Lead-Free Perovskite Narrow-Bandgap Oxide Semiconductors of Rare-Earth Manganates.

Tremendous success has been achieved in photovoltaic (PV) applications, but PV-generated electricity still cannot compete with traditional power in terms of price. Chemically stable and nontoxic all-oxide solar cells made from earth-abundant resources fulfill the requirements for low-cost manufacturing under ambient conditions and thus are promising as the next-generation approach to solar cells. However, the main obstacles to developing all-oxide solar cells are the spectral absorbers. Besides photovoltaics, novel chemically stable, nontoxic, and earth-abundant narrow-bandgap semiconductors are desired for photochemical applications in photodetectors, photoelectrodes, or photocatalysts. Herein, were report novel lead-free perovskite narrow-bandgap rare-earth semiconductors, YMnO3, HoMnO3, ErMnO3, and YbMnO3, which were identified by screening a family of perovskite rare-earth manganates, RMnO3 (R = Y, La, Ce, Pr, Nd, Sm, Gd, Tb, Dy, Ho, Er, and Yb). The sharp edge observed in their absorption spectra indicates the existence of band gaps, further confirmed with laser Raman fluorescence spectra. Good periodic on-off photoelectronic response was observed in 8 of the 12 members (i.e., R = La, Pr, Nd, Sm, Gd, Tb, Dy, and Yb). Among them, YbMnO3 is approved as an n-type semiconductor with a direct band gap near 1.35 eV, whose theoretical Shockley-Queisser efficiency is approximately 33.7% for single-p-n-junction solar cells. This work sheds light on exploring stable oxide semiconductors with a narrow band gap for future applications.

Introduction

Tremendous success has been achieved in photovoltaic (PV) applications, with the total solar energy installation capacity reaching 509.3 GW worldwide in 2018.[1] However, the price of PV-generated electricity in most locations still cannot compete with that of conventional electricity.[2,3] Therefore, novel PV cells are needed to further reduce the cost of PV systems. A higher power conversion efficiency (PCE) is crucial for reducing PV system costs, and great progress has been made on this front. The PCE of emerging perovskite solar cells has soared from 20.1 to 23.3% over the past few years;[4−7] the efficiency of some commercially available solar cell modules is also greater than 20%.[8] Besides the cost of the solar cells, other expenses such as packing and installation costs account for a considerable fraction of the PV system cost. Metal oxides (MOs) are chemically and thermally stable, nontoxic, earth-abundant, and environmentally friendly, thus fulfilling the requirements for low-cost manufacturing under ambient conditions.[2,3] Therefore, solar cells that are entirely based on MOs are very promising for next-generation PV techniques.

Some MOs have been applied in PV cells and modules, but they are mainly used as transparent conducting electrodes. Examples include indium tin oxide (ITO), fluorine-doped tin oxide (FTO), and aluminum-doped zinc oxide (AZO).[2,3] Moreover, studies on historically used narrow-bandgap oxides for PV applications, such as Cu2O, Co2O3, PbO, and ZnO, have been ongoing.[9−11] By employing the latest engineering approaches for fabricating devices, great progress has been made regarding materials such as Cu2O, which is the oldest narrow-bandgap semiconductor and has a direct band gap close to 1.9–2.1 eV.[12−17] Recently, an efficiency of 8.1% was achieved with a MgF2/Al-doped ZnO/Zn0.38Ge0.62O/Cu2O:Na heterojunction solar cell.[17] Based on these findings, we are confident that future research regarding solar cells will evolve toward the ultimate goal of an all-oxide material system. Nevertheless, the key component for developing all-oxide solar cells is searching for novel MO absorbers.[2,9−17]

On the one hand, much effort on utilizing lead-free perovskites, particularly with organic–inorganic hybrid halides,[18,19] for solar cells has been made and additional studies are ongoing;[20−24] on the other hand, full-inorganic rare-earth semiconductors exhibit interesting properties.[25] Rare-earth manganates, with the general formula RMnO3, are a family of interesting compounds that exhibit ferroelectric, ferromagnetic, antiferromagnetic, and colossal magnetoresistance properties.[26] Accordingly, they have been applied in memory devices, sensors, gyrators, and optical devices and have even been used as combustion catalysts.[26,27] RMnO3 crystallizes with a distorted perovskite structure, which can be classified into two groups: the orthorhombic structure for larger rare-earth elements (La–Dy) and the hexagonal structure for smaller rare-earth elements (Sc, Y, Ho–Lu).[26] Recently, RMnO3 has attracted attention for PV applications. Jang et al.[28] examined the switchable PV effect in YMnO3 and LuMnO3 thin films and observed PCE values 1–3 orders of magnitude greater than those of the classic ferroelectric photovoltaics (FPVs), undoped Pb(Zr,Ti)O3 and BiFeO3, under standard AM 1.5 G illumination. Huang et al.[29] predicted the strong light absorption of TbMnO3 in the solar-spectrum range using first-principles methods based on the density functional theory (DFT), which theoretically results in the maximum solar conversion efficiency of up to 33%.

