ Co3+ & Co2+/Pt	0.305	0.415	0.35	0.045	[95]
p-DSC	FTO/CuCr0.9Ga0.1O2/P1/I3− & I−/Pt	0.134	1.56	0.489	0.10	[96]

¹ In type, “p-DSC” indicates “p-type sensitization”; ² “FTO” is “ fluorine-doped tin oxide”. “cp” and “mp” indicate “compact film” and “mesoporous film”. PEDOT:PSS is poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate). “PC” indicates “photonic crystal”. “μBs”, “NR”, and “RDS” indicates “microballs”, “nanorod”, and “rapid discharge sintering”, respectively; ³ The intensity of the incident light of 61 mW/cm² is used; ⁴ The intensity of the incident light of 83 mW/cm² is used; ⁵ power conversion efficiency (PCE, η) = J_sc × V_oc × FF/I₀, where J_sc is the short circuit current density; V_oc is the open circuit voltage; FF is fill factor, and I₀ is the intensity of the incident light.

.

3. p-Type Sensitization for DSCs

Candidates for p-type semiconductors for sensitization must satisfy several fundamental parameters. The position of the valence band should be higher than the HOMO of the dye for effective hole injection from the photoexcited dye while the position should be favourably low enough to render high photovoltage, which is determined with the energy gap between the valence band and the redox potential of an electrolyte. On the other hand, the conduction band should be high enough to block electron injection from the photoexcited dye. Therefore wide-band gap p-type semiconductors such as nickel oxide (NiO, a band gap of ~3.6 eV) that are also resistant to photocorrosion are promising candidates. The surface of the semiconductor must exhibit high chemical affinity to an anchoring group of dye in order to promote the adsorption of the dyes on the surface. Mesoporous films that exhibit large surface area to accommodate high numbers of dyes, leading to high light harvesting efficiency must be easily formed in a cost effective way. The molecular dye films adsorbed on a flat surface can only harvest a negligibly small fraction of the incoming light. Indeed, monolayers of porphyrins, which have among the highest extinction coefficients known, absorb far less than 1% of the sun sepctrum, derived from the Equation (10):

$$LHE\left(\lambda \right)=1-{10}^{-A}=1-{10}^{-\sigma \mathsf{\Gamma }}$$

(10)

where A is the absorption optical density of sensitized film, σ is the absorption cross section (cm²/mol) which equals to the product of ε (the molar extinction coefficient M⁻¹·cm⁻¹) and 1000 (cm³/L), Γ is the dye loading per projected surface area of the film. For a 10 μm-thick film the surface is enlarged over 1000 times allowing for efficient harvesting of sunlight, so high surface area materials needs to increase the light harvesting efficiency (LHE) of a sensitizer layer as a consequence of high number. In this respect, nano-scale materials, e.g., nanoparticles, nanorods, nanowires etc. have been demonstrated as a good candidate.

3.1. NiO

Since He et al. first demonstrated the generation of a cathodic photocurrent from organic dye sensitized nanostructured NiO cathode [68], many sophisticatedly designed dyes have been reported (details about their molecular structures can be found in [26]). Among them, a donor-acceptor type dye, comprising a perylene monoimide (PMI) as the acceptor and an oligothiophene coupled to triphenylamine as the donor holds the record efficiency of 0.41% [27]. The use of the tailored bridge lengths allowed for the control over the charge recombination events and the charge recombination was decelerated by several orders of magnitude [27]. A main drawback of NiO is the hole diffusion coefficient in the range of 10⁻⁸ to10⁻⁷ cm²/s [69,70], which is several orders of magnitude lower than the electron diffusion coefficient in TiO₂. Therefore, intriguing routes to form good quality NiO films have been reported [70,71,72,73,74,75,76,77]. Sumikura et al. demonstrated good quality NiO electrodes obtained by the use of poly(ethylene oxide)-poly(propylene oxide)-poly(ethylene oxide) triblock copolymers as a template [72] and Li et al. demonstrated a PCE of 0.15% with a modified method [70], combined with a p-type dye, P1 (the triphenylamine moiety as the electron donor, malononitrile moiety as the electron acceptor, and a thiophene unit as the conjugated chain) [71]. Nanostructured NiO microballs (μBs) were synthesized by a thermolysis method and applied in p-DSCs. The high surface area and favourable pore size distribution of NiO-μBs facilitate increasing dye loading, leading to high J_sc (7.0 mA/cm²) and IPCE (74%) with a PCE of 0.46% [74]. Electrode films composed of one-dimensional structures are expected to offer improved charge transport over a random polycrystalline network [78]. Crystalline NiO nanorods (NR) with average diameters of sub-10 nm formed through a nanocasting process, using mesoporous SiO₂ as a template was reported and its application in p-DSC achieved a PCE of 0.40% with improved V_oc (292 mV) and fill factor (FF) (41%) [75]. Another interesting result reported by Gibson et al. was demonstration of a scalable method for the preparation of nanoporous NiO electrodes. A film composed of NiO nanoparticles formed by spraying method was sintered using a technique based on microwave-assisted plasma sintering (called rapid discharge sintering, RDS), which was faster than the conventional sintering process. Its application in P1-sensitized p-type DSC achieved overall efficiencies as high as 0.12% and IPCEs of about 50% [76]. Very recently, Zhang et al. demonstrated a solid state p-type DSC in which NiO is employed as a p-SC, and the triphenylamine (TPA) based organic dye P1 as a sensitizer, and a solid electron transport material (ETM), phenyl-C61-butyric acid methyl ester (PCBM), is employed to replace liquid redox couple electrolyte [79]. Although the p-SSDSC device gave a low J_sc of 50 μA/cm², a promising V_oc of 620 mV was demonstrated and that an electron transfer to PCBM and a hole injection into NiO from the excited dye was proved by ultrafast transient absorption spectroscopy.

