Lanthanide doping in metal halide perovskite nanocrystals: spectral shifting, quantum cutting and optoelectronic applications

Lanthanides have been widely explored as optically active dopants in inorganic crystal lattices, which are often insulating in nature. Doping trivalent lanthanide (Ln3+) into traditional semiconductor nanocrystals, such as CdSe, is challenging because of their tetrahedral coordination. Interestingly, CsPbX3 (X = Cl, Br, I) perovskite nanocrystals provide the octahedral coordination suitable for Ln3+ doping. Over the last two years, tremendous success has been achieved in doping Ln3+ into CsPbX3 nanocrystals, combining the excellent optoelectronic properties of the host with the f-f electronic transitions of the dopants. For example, the efficient quantum cutting phenomenon in Yb3+-doped CsPb(Cl,Br)3 nanocrystals yields a photoluminescence quantum yield close to 200%. Other approaches of Ln3+ doping and codoping have enabled promising proof-of-principle demonstration of solid-state lighting and solar photovoltaics. In this perspective article, we highlight the salient features of the material design (including doping in Pb-free perovskites), optical properties and potential optoelectronic applications of lanthanide-doped metal halide perovskite nanocrystals. While review articles on doping different metal ions into perovskite nanocrystals are present, the present review-type article is solely dedicated to lanthanide-doped metal halide perovskite nanocrystals. Methods for integrating lanthanide materials into light-emitting devices to improve their performance have been reviewed by scientists from India and China. Semiconductors create light when the energy of a high-energy electron is converted to a single photon. Conventionally, the minimum energy of this photon is determined by an intrinsic material property known as the bandgap. Wasim Mir from the Indian Institute of Science Education and Research Pune, and colleagues, summarize developments in using lanthanide ions to enable the generation of lower-energy light. Lanthanide ions provide an energy “step” within the bandgap that means the electron is converted to two low energy photons rather than a single high energy one. The authors review how embedding lanthanum ions within nanoparticles made from so-called metal-halide perovskites makes them easier to integrate into common optoelectronic semiconductors such as silicon. Metal halide perovskites are extraordinary defect-tolerant semiconductors. A unique structural aspect of perovskites is the octahedral coordination for (B-site) metal ions, unlike other semiconductors that exhibit tetrahedral coordination. This octahedral coordination helped to achieve lanthanide doping in halide perovskite nanocrystals in 2017. Fundamental understanding of material design, luminescence and quantum cutting phenomena in lanthanides (with focus on Yb3+) doped in CsPbX3 (X = Cl, Br, I) and Cs2AgInCl6 nanocrystals are reported. Subsequently, these doped systems are applied for solar energy harvesting and lighting in both visible and near infrared region. This perspective article summarizes everything important that has happened so far in field and discusses about the future research directions.


Introduction
A variety of interesting optical properties and applications of inorganic materials depend on the presence of lanthanide ions (Ln 2+ /Ln 3+ ) doped into the crystal lattice. Generally, lanthanide ions, as optically active centers, provide energy levels within the band gap of the material, so they can give rise to the appearance of optical transitions at frequencies lower than that of the fundamental absorption. For example, Ce 3+ -doped Y 3 Al 5 O 12 is a benchmark light emission material for commercial white light-emitting diodes (WLEDs) 1 . The 4f-5d transition of lanthanide ions represented by Eu 2+ or Ce 3+ is a parityallowed electric dipole transition, and thus, it is highly efficient. However, typical Ln 3+ ions, except for Ce 3+ , exhibit 4f-4f forbidden transitions with well-defined energy levels that are nearly invariable in different hosts due to the shielding of 4f orbitals 2,3 . Thus, the photoluminescence (PL) arising from the f-f transition exhibits a long lifetime (~ms), along with a poor light absorption coefficient (~1-10 M −1 cm −1 ) 4 . Therefore, organic dyes and complexes have been used as light absorbing sensitizers to nonradiatively excite the f electrons of Ln 3+ ions, and light is emitted via the subsequent de-excitation of the f electrons 5 . To boost the PL quantum yield (QY) of Ln 3+ ions, Ln 3+ ions have been doped into crystalline sensitizers, shielding the excited f electrons from unwanted nonradiative decay channels.