FPVs are an interesting research field in solar cells.[30−32] The non-centrosymmetry of ferroelectric materials that possess spontaneous electric polarization provides an inherent force to drive exciton separation and carrier extraction, differing from the photogenerated electron–hole pairs separated by the built-in electric field inside traditional p–n-junction solar cells, and the open-circuit voltages could be a few orders of magnitude greater than the band gap of the ferroelectric materials. Recently, a PCE of 8.1% under AM 1.5 G irradiation has been achieved with ferroelectric Bi2FeCrO6 solar cells.[28]

Besides photovoltaics, novel chemically stable and nontoxic narrow-bandgap semiconductors are specifically needed in a wide variety of photoelectrochemical applications; for example, they are used as photoelectrodes in photocatalysts for solar water-splitting,[33−35] photodetectors,[36] and photoelectronic devices.[37] Aiming at these applications, a family of perovskite rare-earth manganates, RMnO3 (R = Y, La, Ce, Pr, Nd, Sm, Gd, Tb, Dy, Ho, Er, and Yb), is examined in this work.

Results and Discussion

To explore the synthesis of RMnO3 powders, the solid-state reaction was performed at 1000, 1200, and 1300 °C for 4 h. As indicated by the X-ray diffraction (XRD) patterns shown in Figure S1, the products synthesized at 1000 °C in ambient air mainly consisted of rare-earth oxides. As the temperature was increased to 1200 °C, the characteristic diffraction peaks of rare-earth oxides decreased and peaks corresponding to rare-earth manganates as the main phases appeared, as shown in Figure S1. As the temperature was further increased to 1300 °C, the minor residual rare-earth oxides PrO2, Nd2O3, Sm2O3, and Er2O3 that had not been consumed in reactions could still be discriminated. Considerable amounts of Y2O3 and Er2O3 were present as residual starting materials in the nominal YMnO3 and ErMnO3 compounds, respectively. Moreover, an anomaly was observed for the LaMnO3 and GdMnO3 phases, whose content increased as the temperature was increased from 1000 to 1200 °C but decreased again as the temperature was increased from 1200 to 1300 °C due to the variation in the La2O3 and Gd2O3 contents. However, when CeO2 and MnO2 were reacted, the XRD patterns of the products indicated that the starting material CeO2 was primarily present and minor Mn3O4 was formed as the temperature increased to 1200 or 1300 °C. For more details about the dependence of the structural transformation on temperature, please refer to Figure S1.

Figure presents XRD patterns of the 12 RMnO3 powders synthesized at 1200 °C for 4 h in an air atmosphere, in which two sets of diffraction patterns could be discriminated, with the exception of the CeO2 raw material. Namely, the manganates of La, Pr, Nd, Sm, Gd, Tb, and Dy crystallize in orthorhombic systems, whereas the manganates of Y, Ho, Er, and Yb crystallize in hexagonal systems. A significant difference between the orthorhombic and hexagonal systems is the 002 plane at a 2θ of approximately 15.5°, which is absent in the former but present in the latter, as marked in Figure . A comparison of the XRD patterns of the YbMnO3 powder and pellet sample, displayed in Figure S2, shows that the minor residual Yb2O3 that presents in the powder sample synthesized at 1200 °C for 4 h disappears in the pellet by extending the reaction to 24 h and with the temperature enhanced to 1350 °C. Table summarizes the crystal system, space group, and cell parameters of the RMnO3 samples. The theoretically calculated band gaps, retrieved from the database of Citrine Informatics,[38] of RMnO3 are also summarized in Table .

Figure 1

XRD patterns of RMnO3 (R = Ce, La, Pr, Nd, Sm, Gd, Tb, Dy, Y, Ho, Er, Yb) powders synthesized by a solid-state reaction at 1200 °C for 4 h in ambient air, in which the orthorhombic (La, Pr, Nd, Sm, Gd)MnO3 and the nominal-composition CeMnO3 with CeO2 indexed are presented in (a) and the orthorhombic (Tb, Dy)MnO3 and the hexagonal (Y, Ho, Er, Yb) MO3 systems are presented in (b).

Table 1

Summary of the Phase Data, Crystal Lattice Parameters, and Theoretical Band Gaps of 12 RMnO3 Compounds

formula	origin code	crystal system	space group	cell parameters (Å)	band gap (eV)a
YMnO3	ICSD 73361	hexagonal	P63/mmc (194)	a = 3.6100, c = 11.3900	0
LaMnO3	ICSD 51653	orthorhombic	Pbnm (62)	a = 5.5006(5), b = 7.7744(8), c = 5.5253(7)	0
CeMnO3	ref (39)b	orthorhombic	Pbnm (62)	a = 5.537, b = 5.557, c = 7.812	0
PrMnO3	COD 27962	orthorhombic	Pbnm (62)	a = 5.5450, b = 5.7870, c = 7.5750	0
NdMnO3	ICSD 53214	orthorhombic	Pbnm (62)	a = 5.4168(5), b = 5.8518(5), c = 7.5479(7)	0
SmMnO3	ICSD 57391	orthorhombic	Pbnm (62)	a = 5.369(1), b = 5.866(1), c = 7.484(1)	0
GdMnO3	ICSD 57393	orthorhombic	Pbnm (62)	a = 5.3160(1), b = 5.8686(1), c = 7.4252(1)	0.24/0
TbMnO3	ICSD 57394	orthorhombic	Pbnm (62)	a = 5.301(1), b = 5.847(1), c = 7.401(1)	0.39
DyMnO3	PDF 25-0330	orthorhombic	Pbnm (62)	a = 5.272, b = 5.795, c = 7.38	0.4/0
HoMnO3	ICSD 92838	hexagonal	P63cm (185)	a = 6.1413(1), c = 11.4122(3)	0.41/0
ErMnO3	ICSD 80583	hexagonal	P63cm (185)	a = 6.1121(5), c = 11.4200(14)	0.41
YbMnO3	ICSD 60749	hexagonal	P63cm (185)	a = 6.073(1), c = 11.349(3)	0.82/0