3.2. Cu(I) Delafossite

The valence band of NiO as low as +0.5 V vs. NHE limits the photovoltage from p-DSCs and other type p-SC like CuSCN- [80,81], CuO- [82], GaP- [83], and Cu(I)-based delafossites have been proposed as an alternative. Among them, the Cu(I)-based delafossite materials (CuMO₂, M = Cr, Ga, Al) have shown promising results because of their deeper position of the valence band edge and optical transparency in the visible range. Moreover, in Cu(I)-based delafossite materials, that Cu 3d orbitals strongly hybridize with the O 2p orbitals, making the hole charge carriers be delocalized both on oxygen and copper atoms [84] which facilitates the hole mobility, for instance 10–10⁻² cm²/(V·s) for CuAlO₂ and CuGaO₂ [84,85,86]. A solid-state reaction synthesized CuAlO₂ and hydrothermally synthesized CuGaO₂ proved V_ocs of 333 mV [87] and 357 mV [88], with respectively the following dye and redox mediator couples: PMI-6T-TPA and I₃⁻/I⁻, and PMI-NDI and Co³⁺/Co²⁺ (dtb-bpy). However, the limited film surface area resulted in small currents below 0.4 mA/cm². More effort to decrease the size of CuAlO₂ and CuGaO₂ improved the current to 0.954 mA/cm² [89] and 2.05 mA/cm² [90] and the hydrothermally synthesized CuCrO₂ nanocrystals with a high surface area of 87.9 m²/g, combined with the use of 1-methy-1H-tetrazole-5-thiolate (T⁻)/disulfide dimer (T₂) redox shuttle, P1, and CoS counter electrode showed 1.73 mA/cm² [91]. Recently metal doped Cu-based delafossite compounds have been reported to increase electrical conductivity [92] or change surface affinity to increase dye adsorption [93] or control crystal size [94]. Ursu et al. demonstrated relatively higher J_sc (0.093 mA/cm²) with 5% Al-doped CuGaO₂ nanoparticles with N719 sensitizer compared to J_sc (0.046 mA/cm²) of the cells using undoped CuGaO₂ [92]. The same group reported also Fe-doped CuGaO₂ as a photocathode in p-type DSCs [93]. Fe-doped CuGaO₂ increased hydrophilicity and adsorptivity of sensitizer compared to pure CuGaO₂ by formation of Fe³⁺-OH species on the surface. More hydrophilic Fe-doped CuGaO₂ films resulted in slightly increased J_sc from 0.046 mA/cm² to 0.056 mA/cm². Renaud et al. reported an improved performance from Mg-doped CuGaO₂, which is attributed to an increased surface obtained by addition of small amounts of Mg, and a higher V_oc of 305 mV with a higher J_sc of 0.415 mA/cm² was achieved [95]. Very recently, Ga-doped CuCrO₂ nanocrystalline synthesized by a hydrothermal method has achieved an improved J_sc of 1.56 mA/cm² [96]. We summarize the performance of p-type DSCs incorporating p-type inorganic SC with their photovoltaic parameters in Table 1.

3.3. Tandems DSC

The development of p-type DSCs opens possibilities for significant improvement of existing n-type DSCs by making tandem cells. In a tandem DSC, the cathode of conventional n-type DSC is replaced by a photocathode based on a p-type semiconductor, for example, based on NiO. The architecture is shown in Figure 2.

The first promising advantage of the tandem DSC design is the obvious possibility to use two sensitizers with complementary spectral response, one on each type of semiconductor. The second advantage of such a design is the expected increase of the open circuit voltage. Indeed, in an n-type DSC, the maximum V_oc value, V_oc(n-SC) is considered to be the energy difference between the oxidation potential of the redox mediator and the energy level of the quasi-Fermi level in the n-type semiconductor. In a p-type DSC, the maximum V_oc value, V_oc(p-SC) is the energy difference between the oxidation potential of the redox mediator and the energy level of the quasi-Fermi level of the free electrons in the p-type semiconductor. In the tandem DSC, the maximum value for the open circuit voltage is the sum of both components, V_oc = V_oc(n-SC) + V_oc(p-SC). The proof of concept of this strategy has been reported in 2000, using N719 as a sensitizer on TiO₂ for the photoanode and erythrosine B as a sensitizer on NiO for the photocathode [98]. The performance of this system has been improved by more than 50% (from η = 0.39% to 0.66% under air mass (AM) 1.5) by using a mesoporous NiO film of higher quality [99].

Due to the definition of V_oc in a tandem DSC, it can be deduced that the electrochemical potential of the redox mediator has no impact on its value. Therefore it is possible to replace the I⁻/I₃⁻ couple by the Co³⁺/Co²⁺ (dtb-bpy) in the electrolyte without impacting the V_oc [25]. Finally, using a dye comprising triphenylamine-oligothiophene-PMI, a tandem DSC with a power conversion efficiency of 2.42% was obtained [27]. Taking into account that in this case, the spectral responses of both dyes were similar, this is a very promising result for future devices using dyes with a complementary spectral response.

4. p-Type Inorganic Hole Transport Materials for PSCs

In this section, we focus on recent developments in perovskite solar cells based on inorganic p-type semiconductors, including NiO, CuSCN, CuI, Cu₂O and CuO. Ideal HTMs for highly efficient PSCs are required to offer the following features: appropriate valence band for efficient electron blocking and hole collection, high hole mobility, stable thermal and optical properties. Based on the contact electrode polarity, DSC-like structures that work with n-i-p heterojunction are firstly introduced, followed by the organic photovoltaic (OPV)-like inverted structures of p-i-n heterojunction. With the evolution from mesoscopic sensitized junction, non-injecting super mesostructure scaffolds to planar perovskite solar cells, the configuration of photovoltaic devices can be categorized into mesoscopic and planar junction devices. For both DSC-like and OPV-like architecture, mesoscopic and planar PSCs schematically illustrated in Figure 3 are also reviewed.

4.1. NiO

As mentioned above, NiO is a low-cost material with superior properties, such as thermal and chemical stability, optical transparency, which has been demonstrated to be a good candidate of hole selective contact for PSCs in virtue of its suitable work function and suitable conduction band position to effectively transport holes and block electrons. In contrast to spiro-OMeTAD that requires multi-step organic synthesis and tedious purification, NiO can be prepared by simple and low-cost synthetic processes. A significant quenching of photo-luminescence emission from NiO/perovskite heterojunction confirms that NiO film effectively extracts holes from perovskite; though the first planar architecture of NiO/CH₃NH₃PbI_3−xCl_x/PCBM obtained a low efficiency (<0.1%) mainly due to very poor perovskite coverage on the NiO surface [100]. Since then, many efforts have been devoted to improving the device efficiency using NiO as a hole transporter. The details are discussed in various device configurations.

4.1.1. Mesoporous NiO-Based n-i-p PSCs

Liu et al. introduced NiO nanoparticles to form a mesoporous NiO space separator between the mesoporous TiO₂ and the carbon counter electrode. A hetero p-n junction formed by NiO/TiO₂ junction easily separates electron and hole flow in opposite directions; hence, lower recombination could occur at this interface. The PSCs were composed of FTO/compact TiO₂/mesoporous TiO₂/mesoporous NiO/carbon, in which the light absorber CH₃NH₃PbI₃ was infiltrated inside the pores of mesoporous TiO₂/mesoporous NiO/carbon via sequential deposition [101]. Application of mesoporous NiO as a HTM effectively suppresses charge recombination and facilitates hole extraction, resulting in a cell with a PCE of 11.4% [102]. Furthermore, the same groups applied highly-crystalline NiO nanosheets as top nanostructured charge transport layers and inserted an additional mesoporous ZrO₂ layer to achieve an n-i-p heterojunction. The ZrO₂ layer was inserted to separate the mesoporous TiO₂ layer and the mesoporous NiO layer to suppress TiO₂/NiO interface charge recombination, leading to an enhancement of charge collection. A full device was fabricated as a structure of FTO/compact TiO₂/mesoporous TiO₂/mesoporous ZrO₂/mesoporous NiO/carbon/ sequentially deposited CH₃NH₃PbI₃, in which the mesoporous layers were prepared by a screen printing technique. The PSCs using all metal oxide semiconductors as a framework achieved a promising energy conversion efficiency of 14.2% [103]. Similar works were reported by Xu et al. who used a doctor-blade technique to fabricate mesoscopic TiO₂/ZrO₂/NiO/carbon structures that exhibited an appreciable PCE of 14.9% [104]. Under a similar configuration, a quadruple layer consisting of mesoscopic TiO₂/Al₂O₃/NiO/carbon was employed as a scaffold and deposited by screen-printing to fabricate mesoscopic PSCs. By optimizing the thickness of each layer, a considerable PCE of 15.03% was achieved for 800 nm-thick mesoporous NiO layer [105]. The above-mentioned mesoscopic devices displayed excellent stability due to the protection provided by the thick carbon counter electrode.