All-inorganic, metal-halide perovskite CsPbX 3 (X = Cl, Br, I) nanocrystals (NCs) have recently been the subject of intense research [6][7][8][9][10][11][12][13] . Morphology modification, size control and compositional alloying play important roles in PL tuning and optoelectronic applications, such as LEDs, lasers, displays, solar cells, and photodetectors. Moreover, the incorporation of impurity ions or doping is a promising method for controlling the electronic and optical properties and the structural stability of halide perovskite NCs [14][15][16][17][18][19] . Lanthanide ions with unique optical properties have been successfully doped into CsPbX 3 NCs 20-23 . These materials demonstrate the rich, efficient, and inherently narrow 4f-4f emission features of the lanthanides sensitized by the perovskite NCs. Except for the normal spectral shifting behavior, Yb 3+ -doped CsPbCl 3 NCs particularly show excellent near-infrared (NIR) PL with QY approaching 200% because of a rarely observed phenomenon, namely, quantum cutting. The combination of efficient quantum cutting and the strong absorption cross-section of CsPbCl 3 for the UV-blue region of the solar spectrum lead to novel strategies for solar light harvesting, such as (i) depositing a quantum cutting layer on a Si solar cell to improve its efficiency and (ii) utilizing luminescent solar concentrators.
In this perspective article, we discuss the exciting research developments in Ln 3+ -doped metal halide perovskite NCs in the following subsections: (i) lanthanide doping in octahedral (perovskite) semiconductor NCs, (ii) luminescence via spectral shifting from lanthanide-doped CsPbCl 3 perovskite NCs, (iii) quantum cutting and band gap tuning, (iv) optoelectronic applications, (v) prospects of lanthanide doping in Pb-free perovskites, and (v) conclusions and future outlook.
Lanthanide doping in octahedral (perovskite) semiconductor NCs Figure 1 depicts Ln 3+ doping in widely studied crystal lattices from insulators to semiconductors. Ln 3+ doping in oxide and fluoride lattices, such as Y 3 Al 5 O 12 , NaYF 4 , and NaGdF 4 , has been successfully realized due to the higher coordination (coordination number, CN = 8 or 9) environment possible for Ln 3+ dopants [24][25][26] . However, a drawback of such host materials is their wide band gap, yielding both poor visible light absorption and poor charge transport. Therefore, for visible light optoelectronic applications, including solar light harvesting and LEDs, efforts have been made to dope Ln 3+ into the lattice of semiconductors. However, Ln 3+ doping in semiconductors has remained challenging because most semiconductors, such as CdSe, CdS, Si, GaAs, and InP, offer a tetrahedral (CN = 4) coordination environment for Ln 3+ dopants, whereas Ln 3+ ions prefer sites with CN ≥ 6 27,28 . It is noteworthy that lead halide perovskites are a rare class of semiconductors that possess octahedral coordination (CN = 6) for Pb (B-site cation); therefore, they provide an opportunity to dope lanthanide ions into optoelectronically active semiconductors. CsPbX 3 NCs exhibiting strong visible light absorption and intense emission arising from their excitonic (band gap) transitions, along with reasonable charge transport properties, are therefore ideal hosts for Ln 3+ ion doping.

perovskite NCs
The first manuscript on Ln 3+ (Yb 3+ , Ce 3+ , Er 3+ ) doping in CsPbCl 3 and mixed halide CsPb(Cl,Br) 3 perovskite NCs was reported in 2017 after the seminal work of Song and coworkers 20 . The same group expanded the scope of optical properties by doping a large number of Ln 3+ ions into CsPbCl 3 NCs (Fig. 2 29,30 . However, easy and flexible Ln 3+ doping in narrower band gap hosts, such as CsPbBr 3 and CsPbI 3 NCs, is challenging. Very few reports were able to achieve Ln 3+ doping in CsPbBr 3 NCs by employing a direct synthesis using halide precursors of Ln 3+ 31,32 . This challenge of doping narrower band gap hosts could alternatively be handled by postsynthesis strategies, as Schematics showing the coordination environment provided to Ln 3+ by different host lattices. Ln 3+ doping preferentially occurs in lattices with coordination number (CN) ≥ 6. Therefore, metal halide perovskites with CN = 6 exhibiting good optoelectronic properties are ideal semiconductor host NCs for Ln 3+ doping. discussed later in the quantum cutting and band gap tuning section.