The band gaps of RMnO3 compounds were retrieved from the database of Citrine Informatics.[38]

Cell parameters of CeMnO3 were obtained from ref (39).

To explore the potential of RMnO3 for PV applications, the photon response was examined. Figure a,b shows the plot of the illuminated open-circuit photovoltage (VOC) and short-circuit photocurrent (JSC), respectively, of LaMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, DyMnO3, and YbMnO3 pellets that were sintered at 1350 °C for 24 h in an Ar atmosphere with the light sequentially switched on and off. In Figure , both VOC and JSC increase nearly linearly with an increase in the radiation intensity of the solar simulator from 0.1 to 1.1 sun (AM 1.5 G). Among the RMnO3 samples, the largest voltaic response belongs to YbMnO3 while LaMnO3 exhibits the largest current. The observed photoresponse during the simultaneous heating process confirms that the PV effect is caused by photon-generated carriers, as shown in Figure S3. Interestingly, the direction of the voltage and current measured in YbMnO3 is different from that measured in LaMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, and DyMnO3. It is possible that the carriers in YbMnO3 are different from those in the other compounds. The electrochemical workstation adopted in our measurement shuts off automatically if the forward voltage is greater than 13 V, thus terminating the measurements of the photon-excited voltage of YMnO3, CeMnO3, ErMnO3, and HoMnO3, as observed in Figure S4a. When the voltage is too high, the photogenerated current is accordingly very weak, as shown in Figure S4b. Therefore, the photogenerated voltage may be very high, or it is possible that these compounds do not conduct at all.

Figure 2

Photon response of the LaMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, DyMnO3, and YbMnO3 pellets fired at 1350 °C for 24 h in an Ar atmosphere.

Absorption, including the wavelength range and intensity, is the key factor determining whether a material is an applicable absorber for solar cells. Herein, the absorption spectra of the samples were collected to obtain this information. Figure a,b presents UV–vis–near-infrared (NIR) absorption spectra of the RMnO3 powders synthesized at 1200 °C. No sharp edge is observed in the absorption spectra of orthorhombic La–Dy manganates displayed in Figure a, which is in accordance with the calculated value of Eg of approximately 0, as summarized in Table . This result indicates that there is no band gap in these compounds. By contrast, absorption bands with sharp edges were observed in the spectra of the hexagonal Y and Ho–Yb manganates in Figure b. The cutoff wavelength of the absorption spectra at approximately 1000 nm indicates that the band gap is close to 1.24 eV. The absorption spectra of the 12 samples each synthesized at 1000, 1200, and 1300 °C, respectively, in ambient air are displayed in Figure S5. Figure S6a presents the absorption spectra of the 12 samples synthesized at 1200 °C, by plotting the spectra in Figure a,b together under the same horizontal and vertical scales. Besides, we should bear in mind that the concave region at 1800–2100 nm in Figures and S5–S7 for all absorption spectra is caused by the Shimadzu UV-3600 spectrophotometer adopted here rather than by the intrinsic absorption of samples.

Figure 3

Absorption spectra, band gaps, and fluorescence spectra. (a, b) Absorption spectra of RMnO3 powders; (c) Tauc plots of the absorption spectra of YMnO3, HoMnO3, ErMnO3, and YbMnO3 using the relationship αhν ∝ (Eg – hν)2 for direct transition, respectively; (d) absorption spectrum of YbMnO3 compared with those of Yb2O3, MnO, Mn2O3, and MnO2; (e) fluorescence spectra of YMnO3, HoMnO3, ErMnO3, and YbMnO3; (f) band-edge positions of YMnO3, HoMnO3, ErMnO3, and YbMnO3 with respect to the vacuum energy level.