4.1.2. Planar NiO-Based n-i-p PSCs

Nejand et al. used a sputter-deposited NiO_x thin film on the perovskite light absorber to extract the generated holes from the perovskite layer. A nickel electrode was subsequently sputtered to replace the noble Au contact due to its vicinity of work function close to the valance band of perovskite. A planar n-i-p heterojunction PSC of FTO/compact TiO₂/CH₃NH₃PbI_3−xCl_x/NiO_x/Ni was fabricated. The substrate was tilted 45° against the sputtering target for the formation of compact and uniform NiO_x layer covering on the perovskite surface. Also, periodic deposition by controlling the temperature was further employed to lower the substrate temperature without damaging the perovskite layer. The fabricated device initially displayed a low photovoltaic performance. After the strain release of the sputter-deposited NiO_x layer, high hole mobility in the NiO_x film was achieved and led to improve both V_oc and J_sc. Eventually, PSCs employing sputter-deposited NiO_x reached a maximum efficiency of 7.24% and impressive long-term durability of over 2 months. The results indicated that applications of inorganic ETM and HTM for PSCs would effectively hinder the water and oxygen penetration [36].

4.1.3. Mesoporous NiO-Based p-i-n PSCs

The original application of mesoporous NiO in PSCs was reported by Tian et al. [106] and our group nearly at the same time. Tian et al. fabricated mesoporous NiO-based PSCs with an architecture of FTO/compact NiO/mesoporous NiO/sequentially deposited CH₃NH₃PbI₃/PCBM/Al. After controlling the thickness of compact NiO film of 80 nm and that of mesoporous NiO layer of 120 nm, the best device exhibited an efficiency of 1.5% with extremely low FF and J_sc which is caused by high resistance at NiO/perovskite interface [106]. Our group incorporated a mesoporous NiO as a host to adsorb more amount of perovskite for the improvement of light harvesting [107]. A similar device structure was constructed as indium-doped tin oxide (ITO)/compact NiO/mesoporous NiO/sequentially deposited CH₃NH₃PbI₃/PC₆₁BM/bathocuproine (BCP)/Al. The energy level arrangement, in which the VB position of NiO is close to that of perovskite and the lowest unoccupied molecular orbital (LUMO) of PC₆₁BM is nearly identical with the conduction band (CB) edge of perovskite, is suitable for receiving high output voltage and minimizing energy losses for the charge separation process between the absorber and selective contacts. The champion cell delivered a V_oc of 1.04 V, a J_sc of 13.24 mA/cm², and a FF of 69%, leading to an overall efficiency of 9.51%. Compared with our previously reported planar heterojunction junction (PHJ) device [108], the implementation of the mesoscopic NiO layer evidently ameliorates device performance for increased current response in the long wavelength part due to the improvement of light harvesting caused by a thicker perovskite layer. With elimination of a NiO compact layer, Zhu et al. employed an extremely thin mesoporous NiO layer as hole extraction and transport layer for efficient perovskite solar cell that was consisted of FTO/mesoporous NiO/sequentially deposited CH₃NH₃PbI₃/PCBM/Au. A corrugated surface resulted from aggregation of faceted NiO nanocrystals facilitated perovskite formation with good coverage and interconnectivity. By optimizing the thickness of the mesoporous NiO layer of 40 nm, its electron blocking ability was improved to reduce the charge recombination and leakage current. The best performance of NiO-based PSC displayed a V_oc of 0.882 V, a J_sc of 16.27 mA/cm², a FF of 0.635 and a PCE of 9.11% [109].

Furthermore, we introduced a low-temperature sputtered NiO_x film to serve as a pinhole-less compact layer for optimizing charge collection losses in the NiO/perovskite interface. A high quality NiO_x compact layer formed by sputter deposition has benefits in the charge collection, leading to a remarkable improvement in device performance and reproducibility as compared to the PSCs employing solution-processed NiO_x [107]. A full ITO/sputtered-compact NiO/mesoporous NiO/sequentially deposited CH₃NH₃PbI₃/PCBM/BCP/Al device was demonstrated, in which the compact NiO_x layer was deposited by RF magnetron sputtering system in pure Ar gas or a reactive sputtering method under a mixture of Ar and O₂ gas. By optimizing the oxygen doping ration of 10% and the thickness of mesoporous NiO layer, a decent PCE of 11.6% was obtained [110].

As revealed from the transient photo-induced absorption (PIA) spectrum, long-lived holes in the mesoporous NiO have been identified to prove the charge separation occurred at the NiO/perovskite interface. For efficient mesoporous NiO-based PSCs, it is critical to reduce the mesoporous NiO thickness and optimize the NiO/perovskite interface. In addition, replacement of non-injection scaffold is suggested to minimize the interface recombination [35]. Chen et al. reported the implementation of hybrid interfacial layer by incorporating an ultrathin NiO compact layer with a thin mesoporous Al₂O₃ scaffold to fabricate inverted PSCs with a FTO/compact NiO/mesoporous Al₂O₃/CH₃NH₃PbI₃/PCBM/BCP/Ag configuration [111]. The hybrid interfacial layer of NiO/mp-Al₂O₃ significantly minimizes light harvesting and interfacial recombination losses, in which thinner compact NiO layer and highly transparent mp-Al₂O₃ were utilized to prevent light absorption losses. Possible shunt paths between FTO and perovskite film are inhibited by introducing Al₂O₃ nanoparticles on the pinhole of NiO compact layer; in addition, mp-Al₂O₃ layer effectively blocks the shunt paths between NiO and PCBM. Devices using hybrid interfacial layer displayed a high FF of 0.72 with an acceptable PCE of 13.5%.