The PL spectra in Fig. 2b show that the undoped sample exhibits only excitonic emission, but all the doped samples show multiple emission peaks arising from both excitonic and dopant emissions. None of these Ln 3+ dopants absorbs a measurable amount of light, and therefore, the absorption of light is dominated by the excitonic (or band gap) transition of the host CsPbCl 3 at wavelengths < 420 nm 21 . Therefore, the Ln 3+ emissions are sensitized by the host. Figure 2c shows the absolute PL QYs of the excitonic and total (excitonic + dopant) emission for various Ln 3+ -doped CsPbCl 3 NCs. The PL QY of Yb 3+ emission is by far the highest. Surprisingly, the first three reports of Yb 3+ doping in CsPbCl 3 or CsPb(Cl,Br) 3 NCs reported a very unusual PL QY exceeding the ideal value 100% [20][21][22] . A long PL lifetime >2 ms was reported for the Laporte forbidden NIR Yb 3+ emission, suggesting incorporation of the dopants into the host lattice 22 . Local structure studies also confirm that Yb 3+ occupies the octahedral Pb 2+ site of CsPbCl 3 NCs 33 . Gamelin and coworkers reported añ 190% PL QY from not only NCs but also bulk-like thin films of Yb 3+ -doped CsPbCl 3 34 . This extraordinarily high PL QY for Yb 3+ emission received significant attention from researchers for both fundamental studies and possible future applications. Consequently, both the prior literature on lanthanide-doped perovskite NCs and our present perspective are dominated by the studies on Yb 3+ doping.

Quantum cutting and band gap tuning
In the previous subsection, we discussed the PL QY of Yb 3+ -doped CsPbCl 3 NCs approaching~200%. Figure 3a shows that the absorption onset (420 nm, 2.95 eV) is more than twice the Yb 3+ emission energy (992 nm, 1.25 eV) 22 . Therefore, in principle, it is possible that one absorbed photon of energy ≥ 2.95 eV can yield two emitted photons of 1.25 eV energy, following a phenomenon known as quantum cutting (Fig. 3a). Moreover, the PL QY of NIR Yb 3+ emission in Yb 3+ -doped CsPbCl 3 NCs shows a fluence-dependent nature under varying excitation power at λ exc = 380 nm 22 . At lower excitation power, the PL QY of the Yb 3+ emission increases and reaches close to 200%, showing a very efficient quantum cutting phenomenon. This high PL QY at low excitation power is possible due to the high absorption coefficient of the CsPbCl 3 NC host, which directly transfers excitonic energy to Yb 3+ ions upon photoexcitation. Unlike in conventional quantum cutting 20,35 , a high-energy photon absorbed by the host is directly converted to two low-energy photons without the requirement of two different lanthanide ions. Gamelin and coworker reported an efficient quantum cutting process at the picosecond scale 22 . To maintain charge neutrality, the incorporation of two Yb 3+ ions will lead to the removal of three Pb 2+ ions, creating Pb 2+ vacancies (V Pb ), thereby forming a Yb-Cl-V Pb -Cl-Yb defect complex (Fig. 3b). It was suggested that this defect forms a shallow state below the conduction band minimum, facilitating picosecond nonradiative energy transfer from the photoexcited host NC to two Yb 3+ ions simultaneously, yielding quantum cutting 22 . Later, theoretical studies extended the mechanistic insight 36 . They suggested that the two Yb 3+ dopants are excited in a concerted manner through nonradiative energy transfer from the PbCl 6 octahedra closest to the two Yb 3+ dopants (see Fig. 3c) 36 .