The band gap can be further evaluated based on Tauc plots using the relationship αhν ∝ (Eg – hν)1/, where α is the coefficient of optical absorption, hν is the energy of photons, Eg is the band gap, and n = 2 for direct transitions and n = 1/2 for indirect transitions;[40] these plots indicate that a direct band gap near 1.35 eV (Figure c) and an indirect band gap near 1.27 eV (Figure S6b) were obtained for these four samples. After data transformation, however, no tangential line could be discriminated in Figure S6b. On comparing the spectral configuration of Figure c with that of Figure S6b, we can come to the conclusion that hexagonal YMnO3, HoMnO3, ErMnO3, and YbMnO3 have direct band gaps. Moreover, a band gap near 1.35 eV is consistent with the value of approximately 1.4 eV for hexagonal TbMnO3 theoretically predicted by Huang et al.[29] It should be noted that the orthorhombic TbMnO3, rather than the hexagonal structure model of TbMnO3 adopted by Huang et al.,[29] was obtained in this work. For Tauc plots of the absorption spectra of orthorhombic LaMnO3, CeMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, and DyMnO3 obtained using the relationships αhν ∝ (Eg – hν)2 and αhν ∝ (Eg – hν)1/2, please refer to Figure S6c,d, in which the smooth decrease in the intensity with the photon energy further suggests that there are no band gaps in these compounds. In contrast to the band gaps of the inorganic lead halide perovskite CsPbBr3 (2.21 eV),[41] the organic–inorganic hybrid lead-free halide perovskite (C6H5NH3)BiI4 (2.14 eV),[19] and the phenethylammonium (PEA) bismuth halides (PEA)3Bi2I9/(PEA)3Bi2Br9/(PEA)3Bi2Cl9 (2.23/2.66/3.28 eV, respectively),[42] the band gap of hexagonal YMnO3, HoMnO3, ErMnO3, and YbMnO3 of about 1.35 eV is closer to that of methylammonium (CH3NH3+ or MA) lead iodide MAPbI3 (1.64 eV) and formamidinium (NH2CH=NH2+ or FA) FAPbI3 (1.55 eV).[43] Among all of the available published data,[41−44] the band gap of hexagonal YMnO3, HoMnO3, ErMnO3, and YbMnO3 at 1.35 eV is the most proximate to the ideal bandgap energy of 1.34 eV for the maximum power conversion efficiency of single-junction solar cells, predicted by the Shockley–Quesser limit.[45,46] The theoretical maximum limit of the Shockley–Queisser efficiency corresponding to the band gap of 1.35 eV is approximately 33.7% for a single-p–n-junction solar cell.[45,46]

To reveal the mechanism of absorption, the absorption spectrum of YbMnO3 was compared with those of MnO2, Mn2O3, MnO, and Yb2O3. As shown in Figure d, the sharp edge observed in the absorption spectrum of YbMnO3 indicates the existence of a band gap; by contrast, the absorption spectra of MnO2 and Mn2O3 are spread over a wide range without absorption edges due to the oxidation along with the increased number of electrons around O, resulting in a lack of an obvious gap between the valence band (VB) and the conduction band (CB) of MnO2 and Mn2O3 (namely, Eg ≤ 0). With more O atoms deprived of electrons, a sharp edge appears in the absorption spectrum of MnO. Moreover, some fine structures are observed in the absorption spectrum of MnO, which is in accordance with ref (47). We can conclude from the differences among the absorption spectra of MnO2, Mn2O3, MnO, and YbMnO3 that Yb plays a significant role in localizing electrons, equivalent to withdrawing electrons from the O atoms in manganese oxides. This behavior is because the Yb 4f orbitals hybridize strongly with the O 2p orbitals in the ground state, causing the Yb 4f orbitals to be filled with electrons while generating ligand holes in the O 2p orbitals.[48]

Distinguished from the spectral configuration, the absorption spectra of LaMnO3, PrMnO3, and NdMnO3 in Figure a are similar to that of MnO2 in Figure d, the absorption spectra of SmMnO3, GdMnO3, TbMnO3, and DyMnO3 in Figure a are similar to that of Mn2O3 in Figure d, whereas the absorption spectra of YMnO3, HoMnO3, ErMnO3, and YbMnO3 in Figure b are not similar to MnO2, Mn2O3, or MnO in Figure d. It is easy to imagine that the Mn may exist in variable chemical states in these RMnO3 compounds. Arguably, the absorption must be related to the crystal structure. As can be found by combining Figure a,b with Figure , the La–Dy manganates that crystallize in orthorhombic systems have no band gap, whereas the manganates of Y, Ho, Er, and Yb that crystallize in hexagonal systems have a band gap.

In Figure d, the comparison of the absorption spectrum of YbMnO3 with that of Yb2O3 shows that on the one hand, a minor peak at approximately 967 nm of YbMnO3 is consistent with the absorption of Yb2O3, and on the other hand, the absorption spectrum of YbMnO3 differs significantly from that of Yb2O3. The characteristic absorption of Yb3+ at approximately 967 nm, corresponding to the 2F7/2–2F5/2 transition,[49] is observed at the foot of the absorption edge of YbMnO3 in Figure b, suggesting that the interval energy between 2F7/2 and 2F5/2 is similar to but smaller than the band gap of YbMnO3. Therefore, the energy level of the excited state 2F7/2 must be located in the band gap. In addition, the typical 5I8–5I6 and 5I8–5I7 transitions of Er3+ at 1150 and 1950 nm, respectively, are observed in the absorption profile of ErMnO3, and the 4I15/2–4I13/2 transition of Ho3+ at 1500 nm is observed in the absorption profile of HoMnO3.[50] However, their interval energies are far smaller than the band gaps of ErMnO3 and HoMnO3. In Figure b, five bands located at approximately 830, 660, 550, 400, and 310 nm, marked as A, B, C, D, and E, respectively, could be distinguished. Clearly, band A differs from all of the absorption bands of MnO2, Mn2O3, MnO, and Yb2O3. Therefore, the low-energy absorption edge of band A, that is, the minimum of the VB, must be caused by the synergetic effect among Yb, Mn, and O atoms rather than by the individual effects of O–Mn and O–Yb or electron transitions within variant levels of the Mn or Yb atoms.