4.1.4. Planar NiO-Based p-i-n PSCs

For the demonstration of the first planar NiO-based perovskite solar cells, we employed a solution-processed NiO_x thin film as an electrode interlayer on the ITO substrate to realize CH₃NH₃PbI₃/PCBM PHJ hybrid solar cells. As evidence of effective PL quenching, NiO_x was expected to be a potential electrode for extracting hole carrier. To further overcome the coverage issue, a UV-ozone (UVO) treatment of the ITO/NiO_x substrate was performed to improve the surface wetting capability and resulted in an improvement of the perovskite coverage. The work function of NiO_x thin film was further modified to be 5.4 eV for better matching with the VB edge of CH₃NH₃PbI₃. The hybrid cell composed of ITO/spin-coated NiO_x/one-step deposited CH₃NH₃PbI₃/PCBM/BCP/Al was fabricated to deliver an efficiency of 7.8% [108]. Lai et al. proposed oxidized-Ni metal film with variant annealing temperature to realize NiO_x compact layer as a HTM. A full device with an ITO/compact NiO_x/one-step deposited CH₃NH₃PbI₃/PCBM/BCP/Al structure was fabricated. By oxidizing Ni metal with an annealing temperature of 450 °C, NiO_x film shows a smooth surface morphology that facilitates perovskite interconnectivity with pinhole-less surface coverage, leading to the best PCE of 7.75% [112]. The same group further employed Ni/Au bilayer to simultaneously function as HTM and conductive electrode by oxidizing this e-beam-sputtered film in O₂ atmosphere with annealing process. The bifunctional p-type electrode composed of proper thicknesses of Ni (10 nm) and Au (7 nm) under an annealing temperature of 500 °C shows acceptable optical transparent (transmission >60% in the visible spectrum), good electrical conductivity and favorable work function matching the VB of perovskite. A TCO-free PSC was fabricated with a device configuration of glass/Au:NiO_x/gas-assisted deposition CH₃NH₃PbI₃/C₆₀/BCP/Al, delivering a J_sc of 13.04 mA/cm², a V_oc of 1.02 V, a FF of 0.77 and a decent PCE of 10.24%. Improvements on V_OC and FF were attributed to the minimizing energy loss and superior electrical conductivity, as compared their last work [113]. Subbiah et al. used electrodeposited NiO and CuSCN films as HTMs for their FTO/HTM/CH₃NH₃PbI_3−xCl_x/PCBM/Al device, in which the perovskite layer was deposited via coevaporation of PbCl₂ and CH₃NH₃I [114]. The results indicated that the NiO-based PSCs exhibit superior performance than the CuSCN-based PSCs because of lower series resistance and higher shunt resistance. The NiO-based device with UVO treatment displayed a J_sc of 14.2 mA/cm², a V_oc of 0.786 V, a FF of 0.65 and a PCE of 7.26% [115]. Hu et al. fabricated a solution-processed NiO compact layer by spin-coating a nickel acetate methoxy-ethanol solution. Similar structure of ITO/NiO/sequentially deposited CH₃NH₃PbI₃/PCBM/Al with UVO treatment was achieved to present device efficiency of 7.6% [116]. To further improve the quality of the NiO, Cui et al. exploited the reactive magnetron sputtering method to make a compact NiO film for a cell structure of FTO/sputtered NiO/solvent-engineering CH₃NH₃PbI₃/PCBM/BCP/Au. The solvent-engineering method was introduced to form a homogeneous surface coverage and uniform interlayer morphology of perovskite absorber onto the compact layer. With a 240 nm-thick perovskite and effective NiO electrode, the as-fabricated device exhibited an increased PCE of 9.84% [117].

On the whole, NiO-based PSCs generally exhibited a lower FF than the PEDOT-based devices although they displayed a higher V_oc. PSCs applying hybrid NiO_x/PEDOT HTM were further proposed by Park et al. to improve FF. With the addition of a PEDOT covering on a NiO_x-coated substrate, the NiO_x surface morphology was modified to be rough, which benefits increased interfaces of NiO_x/CH₃NH₃PbI₃ heterojunction and reduces interface resistance. A planar p-i-n PSC of ITO/hybrid NiO_x-PEDOT/CH₃NH₃PbI₃/PCBM/Ag with high conversion efficiency of 15.1% was presented by using PEDOT treatment of 1.0% (v/v). As proven by the PL spectra and impedance spectroscopy, the capability of PL quenching and interface charge transfer resistance (R_CT) at the NiO_x/perovskite interface significantly improved with the PEDOT treatment. The experimental results also exhibited neglected-hysteresis of the fabricated inverted PSC device [118]. Li et al. used an extremely thin NiO_x film of 10 nm as a hole extraction layer for a device composed of ITO/spin-coated NiO_x/ CH₃NH₃PbI₃/C₆₀/BCP/Ag. A high-quality perovskite layer was deposited by fast deposited crystallization [119] to form a compact surface coverage and flat interlayer morphology, and the NiO-based planar PSCs delivered a V_oc of 994 ± 15 mV, a J_sc of 20.4 ± 0.7 mA/cm², a FF of 0.669 ± 0.022 and a PCE of 13.6% ± 0.6%. It is surprising that a HTM-free PSCs was further demonstrated to exhibit superior performance compared to NiO_x-based PSCs. In the HTM-free PSCs, perovskite simultaneously functions as the light absorber and hole transporter [120]. With a UVO treatment of ITO substrate, the work function of ITO was improved to a higher value of 5.2 eV that reduced the energy barrier between ITO and perovskite layer. Moreover, UVO treatment effectively removes the residual organic species to lower surface energy, which benefits the deposition of perovskite layer. On the other hand, the perovskite layer deposited on the bare ITO substrate with a nano-sized rough surface morphology improved the perovskite/ITO heterojunction interface. Eventually, the HTM-free PSCs showed a higher J_sc of 21.7 ± 0.5 mA/cm², a higher FF of 0.718 ± 0.015 and a promising PCE of 15.0% ± 0.5% [121].

Recently, significant improvements on NiO-based PSCs were made by doping the NiO thin film or modifying the NiO/perovskite interface heterojunction [98,99,100,101,102]. Kim et al. demonstrated a high-efficiency planar PSC by using solution-processed copper (Cu)-doped NiO_x (Cu:NiO_x) as a HTM [122]. With copper acetate doping in nickel acetate as precursor, a Cu-doped NiO_x was fabricated via the conventional sol-gel route with a sintering process at 500 °C to achieve high crystallinity. The device structure of ITO/Cu:NiO_x/CH₃NH₃PbI₃/PC₆₁BM/C₆₀/Ag was demonstrated with a high PCE of 15.4%, and the improvement was attributed to the enhanced electrical conductivity and charge extraction capability of Cu:NiO_x film, and favorable perovskite crystallization on Cu:NiO_x film. Moreover, the application of wider bandgap Br-doped perovskite solar absorber in Cu:NiO_x-based PSCs was further demonstrated to exhibit a high voltage output of 1.13 V and 1.16 V for CH₃NH₃Pb(I_0.8Br_0.2)₃ and CH₃NH₃Pb(I_0.6Br_0.4)₃ devices, respectively, because of less potential loss between work function of NiO_x and VB of Br-contained perovskite. In the following progress, the same group introduced low-temperature combustion process to fabricate Cu:NiO_x compact layer that was prepared by doping copper nitrate trihydrate into nickel nitrate hexahydrate and acetylacetone as a precursor. A highly crystalline Cu:NiO_x via combustion process at 150 °C exhibited similar work function of −5.3 eV compared to that of sol-gel processed Cu:NiO_x [98]. Moreover, combustion-processed Cu:NiO_x exhibited higher electrical conductivity than that of high-temperature-sintered film by almost two fold. PSCs fabricated with the configuration of ITO/Cu:NiO_x/CH₃NH₃PbI₃/C₆₀/bis-C₆₀/Ag displayed an recorded PCE of 17.74% [123]. Park et al. modified the NiO/perovskite interface heterojunction by employing the pulse laser deposition (PLD) method to fabricate a well-ordered nanostructured NiO thin film with a good optical transparency and preferred orientation of (111) [124]. The (111)-oriented NiO showed a lower resistivity and sheet resistance compared to other growth directions, which have the benefit of effectively extracting hole carriers and preventing recombination. Also, the well-ordered nanostructured NiO provides more contact interface with perovskite to slow the recombination rate during the charge collection. The full device was constructed as ITO/PLD-NiO/CH₃NH₃PbI₃/PCBM/LiF/Al with a dense perovskite prepared by a solvent engineering technique. By introducing a partial oxygen pressure of 200 mTorr and controlling the deposition time, a nanostructured NiO film with controlled thickness was made for high efficiency planar p-i-n heterojunction PSC. The best-performing device exhibited a J_sc of 20.2 mA/cm², a V_oc of 1.06 V and a remarkable efficiency of 17.3% with a very high FF of 0.813.