All these studies were performed with Yb 3+ doping in CsPbCl 3 or CsPb(Cl/Br) 3 perovskite NCs with a band gap >2.88 eV (430 nm). Doping Yb 3+ into narrower band gap perovskites, such as CsPbBr 3 and CsPbI 3 NCs, was found to be challenging. Such tuning of the band gap of the host NCs is interesting for both a fundamental understanding of the role of the band gap in quantum cutting and possible applications requiring more visible light absorption. To address this challenge of band gap tuning in Yb 3+ -doped CsPbX 3 NCs, two strategies are adopted: (i) postsynthesis doping of Yb 3+ into preformed NCs with the desired band gap 23 and (ii) anion exchange of Yb 3+ -doped CsPbCl 3 perovskite NCs 37 . The postsynthesis doping of Yb 3+ into all compositions of CsPbX 3 (X = Cl, Br, I) NCs (also into CsPbBr 3 nanoplatelets (NPLs)) yielded NIR Yb 3+ emission (see Fig. 3d) 23 . However, the relative intensity of the NIR Yb 3+ emission is significantly decreased for CsPbBr 3 and CsPbI 3 NCs. In the anion exchange approach, the PL QY of the NIR Yb 3+ emission sharply dips for band gaps smaller than 2.5 eV, as shown in Fig. 3e and its inset 37 . The band gap of 2.5 eV is thus termed the threshold value above which the quantum cutting phenomenon is realized in Yb 3+ -doped CsPb(Cl 1−x Br x ) 3 NCs. The schematic in the inset shows the quantum cutting of one high-energy absorbed photon into two low-energy photons in Yb 3+ emission, giving rise to PL QY > 100%. Figure a is adopted from ref. 22 with Copyright © 2018, American Chemical Society. b Atomic model in which 2 Yb 3+ ions replace 3 Pb 2+ ions in the crystal lattice of CsPbCl 3 , leading to the formation of the right angled charge neutral Yb 3+ -V Pb -Yb 3+ defect complex. Black spheres represent Cs, white sphere represents Pb vacancies (V Pb ), gray colored octahedra correspond to PbCl 6 , and pink colored octahedra correspond to YbCl 6 . c Nonradiative excitonic energy transfer from PbCl 6 octahedra to the two nearest Yb 3+ dopants in a concerted manner. d PL spectra of undoped and x% Yb 3+ -doped CsPbX 3 (X = Cl, Br, I) perovskite NCs along with CsPbBr 3 perovskite nanoplatelets (NPLs, blue spectra) obtained though postsynthesis Yb 3+ doping. Figure d is adopted from ref. 23

Optoelectronic applications
The light emitted by Yb 3+ at~990 nm is suitable for absorption by a commercial Si solar cell with reasonably good incident photon-to-current conversion efficiency (IPCE). Consequently, efforts have been made to improve the power conversion efficiencies (PCEs) of Si solar cells via two mechanisms: (i) using a quantum cutting layer on a Si solar cell (Fig. 4a-c) and (ii) using luminescent solar concentrators (LSCs) (Fig. 4d-f) to build integrated photovoltaics. In a different direction, lanthanide doping was found to increase the grain size of CsPbBr 3 microcrystals in a film, thereby increasing the carrier lifetime and efficiencies of perovskite solar cells (Fig. 4g, h) 38 . In addition, the different luminescence colors arising from different lanthanide dopants are being explored for lightemitting applications, including white LEDs (Fig. 4i). In this subsection, the progress in and challenges for these possible applications will be discussed.