The existence of band gaps in hexagonal YMnO3, HoMnO3, ErMnO3, and YbMnO3 is further confirmed by laser fluorescence spectroscopy. As shown in Figure e, two emission bands, one located at 760 nm (H center) and the other located at approximately 878 nm (low center), were observed under an excitation of 633 nm. As shown in Figure S7, the overlap of the absorption spectra of YMnO3, HoMnO3, ErMnO3, and YbMnO3 with their fluorescence spectra confirms that these compounds have semiconductor band structures (BSs). Regarding the L-center emission, with an increase in the atomic number in the order Y, Ho, Er, and Yb, the peak wavelength practically does not change, but the peak intensity increases drastically. In addition, the H-center emission peak shifts toward lower energy with an increase in the atomic number. We believe that the L-center emission is irradiated by electrons returning from the bottom of the CB to the hole in the VB and that the H-center emission is irradiated by electrons returning from the excited states to the ground states of Mn 3d. Since the absorption edges of YbMnO3, Er MnO3, Ho MnO3, and Y MnO3 are nearly the same, there is no variation in the L-center emission peak. However, the splitting of the Mn 3d orbital depends intensively on the coordinated ligands. Due to the lanthanide radius contraction, the emission peak of the H center shifts toward lower energy with an increase in the atomic number.

The surface potential is a key parameter for matching electrodes for solar cells. The measured work function values for YMnO3, HoMnO3, ErMnO3, and YbMnO3 are approximately −4.87, −4.69, −4.36, and −4.71 eV, respectively. According to ultraviolet photoelectron spectroscopy (UPS) VB spectra, the difference between the Fermi level (Ef) and the uppermost valence band (Ev), that is, Ef – Ev, is about 0.86 eV (as can be referred to in Figure c). By combining the surface potential values with the band gaps estimated in Figure c, the band and band-edge positions of YMnO3, HoMnO3, ErMnO3, and YbMnO3 can be described in Figure f.

Figure 6

Electronic band structure, VB spectrum, and electron transitions. (a, b) BS and DOS of YbMnO3 around the Fermi level; (c) VB spectra of YbMnO3 and ErMnO3; (d) mechanism of photoexcited charge carriers.

A Hall effect experiment was performed to examine the semiconductor properties of YMnO3, HoMnO3, ErMnO3, and YbMnO3 for use in solar cells. The results show that HoMnO3 and ErMnO3 are p-type semiconductors, while YMnO3 and YbMnO3 are n-type semiconductors. As shown in Table , the carrier concentration is in the range of 1011–1012 cm–3, the resistivity is in the range of 106–107 Ω·cm, and the mobility is in the range of 0.1464–26.92 cm2·V–1·s–1. The carrier concentration in the range of 1011–1012 cm–3 and the mobilities in the range of 0.1464–26.92 cm2·V–1·s–1 of hexagonal YMnO3, HoMnO3, ErMnO3, and YbMnO3 are on par with the experimental parameters (108–109 cm–3 for carrier concentration; 11–52 cm2·V–1·s–1 for mobility) of CsPbBr3 single crystals reported by Bakr[41] and the theoretical data of mobility (1–10 cm2·V–1·s–1) in hybrid halide perovskites calculated by Motta.[51] However, the carrier concentration of YMnO3, HoMnO3, ErMnO3, and YbMnO3 is far below the values of about 1016–1018 cm–3 reported in MAPbI3 polycrystalline thin films,[52] about 1012–1015 cm–3 in the electron-selective layer of SnO2 quantum dots films,[53] and 1018–1019 cm–3 in the hole-transport layer of NiO- and Cu-doped NiO thin films.[54] With respect to the conductivity in the order of 10–4 S cm–1 as reported in the lead-free perovskite (C6H5NH3)BiI4[19] and in SnO2 quantum dot films,[53] the resistivity of hexagonal YMnO3, HoMnO3, ErMnO3, and YbMnO3 in the range of 106–107 Ω·cm is too high. Nevertheless, it should be noted that pellets instead of high-quality thin films were adopted in this work for measurements. From scanning electron microscopy (SEM) images of the pellets shown in Figure S8, we observe that there are many grain boundaries, pores, and tiny cracks in the pellets of YMnO3, HoMnO3, ErMnO3, and YbMnO3. These defects will increase the resistivity and decrease the carrier mobility. To avoid defects, high-quality thin films should be adopted to fabricate solar cells. More work along these lines is underway in our group.