To further simplify the deposition procedure, Yin et al. employed solution-processed NiO_x film to fabricate an inverted planar PSC which was composed of FTO/NiO_x/CH₃NH₃PbI₃/PCBM/Ag [125]. The thickness of the NiO_x film significantly influences the photovoltaic performance, hysteresis effect and air storage stability of the as-fabricated devices. The thickness-dependent roughness would have impact on the charge collection. With the suitable thickness of 90 nm and nanostructure roughness of 3.835 nm for NiO_x compact layer, a hysteresis-free NiO_x-based PSC was fabricated with a PCE of 14.42%. Moreover, spin-coating NiO_x film via solution process was further applied to flexible PET substrate. Zhang et al. demonstrated flexible PSCs employing a room-temperature formation of NiO_x film to display an impressive PCE of 14.53% for the first time. A PET/ITO/NiO_x/CH₃NH₃PbI₃/C₆₀/bis-C₆₀/Ag structure was achieved, in which the formed NiO_x film is flawless with a nanostructure morphology feature that improves the hole extraction, and reduces the interfacial recombination and monomolecular Shockley-Read-Hall recombination of as-fabricated PSCs. Hysteresis-free with good air and mechanical stability of flexible NiO_x-based PSCs were thus achieved [126].

It is worth noting that the above-mentioned structures use PCBM as an ETM, whose stability and cost are issues of concern. Replacing organic charge extraction layers with inorganic materials would improve the devices’ stability and provide versatile choices for material selection and device design. For the fabrication of durable and cost-effective PSCs, You et al. reported planar PSCs with all-solution-processed metal oxide charge transport layers, whose device structure consisted of ITO/NiO_x/CH₃NH₃PbI₃/ZnO/Al [33]. Favorable band alignment between the NiO_x film and perovskite material, and improved crystallinity of the perovskite film on NiO_x lead to cells with higher V_OC as compared to cells using PEDOT:PSS. On the other hand, ZnO film possesses strong hole-blocking effects originating from its deep VB, resulting in a higher J_SC for the ZnO-based PSCs. The all metal-oxide-based PSCs achieved a V_oc of 1.01 V, J_sc of 21.0 mA/cm² and FF of 76.0%, leading to a remarkable efficiency of 16.1%. With replacement of the aforementioned PCBM with ZnO, a significant enhancement in stability of un-encapsulated cells is achieved with a maintained PCE for 60 days under an ambient environment. By heavily doping metal oxide semiconductor as charge transport layer in planar PSCs, device performance of large active-area (>1 cm²) with high-certified efficiency (>15%) and high stability under light soaking for 1000 h were further achieved [34]. Chen et al. employed heavily p-doped (p⁺) Ni_xMg_1−xO and n-doped (n⁺) TiO_x as HTM and ETM, respectively, in inverted planar PSCs to improve charge transport capability and minimize resistive losses. As confirmed by transient photocurrent and photovoltage decay measurements, faster charge transport through the heavily-doped charge carrier extraction layers facilitate the charge extraction and prevent charge accumulation at the interface of HTM/CH₃NH₃PbI₃/ETM heterojunction. A remarkable efficiency of 18.3% is obtained with a very high FF of 0.83 for a small aperture area of 0.09 cm². So far, all high efficiency PSCs were measured for small areas of 0.1 cm²; however, record efficiencies with cell sizes larger than 1 cm² are recommended by National Renewable Energy Laboratory (NREL) to compete with other solar devices. The PSCs employing highly-doped-metal-oxide charge extraction layers were fabricated with active area >1 cm², and the cell showed a considerable efficiency of 16.2%. Based on the inverted configuration, all metal-oxide-based PSCs display a small hysteresis with different scan directions. We summarize the NiO-based perovskite solar cells with their champion photovoltaic parameters in

Table 2. Summary of photovoltaic characteristics of organometallic lead halide perovskite solar cell (PSC) based on inorganic hole transport material (HTM).