Quantum cutting (QC) layer on a Si solar cell
Zhou et al. improved the PCE of a Si solar cell from 18.1% to 21.5% simply by coating the solar cell with a Yb 3+ (7.1%)-Ce 3+ (2%) codoped CsPbCl 1.5 Br 1.5 NC layer (Fig. 4a, b) 20 . As shown in Fig. 4c, the IPCE of the Si solar cell significantly decreases for the solar spectrum below 450 nm. The idea here is that the layer of Yb 3+ -doped (or Yb 3+ -Ce 3+ codoped) CsPbX 3 NCs absorbs this part (<450 nm) of the solar light and converts it to~990 nm Yb 3+ emission (PL QY > 100%), which is then reabsorbed by the Si solar cell. The quantum cutting layer increases the IPCE of the Si solar cell in the 350-450 nm region, improving the solar cell efficiency. The thickness of the quantum cutting NC layer needs to be optimized tõ 230 nm so that the layer sufficiently absorbs solar light below <450 nm but remains transparent to 450-1000 nm light, which is directly absorbed by the Si solar cell with high IPCE.
Another important point is that the absorbed ultraviolet-blue light possesses energy much higher than the band gap of the solar cell materials, including Si. This excess energy is lost as heat during relaxation of the hot carriers to the band edges. Such thermalization losses constitute a significant part of the total losses causing the Shockley-Queisser thermodynamic limit of the PCE for single-junction Si solar cells of 31% 39 . Converting UVblue light to NIR light using a quantum cutting layer has the potential to reduce the thermalization losses, thereby achieving a theoretical efficiency of a single-junction solar cell beyond the Shockley-Queisser limit, similar to the case of tandem solar cells 40 . More theoretical and experimental works are required to verify the efficacy of Yb 3+ -doped quantum cutting NC layers in reducing the thermalization losses.

Luminescent solar concentrators (LSCs)
LSCs absorb incident solar light, and then, the emitted lower energy light is waveguided to the edges of the device by total internal reflection. A photovoltaic device attached to the edge of the LSC absorbs the waveguided light and converts it into usable power. The internal optical efficiency (η int ) of an LSC is the ratio of edge-emitted photons to absorbed solar photons. Therefore, to achieve high η int , we need both high PL QY and minimal reabsorption of the emitted light by the medium of the LSC such that the emitted light can travel to the edges of the LSC without any loss. Yb 3+ -doped CsPbX 3 NCs fulfill both criteria because of the quantum cutting that provides >100% PL QY and the large redshift between absorption and Yb 3+ emission (see Fig. 3a). Wu and coworker reported a very high η int (118.1%) using Yb 3+ -doped CsPbCl 3 NCs as a luminophore in a 25 cm 2 sized LSC (Fig. 4d) 41 . The upper inset of Fig. 4d shows that the Yb 3+ -doped LSC is largely transparent to visible light, suggesting that such LSCs can be used in the glass window material for a building, whereas the lower inset shows that the luminescence generated in the LSC travel to its edges. Extrapolation of the data (Fig. 4e) shows the possibility of high η int using Yb 3+ -doped CsPbCl 3 NCs in a large-size LSC. Interestingly, the η int of the Yb 3+ -doped sample is approximately 2-fold higher than that of the Mn 2+ -doped quantum dot LSC 42 . The external optical efficiency (η ext ) of an LSC also depends on the solar photon absorption efficiency (η abs ) of the luminophore. The wide band gap of CsPbCl 3 NCs reduces their solar absorption capability to only 3.1% η abs , leading to an η ext of 3.7% from a 5 cm 2 LSC. This η ext is not high compared with previous quantum LSCs 41,42 . Yb 3+ doping in narrower band gap CsPbCl x Br 3−x NCs with a higher η abs of 7.6% shows a much-improved η ext of 9.0% for a 5 cm 2 LSC (Fig. 4f) 41 . However, this narrowing of the host band gap will reduce the visible light transparency of LSCs.