Table 2

Semiconductor Parameters of Volume Resistivity, Volume Hall Coefficient, Bulk Carrier Concentration, and Mobilitya

	resistivity (Ω·cm)	Hall coefficient (cm3·C–1)	carrier concentration (cm–3)	mobility (cm2·V–1·s–1)
YMnO3	2.669 × 107	–3.1589 × 108	1.976 × 1010	11.83
HoMnO3	9.4739 × 106	1.387 × 106	4.501 × 1012	0.1464
ErMnO3	6.0439 × 106	1.627 × 108	3.836 × 1010	26.92
YbMnO3	2.7119 × 106	–2.139 × 107	2.918 × 1011	7.890

To eliminate surface effects, the surface of the pellets was polished using abrasive paper before carrying out the Hall effect measurements. Thus, only volume (or bulk) parameters, rather than surface parameters, are listed, that is, the volume resistivity, volume Hall coefficient, and bulk carrier concentration.

The absorption of incident light by a compound is determined by its composition, the chemical state of each element, and, ultimately, the electron transitions occurring in the ions. To reveal the mechanism and the compositions, chemical states of RMnO3 pellets were investigated with X-ray photoelectron spectroscopy (XPS) analysis by choosing YbMnO3 as a representative example. All expected elements were identified from the XPS survey, as shown in Figure a. The elemental ratio of Yb to Mn was found to be ∼1:1. Five peaks at 184.6, 188.1, 192.4, 198.8, and 205.9 eV in Figure b, which are denoted a, b, c, d, and e, respectively, are fitted from the Yb 4d XPS spectrum, and they are attributed to Yb3+.[55−59] The one-shoulder neighbor to Yb3+ at 181.9 eV may be caused by Yb2+, but this peak is by far very weak. By combining this spectrum with the Yb 5p XPS spectrum shown in Figure c, we can conclude that a small amount of Yb2+ exists in YbMnO3. Figure c shows that the Yb in YbMnO3 exhibits spin-orbital splitting of Yb 5p with 5p1/2 and 5p3/2 peaks at 32.6 and 26.1 eV, respectively, corresponding to Yb3+. In addition, one minor peak at approximately 20.9 eV in Figure c is ascribed to Yb2+ and the other peak of Yb2+ in the region of 23–28 eV possibly is submerged by the strong peak of Yb3+.[55] Analysis of the magnitude of peak splitting is an approach to determine the oxidation state. Typically, the ΔE of Mn 3s for MnO (Mn2+) is 6.0 eV, that of Mn2O3 (Mn3+) is ≥5.3 eV, and that of MnO2 (Mn4+) is 4.7 eV. However, the magnitude of Mn 3s splitting was found to be ∼5.1 eV in Figure d, indicating that the main oxidation state of Mn in YbMnO3 is +3, accompanied by a few compounds in the +4 oxidation state. From the point of view of charge balance, minor Yb2+ might exist to balance with Mn4+ for three O2– ions in YbMnO3 due to the existence of Mn4+. The Mn 2p XPS spectrum was also acquired. As shown in Figure S9, the spin-orbital splitting of Mn 2p with 2p1/2 and 2p3/2 peaks at 653.4 and 642.1 eV, respectively, was observed, but the MnO satellite (∼647 eV) feature was not. This conclusion further confirms the existence of +3 as the main oxidation state of Mn and excludes the +2 oxidation state.

Figure 4

XPS spectra of the YbMnO3 pellet. (a) Survey scan of the element binding energies; (b) Yb 4d binding energies; (c) Yb 5p binding energies; (d) Mn 3s binding energies.

YbMnO3 crystallizes in a hexagonal system with space group P63cm (185) and cell parameters of a = 6.073(1) and c = 11.349(3). In the crystal lattice of YbMnO3, there are two Yb sites and one Mn site. Yb1 and Yb2 are both sevenfold-coordinated, as shown in Figure a. Mn is coordinated with five neighboring O atoms, which build a pentahedron around Mn, as shown in Figure b. As Mn is fivefold-coordinated, the 3d orbital will split into high-energy singlet, ag (3z2 – r2), and low-energy parallel doublet, eg (xy/x2 – y2 and yz/zx), states.[29] The projection of the unit cell along the b and c axes is displayed in Figure c,d, respectively. Therefore, the three-dimensional structure of YbMnO3 could be considered staggered by the layers of Mn–5O pentahedra and Yb ions (as Figure c).

Figure 5

Three-dimensional crystal structure of YbMnO3 and atom coordination. (a, b) Coordination around the Yb and Mn centers, respectively; (c, d) three-dimensional crystal structure of YbMnO3 viewed along the b and c axes, respectively.