Type 1	Architecture 2	Voc (V)	Jsc (mA/cm2)	FF (%)	η (%) 3	Ref.
NiO
M	FTO/cp-TiO2/mp-TiO2/mp-NiO/Psk/carbon	0.89	18.2	0.71	11.4	[102]
M	FTO/cp-TiO2/mp-TiO2/mp-ZrO2/mp-NiO/Psk/carbon	0.965	20.4	0.72	14.2	[103]
M	FTO/cp-TiO2/mp-TiO2/mp-ZrO2/mp-NiO/Psk/carbon	0.917	21.36	0.76	14.9	[104]
M	FTO/cp-TiO2/mp-TiO2/mp-Al2O3/mp-NiO/Psk/carbon	0.915	21.62	0.76	15.03	[105]
M	FTO/cp-TiO2/mp-TiO2/mp-NiO/Psk/carbon	0.89	18.2	0.71	11.4	[102]
P	FTO/cp-TiO2/Psk/NiOx/Ni	0.77	17.88	0.53	7.28	[36]
M	FTO/cp-NiO/mp-NiO/Psk/PCBM/Al	0.83	4.9	0.35	1.5	[106]
M	ITO/cp-NiO/mp-NiO/Psk/PC61BM/BCP/Al	1.04	13.24	0.69	9.51	[107]
M	FTO/mp-NiO (sol-gel)/PsK/PCBM/Au	0.882	16.27	0.635	9.11	[109]
M	ITO/sputtered-NiO/mp-NiO/Psk/PCBM/BCP/Al	0.96	19.8	0.61	11.6	[110]
M	FTO/NiO/meso-Al2O3/Psk/PCBM/BCP/Ag	1.04	18.0	0.72	13.5	[111]
P	ITO/NiOx/Psk/PCBM/BCP/Al	0.92	12.43	0.68	7.8	[108]
P	ITO/NiOx (Ni-oxidized)/Psk/PCBM/BCP/Al	0.901	13.16	0.65	7.75	[112]
P	(TCO-free) Au:NiOx/Psk/C60/BCP/Al	1.02	13.04	0.77	10.24	[113]
P	FTO/NiO/Psk/PCBM/Al	0.786	14.2	0.65	7.26	[115]
P	ITO/NiO/Psk/PCBM/Al	1.05	15.4	0.48	7.6	[116]
P	FTO/sputtered-NiO/Psk/PCBM/BCP/Au	1.10	15.17	0.59	9.84	[117]
p	FTO/NiOx/Psk/PCBM/Ag	1.09	17.93	0.74	14.42	[125]
P	ITO/Cu:NiOx(sol-gel)/Psk/PC61BM/C60/Ag	1.11 ± 0.01	18.75 ± 0.42	0.72 ± 0.01	14.98 ± 0.33	[122]
P	ITO/Cu:NiOx(combustion)/PsK/C60/bis-C60/Ag	1.05	22.23	0.76	17.74	[123]
P	ITO/PLD-NiO/Psk/PCBM/LiF/Al	1.06	20.2	0.813	17.3	[124]
P	PET/ITO/NiOx/ Psk/C60/Bis-C60/Ag	0.97 ± 0.01	20.55 ± 0.71	0.68 ± 0.03	13.33 ± 0.78	[126]
P	ITO/hybrid NiOx-PEDOT/Psk/PCBM/Ag	1.02 ± 0.006	19.4 ± 0.3	0.70 ± 0.016	13.9 ± 0.4	[118]
P	ITO/NiOx/Psk/C60/BCP/Ag	0.994 ± 0.015	20.4 ± 0.7	0.669 ± 0.022	13.6 ± 0.6	[121]
P	ITO/NiOx/Psk/ZnO/Al	1.01	21.0	0.76	16.1	[33]
P	FTO/NiMgLiO/Psk/PCBM/Ti(Nb)Ox/Ag (active area >1 cm2)	1.072	20.21	0.748	16.2	[34]
CuSCN
M	FTO/cp-TiO2/mp-TiO2/Psk/CuSCN/Au	0.63	14.5	0.53	4.85	[127]
M	FTO/cp-TiO2/mp-TiO2/Sb2S3/Psk/CuSCN/Au	0.57	17.23	0.52	5.12	[128]
M	FTO/cp-TiO2/mp-TiO2/Psk/CuSCN/Au	1.025	17.91	0.57	10.51	[130]
M	FTO/cp-TiO2/mp-TiO2/Psk/CuSCN/Au	1.016	19.7	0.62	12.4	[131]
M	FTO/cp-TiO2/mp-TiO2/Psk/CuSCN/Au	0.96	18.23	0.68	11.96	[132]
P	FTO/cp-TiO2/Psk/CuSCN/Au	0.97	18.42	0.40	7.19	[132]
P	FTO/cp-TiO2/Psk/CuSCN/Au	0.727	14.4	0.62	6.4	[133]
P	FTO/CuSCN/Psk/PCBM/Ag	0.677	8.8	0.63	3.8	[115]
P	ITO/CuSCN/Psk/C60/BCP/Ag	0.97 ± 0.02	21.7 ± 0.4	0.742 ± 0.014	15.6 ± 0.6	[135]
P	ITO/CuSCN/Psk/PC61BM/bis-C60/Ag	1.07 ± 0.01	19.6 ± 0.3	0.74 ± 0.03	15.6	[136]
P	ITO/CuSCN/Psk/PC61BM/bis-C60/semitransparent Ag	1.06 ± 0.02	13.0 ± 0.4	0.73 ± 0.02	10.06	[136]
P	ITO/CuSCN/Psk/PC60BM/LiF/Ag	1.06 ± 0.01	15.76 ± 0.02	0.632 ± 0.052	10.5 ± 0.16	[137]
CuI
M	FTO/cp-TiO2/mp-TiO2/Psk/CuI/Au	0.55	17.8	0.62	6.0	[138]
P	FTO/TiO2/Psk/CuI/graphite	0.78	16.7	0.57	7.5	[140]
P	FTO/CuI/Psk/PCBM/Al	1.04	21.06	0.62	13.58	[141]
Cu2O
P	(Simulation) FTO/TiO2/Psk/Cu2O/Au	1.2	24.75	0.82	24.4	[143]
P	ITO/Cu2O/Psk/PC61BM/Ca/Al	1.07	16.52	0.755	13.35	[144]
P	ITO/Cu2O/Psk/PCBM/Al	0.89	16.52	0.58	8.30	[145]
P	(Simulation) FTO/Cu2O/Psk/PCBM/Al	0.91	20.22	0.74	13.6	[146]
CuO
P	ITO/CuO/Psk/PC61BM/Ca/Al	1.06	15.82	0.725	12.16	[144]

¹ In type, “M” and “P” means “mesoporous” and “planar”; ² “cp” and “mp” indicate “compact film” and “mesoporous film”; ³ PCE (η) = J_sc × V_oc × FF/I₀, where J_sc is the short circuit current density, V_oc is the open circuit voltage, FF is fill factor, and I₀ is the intensity of the incident light; BCP, bathocuproine; FTO, fluorine-doped tin oxide; ITO, indium-doped tin oxide; PCBM, phenyl-C61-butyric acid methyl ester; Psk, perovskite; TCO, transparent conductive oxide.

and efficiency trends in Figure 4.

4.2. CuSCN

As mentioned above, copper thiocyanate (CuSCN) serves as a universal hole-transport interlayer material for numerous optoelectronic applications because it has intrinsic hole-transport (p-type) characteristics with a large bandgap (>3.5 eV) and good optical transparency throughout the visible to near infrared spectrum. This material has been used both in n-i-p heterojunction and p-i-n heterojunction [91,103,104,106,107,108,109,111,112,113].