Perovskite solar cells
In a different direction from quantum cutting for improving the efficiency of Si solar cells, lanthanide doping has been reported to increase the efficiency and stability of perovskite solar cells 38,[43][44][45] . Ln 3+ doping has been reported to reduce lattice and surface defects of both NCs and bulk-like perovskite thin films 32,38 . Figure 4g, h shows that the solar cell performance of CsPbBr 3 thin films significantly improves with various lanthanide doping. For example, Sm doping increases the PCE from 6.99 to 10.14% using a device geometry that does not require hole-transporting materials (HTMs) 38

White light emission
Good-quality white light generation requires the simultaneous emission of red, green and blue colors in appropriate proportions. However, the simultaneous generation of these three colors with high efficiency and without self-absorption (which may change the color ratio) is difficult. Doping allows the generation of multiple emissions and is free from self-absorption. Song and coworkers reported white light emission from lanthanide (2.7% Ce 3+ ) and transition metal (9.1% Mn 2+ ) codoped CsPbCl 1.8 Br 1.2 NCs with a high PL QY of 72% (Fig. 4i) 46 . Optically pumped white light LEDs of these codoped NCs exhibit a luminous efficiency of 51 lm/W and a color rendering index of 89 after excitation with a 365 nm GaN LED chip. Typically, near UV light pumped white LEDs can generate a high color rendering index of 94, but at the cost of a poor luminous efficiency of~23 lm/W, for rare earth ion doping in oxide lattices 47 . Interestingly, codoped perovskite NCs under near UV light excitation result in a reasonably high luminous efficiency while maintaining a high color rendering index.

Conclusions and future outlook
Unlike the poor doping tendency of Ln 3+ ions into traditional (CdSe, InP, etc.) semiconductor NCs, doping of Ln 3+ into CsPbX 3 NCs that provide the desired octahedral coordination environment is relatively easy. Consequently, in less than three years, Ln 3+ -doped metal halide perovskite NCs emerged as new visible light harvesting phosphors. Different Ln 3+ ions emit light with a well-defined sharp spectral linewidth in the visible and NIR regions, including white light emission obtained by codoping different metal ions. The most important finding thus far is the efficient quantum cutting phenomena in NCs and thin films of Yb 3+ -doped CsPb(Cl/Br) 3 perovskites resulting in an extraordinary PL QY approaching 200%. The Yb 3+ emission at~990 nm can be absorbed by a Si solar cell, and such perovskites are therefore being explored as both (i) a quantum cutting layer on top of a commercial silicon solar cell and (ii) an LSC. In a different direction, Ln 3+ doping can also improve the quality and stability of perovskite crystals, improving the performance of perovskite solar cells. Yb 3+ and Er 3+ have also been doped into Pb-free double perovskite NCs, but with poor PL QY.
There are many new opportunities and challenges that need to be addressed in the future. Ln 3+ ions give rise to a strong magnetic moment per free ion, which includes both orbital and spin contributions. For example, Dy 3+ and Ho 3+ with 5 and 4 unpaired electrons can generate a magnetic moment of >10 μ B per ion. Such spin-based properties, including magneto-optical properties, have not yet been explored for lanthanide-doped perovskite NCs.
Another promising direction is to fabricate highefficiency NIR LEDs and sensors. Extending the existing knowledge of device physics for undoped host CsPbX 3 NCs to Yb 3+ -doped samples appears to be a natural next step. If an efficient and stable NIR LED can be prepared from solution-processed and low-cost Yb 3+ -doped CsPbX 3 NCs, then it may find commercial applications. Likewise, extending the LSC properties to applications by integrating a photovoltaic module at the edges of an LSC panel will be interesting. However, fine tuning of the host composition to absorb the optimal level of solar light is required.
Additionally, the characterization of the local structure around the dopant ions and their spatial distribution in NCs are not sufficiently studied. The distinction between properties arising from Ln 3+ dopants on the surface of CsPbX 3 NCs and those from dopants in the core of NCs needs to be further understood by employing various experimental techniques. Likewise, a microscopic understanding of quantum cutting in Yb 3+ -doped CsPbCl 3 NCs, which is believed to proceed through the formation of a defect complex (Yb-Cl-V Pb -Cl-Yb), also requires further experimental validation. Finally, there are scopes to improve the quality of lanthanide-doped double perovskite NCs and, in general, lanthanide-doped Pb-free halide perovskites to explore their potential as optoelectronic materials.