Based on this structural model, the electronic band structure (BS), density of states (DOS), and partial DOS (PDOS) of YbMnO3 were calculated using a virtual cell approximation based on first principles. The calculated band gap is 0 (Figure a), which is in accordance with the data summarized in Table but less than the aforementioned experimental value. It is common for theoretically calculated values to be less than the experimental band gap due to the underestimation of band gaps in DFT.[60] However, the calculations can still provide useful information about the BS. The DOS and PDOS in Figure b clearly show that the VB mainly consists of the O 2p and Yb 4f orbitals and a small part of the Mn 3d orbitals, whereas the CB mainly consists of the Mn 3d orbitals. However, regarding the valence band maximum (VBM), Figure b shows that the uppermost edge comprises Yb 4f, the next one is O 2p, and the third one is Mn 3d. To verify the theoretical predictions, the VB spectrum of YbMnO3 was measured and compared with the VB spectrum of ErMnO3 to reveal the effects of the Yb and Mn elements on the BS. In Figure c, two main peaks in regions located at 4–9 and 9–12 eV are assigned to the multiplet structures of Yb3+, which arise from the resonant photoemission in the 4d–4f excitation region.[55−58] A bar diagram showing the multiplet structures of Yb3+ calculated by Cox and Schmidt-May can be found in refs (57) and (58) consistent with the resonant photoemission of the Yb 4f states presented in Figure c. Upon replacing Yb with Er, the VB spectrum changes completely within 4–12 eV, but the uppermost VB (i.e., VBM) in the region of 0–4 eV does not, suggesting that the VBM should consist of O 2p or Mn 3d orbitals rather than Yb 4f orbitals. The comparison of the VB spectrum of YbMnO3 with that of ErMnO3 confirms not only that the VB of YbMnO3 mainly consists of Yb 4f orbitals, in accordance with the PDOS presented in Figure b, but also the contribution of O 2p or Mn 3d orbitals to the VBM. The difference in the absorption spectrum of YbMnO3 from those of MnO2 and Mn2O3, as displayed in Figure e, reveals that the hybridized states of O 2p–Yb 4f and O 2p–Mn 3d play a significant role in shaping the BS. Based on the available data from XPS,[61] the smallest binding energies for the elements Yb and Mn are 2.0 eV for Yb 4f and 2.8 eV for Mn 3d, respectively, which are consistent with the lower edge of the d and f orbitals in Figure b. The smallest binding energy of O is approximately 7.0 eV for O 2p.[61] Nevertheless, the O 2p states spread throughout the VB,[62] which will overlap with the Mn 3d and Yb 4f states due to hybridization.[48] Thus, the BS of YbMnO3 could be described as shown in Figure S10, and the mechanisms of the absorption and fluorescence spectra displayed in Figure b,e could be well explained using this scheme. In Figure b, absorption band A is attributed to the charge transfer from O 2p to Mn 3d; bands B and C are attributed to electron transitions from the bonding states xy and x2 – y2+, respectively, to the antibonding 3z2 – r2 state of Mn 3d; and bands D and E are attributed to electron transitions from the bonding states yz and zx, respectively, to the antibonding 3z2 – r2 state of Mn 3d. Upon excitation with a 633 nm laser light, as shown in Figure e, electrons from the Mn 3d (xy, x2 – y2+) states were pumped to the CB, which comprises the 3z2 – r2 state of Mn 3d. With electrons returning to the VB and the (xy, x2 – y2+) states of Mn 3d, light with peak wavelengths at 878 nm (L center) and 760 nm (H center) was emitted, respectively. The radiance of YMnO3, HoMnO3, and ErMnO3, which peaked at 878 nm, was much weaker than that of YbMnO3 due to the strong hybridization of Yb 4f with O 2p.[48] In addition, the high-energy H-center emission peak shifts from 760 nm to higher energy (658 nm) with an increase in the atomic number from Y, Ho, and Er to Yb, possibly caused by the effect of the lanthanide radius contraction with the change in the crystal field on Mn 3d orbital splitting.

Based on the above analyses, we reason out that the generation of charge carriers to produce the photon response shown in Figure a,b is caused by the transition from the occupied mixed states of O 2p, Yb 4f and Mn 3d to unoccupied 3d3 states.[29] Intrinsically, this transition could be considered the Mn 3d–3d transition, but the ground state of Mn 3d hybridizes with a considerable amount of the high-energy Yb 4f state through O 2p. The electron transitions that are responsible for the generation of charge carriers are illustrated in Figure d.

Finally, we provide a short discussion. To achieve a high PCE, the photogenerated charge carriers, that is, excitons, must be able to move over long distances with long lifetimes before annihilation. Accordingly, the nuclei of atoms must exert a weak force on the charge carriers, the outer s, p, d, or f orbitals of which should be fully filled by electrons. Moreover, the outermost electrons should be readily excited to form excitons. According to these standards, Pb is an excellent element for PV absorbers due to its [Xe]6s26p2 configuration, and its efficacy has been demonstrated in organohalide–perovskite solar cells.[4−7] Yb has an electronic configuration very similar to that of Pb, as shown in Figure S11. Thus, replacing Pb with Yb promisingly opens a new route toward developing narrow-bandgap semiconductors for all-oxide solar cells.