4.2.1. Meoporous CuSCN-Based n-i-p PSCs

CuSCN was first employed by Ito et al. into mesoporous PSC with a structure composed of FTO/compact TiO₂/mesoporous TiO₂/CH₃NH₃PbI₃/CuSCN/Au [127]. The CH₃NH₃PbI₃ light absorber was fabricated by one-step spin-coating [29] and its thickness is controlled by hot-air drying during spin-coating. With the introduction of doctor-blade processed CuSCN as a HTM and a controlled thickness of perovskite, the first CuSCN-based PSCs achieved a PCE of 4.85%. The experiment also indicates that the CuSCN films limit photovoltaic degradation of moisture-sensitive perovskite material. The same group further inserted Sb₂S₃ at the interface between mesoporous TiO₂ and perovskite layer to facilitate the crystallization of perovskite formation and serve as a passivation layer on TiO₂ to avoid decomposition of the perovskite crystals upon exposure to light [128]. The device showed a slight enhancement in performance to display a PCE of 5.12% and exhibited stability against light soaking without encapsulation. A remarkable improvement under the same configuration was further achieved by the sequential deposition method [129] and preheating substrate before PbI₂ deposition. The efficiency was enhanced to 10.51% because of a pinhole-less perovskite layer which limits the diffusion of CuSCN into the CH₃NH₃PbI₃ active layer [130]. The thickness of the perovskite film, perovskite capping layer, and surface morphology were further controlled with the sequential deposition method. It is noted that loading of PbI₂ upon the mesoporous TiO₂ was carried out by double spin-coating deposition for the purpose of perovskite capping layer formation. With the application of inorganic p-type HTM, CuSCN, made by doctor blading, the CuSCN-based solar cell displayed a promising performance with a J_sc of 19.7 mA/cm², a V_oc of 1.02 V, a FF of 0.62, and a PCE of 12.4% [131]. With addition of a small amount of methyl ammonium iodide (MAI) and DMSO in the PbI₂-DMF precursor, the light absorber of perovskite was made by the modified sequential two-step spin-coating method to form a pinhole-less film. A pinhole-less perovskite film effectively inhibited possible short contact between TiO₂ and CuSCN, resulting in enhancement on photovoltaic performance. A mesoporous PSC with the same structure of FTO/compact TiO₂/mesoporous TiO₂/CH₃NH₃PbI₃/CuSCN/Au displayed a V_oc of 0.96 V, a J_sc of 18.23 mA/cm², a FF of 0.68, and a PCE of 11.96%. A planar structure is fabricated for comparison to show a lower FF of 0.4, leading to a lower efficiency of 7.19% [132].

4.2.2. Planar CuSCN-Based n-i-p PSCs

For the implementation of CuSCN-based planar architecture, Chavhan et al. fabricated a solar cell composed of FTO/compact TiO₂/CH₃NH₃PbI_3−xCl_x/CuSCN/Au. An organometal halide perovskite layer was deposited by spin-coating with a thermal annealing process [133]. The champion solar cell was exhibited for an annealing temperature of 110 °C because of a uniform perovskite morphology, leading to 6.4% power conversion efficiency. A lower V_oc of 727 mV was obtained and it was attributed to non-uniformity and poor surface coverage of CH₃NH₃PbI_3−xCl_x on the compact layer [134].

4.2.3. CuSCN-Based p-i-n PSCs

Subbiah et al. reported the use of CuSCN for planar OPV-like structures consisting of FTO/electrodeposited CuSCN/coevaporation CH₃NH₃PbI_3−xCl_x/PCBM/Ag with a maximum efficiency of 3.8% [115]. When the perovskite film was deposited on the CuSCN film, a considerable PL quenching was observed to prove its effective charge extraction capability. However, without optimizing CuSCN film thickness, high series resistance and low shunt resistance result in poor photovoltaic performance for the CuSCN-based PSCs. Ye et al. used electrodeposited CuSCN film as HTM to fabricate inverted planar PSCs based on structure of ITO/CuSCN/CH₃NH₃PbI₃/C₆₀/BCP/Ag. By adopting a one-step fast deposition-crystallization (FDC) method [119], a high quality perovskite layer was formed on top of the CuSCN film. Lower interface resistance between the perovskite layer and CuSCN leads to pronounced improvements in J_sc and FF from 18.9 ± 1.9 mA/cm² to 21.7 ± 0.4 mA/cm² and 63.2% ± 4.4% to 74.2% ± 1.4%, respectively, as compared with the device prepared by two-step sequential deposition. A significant increase of PCE was achieved from 11.0% ± 1.5% to 15.6% ± 0.6% [135]. To simplify the deposition procedure, a low-temperature processed CuSCN film was spin-coated onto the ITO substrate with nanocrystalline domains, followed by annealing at a low temperature of 100 °C. A planar inverted device of ITO/CuSCN/CH₃NH₃PbI₃/PC₆₁BM/bis-C₆₀/Ag was fabricated with a compact perovskite active layer deposited by the solvent engineering method. By optimizing the thickness of CuSCN of 40 nm, CuSCN-based PSCs achieved a promising efficiency of 16% and presented decent ambient stability. Interestingly, with further reduction of Ag thickness to 20 nm, a semitransparent PSC with an impressive efficiency of >10% was demonstrated [136]. Zhao et al. reported the similar device structure of ITO/spin-coated CuSCN/CH₃NH₃PbI₃/PC₆₀BM/LiF/Ag, in which the perovskite active layer was deposited by the solvent engineering method with a thickness of ~160 nm. A lower short current density of 15.76 ± 0.02 mA/cm² and fill factor of 63.2% ± 0.52% might be attributed to more abrupt perovskite crystalline interface and unsaturated thickness of absorber less than 300 nm, showing a lower PCE of 10.8% as compared with the previous CuSCN-based PSCs [137].

4.3. CuI

Christians et al. reported the first example by using CuI as a HTM in mesoscopic n-i-p PSCs with the best PCE of 6% [138]. The device configuration was constructed as FTO/compact TiO₂/mesoporous TiO₂/CH₃NH₃PbI₃/CuI/Au. Solution-deposited CuI was applied by an automated drop-casting method to form a 1.5 μm thick overlayer on top of the perovskite film. Though over-thick CuI exhibited a higher electrical conductivity than spiro-OMeTAD, CuI-based PSC exhibited a lower V_oc because of a high recombination rate as determined by impedance spectroscopy. Similarly, Sepalage et al. described the planar n-i-p PSCs with a structure of FTO/compact TiO₂/CH₃NH₃PbI₃/CuI/graphite. The planar perovskite layer was deposited using a gas-assisted spin-coating method [139] while the CuI layer and graphite layer were fabricated by doctor blading. Higher photovoltaic performance in V_oc was obtained compared to previously reported mesoscopic CuI-based devices, resulting in a higher efficiency of 7.5%. The superior voltage was caused by a reduced thickness of CuI film (~400 nm); however, it is still lower than the expected V_oc of ~1 V due to rapid recombination at the perovskite/CuI interface. On the other hand, the CuI-based devices displayed an apparent reduced J-V hysteresis under the DSC-like architecture. As regards steady-state current measurement, the CuI-based devices showed a faster response following voltage step. Rapid injection of holes from perovskite into the CuI layer prevents the capacitive current stored in the perovskite material and has positive contribution to diminish the J-V hysteresis [140].

Chen et al. employed solution-processed CuI as a HTM in inverted planar p-i-n PSCs with the architecture of FTO/CuI/CH₃NH₃PbI₃/PCBM/Al [141]. By using a vapor-assisted solution process to fabricate high quality perovskite light absorber [142], the device showed a promising performance with a V_oc of 1.04 V, a J_sc of 21.06 mA/cm², and a PCE of 13.58%. Benefiting from the high transmittance, nanostructure morphology and deep valance band of CuI thin film, CuI-based PSCs exhibit higher V_oc and J_sc than devices using PEDOT:PSS. Similarly, under the inverted architecture, CuI-based p-i-n PSCs showed a negligible hysteresis effect and acceptable ambient stability.

4.4. Cu₂O and CuO

Other well-known p-type semiconductors are Cu-based oxides such as cuprous oxide (Cu₂O) and copper oxide (CuO). They were recently applied as HTM for PSCs to minimize the energy loss because of their low-lying valence bands [119,120,121,122]. The n-i-p heterojunction planar PSC of glass/FTO/TiO₂/CH₃NH₃PbI₃/HTM/Au with NiO, CuI, CuSCN, Spiro-OMeTAD, and Cu₂O as HTM is simulated via a new solar cell simulation software, wxAMPS (the Analysis of Microelectronic and Photonic Structure). The optimized thickness of ETM and HTM are 135–145 nm and 350–450 nm, respectively, and that of the perovskite absorber layer is 350–450 nm. The PCE values with the various HTMs display a ranking of Cu₂O-, CuSCN-, Spiro-OMeTAD-, CuI-, and NiO-based PSC, among which the simulation of Cu₂O-based PSC exhibit the highest efficiency exceeding 24% [143]. The result indicates that it is possible to further increase the performance of PSCs by using copper oxide as HTM.

A planar inverted p-i-n heterojunction architecture of ITO/HTM/CH₃NH₃PbI₃/PC₆₁BM/Ca/Al was experimentally demonstrated by inserting Cu₂O or CuO as HTM. Cu₂O film was fabricated by converting the CuI-spin-coated substrate via aqueous NaOH solution while CuO film was formed by heating Cu₂O film in air [144]. With the addition of NH₄Cl dissolved in CH₃NH₃PbI₃ precursor, it has benefit of making uniform and high coverage perovskite light absorbers on the HTM; thus, more crystalline CH₃NH₃PbI₃ film was formed on Cu₂O and CuO films. The experimental results indicate that Cu₂O-based and CuO-based PSCs display enhanced V_oc of 1.07 and 1.06 V, respectively, and higher J_sc of 16.52 and 15.82 mA/cm², leading to the best PCE of 13.35% and 12.16%, respectively. As compared to the photovoltaic performance of PEDOT:PSS-based PSC (V_oc of 0.95 V, J_sc of 14.82 mA/cm² and PCE of 11.04%), enhanced V_oc resulted from the valance band of Cu₂O and CuO matching with that of the perovskite, and higher crystallinity of CH₃NH₃PbI₃ on Cu₂O and CuO improves the carrier transport to increase J_sc [144]. Chatterjee et al. prepared a high degree of crystallinity Cu₂O as a HTM through a successive ionic layer adsorption and reaction (SILAR) method. A similar device with a structure of ITO/Cu₂O/one-step deposited CH₃NH₃PbI₃/PCBM/Al was demonstrated with an efficiency of 8.23% [145]. Compared to the devices employing NiO or Cu:NiO HTM, Cu₂O-based PSCs exhibited a superior performance by virtue of the higher mobility of Cu₂O, low energy loss determined by scanning tunneling spectroscopy, low Cu₂O/perovskite interface recombination loss, and better perovskite crystallinity on Cu₂O film. Moreover, Wang et al. conducted a theoretical calculation on the Cu₂O/perovskite heterojunction solar cells by using the wxAMPS software. Models of inverted planar PSCs were performed with the configuration of FTO/HTM/CH₃NH₃PbI_3−xCl_x/PCBM/Al, in which NiO_x and Cu₂O were applied as HTM, and the defect states are considered to take part in the interface recombination [116]. An efficiency of 9.88% was first simulated for NiO_x-based device to verify the experimental results with a thickness of 500 nm for the perovskite absorber layer. For high mobility Cu₂O with 10–50 nm thickness, PSCs using Cu₂O with 500 nm-thick perovskite absorber was further simulated to achieve a PCE of above 13% [146]. Photovoltaic parameters of CuSCN-, CuI, Cu₂O-, CuO-based perovskite solar cells are summarized in Table 2 and their efficiency trends are shown in Figure 5.

5. Summary

The versatility of p-type inorganic semiconductors as a HTM, a semiconductor for p-type sensitization, a charge selective layer, etc. for solar cell applications is summarized. As for p-type DSC, the performance is not yet comparable to n-type DSCs. Probably alternatives with deeper VB than NiO as well as with high charge mobility and controlled size and morphology would improve the photovoltage. Not only that, material development of the other components, e.g., sensitizers and electrolytes, and their interfacial engineering in order to mitigate the charge recombination would be another strategy to improve overall performance. Further improvement of p-DSC will be one of keys to achieve high performance tandem solar cells. SSDSC based on p-type inorganic SC has shown a great improvement and the record efficiency of 8.5% obviously proves its potential over organic-based HTMs.

We also comprehensively surveyed the latest progress in PSCs using inorganic p-type semiconductor as HTMs. Briefly, good HTM candidates for efficient PSCs should have the following properties: (1) suitable valance band position to minimize the energy loss; (2) efficient charge transporting and blocking capability; and (3) high carrier mobility. As revealed, diverse inorganic p-type semiconductor having favorable VB, including NiO, CuSCN, CuI, Cu₂O and CuO, have been introduced to demonstrate efficient PSCs with enhanced device stability, among which Cu₂O is a promising candidate for highly efficient PSCs due to its superior hole mobility and deeper valance band. It is noted that choosing a HTM with high transparency throughout the visible and near infrared spectrum minimizes the optical loss for OPV-like architectures. Furthermore, proper control of surface energy will affect the crystallization kinetics and film quality of the perovskite film deposited on top of it. Combined with versatile fabrication processes for producing high quality uniform, dense and pinhole-free perovskite absorbers such as solvent engineering, fast deposition crystallization, vapor deposition, and additives, the device performance is apparently approaching the performance of conventional DSC-like PSC devices. The balance between charge separation, transportation and recombination shall be considered for optimizing the internal surface area and mesoscopic p-type oxide film thickness.

6. Future Trends

For future improvement of device performance, several issues deserve further investigations. For HTM in solid DSCs, pore-filling is one of the major challenges for inorganic p-type SCs. It is critical to form an intimate heterojunction inside the mesoscopic layer. For p-type sensitization DSCs, improving the crystal quality, increasing the energy offset from the potential of redox mediator, increasing the surface area, and inhibiting the interfacial recombination are critical areas. Novel p-type sensitizers are another main area to be explored to optimize energy level matching and charge transfer kinetics. Increasing open circuit voltage could be achieved by choosing oxides with deeper VB in combination with compatible dye HOMO. As for perovskite solar cells, increasing the mobility and conductivity in p-type oxide layer is important for receiving high efficiency PSCs. On the other hand, doping and optical absorption shall be carefully managed for achieve high overall performance. Except for the enhancement of photovoltaic performance, flexibility and long-term stability should be further addressed in the relevant issues. Low temperature procedure for ETM/HTM is required to offer the feasibility on the flexible substrate. MAPbI₃-based perovskite solar cells using solution-processed n-type/p-type metal oxide materials have shown great promises in facile fabrication and ambient stability. Other p-type materials such as spinel, Cu or Cr-based oxide may find their success in PSCs.