Conclusions

Four novel lead-free perovskite narrow-bandgap semiconductors, YMnO3, HoMnO3, ErMnO3, and YbMnO3, were screened from a family of rare-earth manganates RMnO3 (R = Y, La, Ce, Pr, Nd, Sm, Gd, Tb, Ho, Er, and Yb). The results demonstrate that the hexagonal manganates of YMnO3, HoMnO3, ErMnO3, and YbMnO3 have narrow band gaps, whereas the orthorhombic manganates of LaMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, DyMnO3, and YbMnO3 have no band gaps. Through the solid-state reactions between rare-earth oxides and MnO2 powder, the orthorhombic manganates of La, Pr, Nd, Sm, Gd, and Tb and the hexagonal manganates of Y, Ho, Er, and Yb were obtained, but CeMnO3 was not obtained. Eight of the 12 members, that is, LaMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, DyMnO3, and YbMnO3, exhibited a photon response. Nevertheless, no sharp onset was observed from the absorption spectra of orthorhombic LaMnO3, PrMnO3, NdMnO3, SmMnO3, GdMnO3, TbMnO3, and DyMnO3. The hexagonal YMnO3 and YbMnO3 are n-type semiconductors, whereas HoMnO3 and ErMnO3 are p-type semiconductors. The hexagonal manganates of Y, Ho, Er, and Yb have similar band gaps; yet, the surface potentials of YMnO3, HoMnO3, ErMnO3, and YbMnO3 are approximately −4.87, −4.69, −4.36, and −4.71 eV, respectively. Among these family members, YbMnO3 has a direct band gap of approximately 1.35 eV, whose theoretical Shockley–Queisser efficiency is approximately 33.7% for single-p–n-junction solar cells, exhibiting excellent potential as an absorber for next-generation all-oxide solar cells. The top VB of YbMnO3 consists of the Yb 4f orbital, which hybridizes with Mn 3d through O 2p, whereas the bottom of the CB consists of the Mn 3d orbital. The charge carriers that generate the PV effect are mainly produced by the transition from the occupied mixed states of O 2p, Yb 4f, and Mn 3d to the unoccupied 3d3 states. This result represents a substantial step toward the exploration of novel chemically stable and nontoxic narrow-bandgap semiconductors potentially for applications of all-oxide solar cells, photoelectrodes, photodetectors, or photoelectronic devices.

Experimental Section

Among the 17 rare-earth elements, 12 were selected in this study. First, powders of RMnO3 (R = Y, La, Ce, Pr, Nd, Sm, Gd, Tb, Dy, Ho, Er, and Yb) were synthesized in a solid-state reaction of MnO2 with rare-earth oxides by mixing them together at a stoichiometric ratio of 1:1 for R to Mn and firing them under ambient air at 1000, 1200, or 1300 °C for 4 h. Then, the powders were pressed into disks with diameters of 11 mm and thicknesses of approximately 0.1–0.2 mm. To prevent grain particles from growing too large, the powders that were synthesized at 1000 °C for 2 h were used as precursors to fabricate pellets by mixing with approximately 5% poly(vinyl acetate) as a binder for granulation, pressing under 5 MPa, and sintering at 1350 °C for 24 h in an Ar atmosphere to obtain sufficient densification for photon response measurements. Next, one side of the pellets was sputtered with ITO and the other side was sputtered with gold to serve as electrodes. After being sintered, the surface profile of the pellets was examined with SEM (JSM-6490LV, JEOL). The crystal structures of the RMnO3 powders and pellets were investigated using an X’Pert PRO MPD X-ray diffractometer (PANalytical B.V., Almelo, the Netherlands). Absorption spectra of the powders were recorded with a Shimadzu UV-3600 UV–vis–NIR spectrophotometer equipped with an integrating sphere for the measurement of solid powders. Fluorescence spectra of the powders fired at 1300 °C were first collected using an FLS 920 spectrometer at 77 K, which was cooled with liquid nitrogen, but no spectra were acquired. Then, the fluorescence spectra were recorded using the pellets with a laser confocal Raman spectrometer (LabRAM HR Evolution Systems, HORIBA France SAS) pumped with a 633 nm laser at room temperature. XPS and UPS VB spectra of the pellets were measured using a Kratos Axis Ultra equipped with a monochromatic Al Kα X-ray source (hν = 1486.6 eV) for XPS and a He Iα photon source (hν = 21.2 eV) for UPS. The XPS and UPS spectra were calibrated with respect to the C 1s signal (284.8 eV). All XPS/UPS samples were analyzed using XPS PEAK4.1 software, through which the atomic ratio was derived by computing the area under the peaks. The photon response was measured using a multichannel scanning electrochemical workstation (Uniscan 370 Princeton Applied Research Scanning Electrochemical Microscopy (SCEM) Workstation). The Hall coefficient, resistivity, carrier concentration, and carrier mobility were tested using a Hall effect measurement system (ET9000, East Changing, China). The surface potential was measured using a Kelvin Probe, equipped at an atomic force microscope (Dimension Icon, Bruker). The BS, DOS, and PDOS of YbMnO3 were calculated using the virtual-crystal approximation (VCA) based on DFT of periodic quantum chemistry and the crystal structure ICSD 16-0749.[63] Yb, Mn, and O atoms arbitrarily occupy the same lattice position. The generalized gradient approximation (GGA)[64] with the Perdew–Burke–Ernzerhof (PBE)[65] functional and norm-conserving pseudopotentials was selected as the exchange–correlation functional. When running the geometry optimization, the maximum force and energy tolerances were set as 0.03 eV Å–1 and 1.0 × 10–5 eV atom–1, respectively, and the maximum displacement was set as 1.0 × 10–3 Å. A 500 eV cutoff energy and a 1 × 3 × 3 k-point sampling set were used for convergence.