Mn2+/Yb3+ Codoped CsPbCl3 Perovskite Nanocrystals with Triple‐Wavelength Emission for Luminescent Solar Concentrators

Abstract Doping metal ions into lead halide perovskite nanocrystals (NCs) has attracted great attention over the past few years due to the emergence of novel properties relevant to optoelectronic applications. Here, the synthesis of Mn2+/Yb3+ codoped CsPbCl3 NCs through a hot‐injection technique is reported. The resulting NCs show a unique triple‐wavelength emission covering ultraviolet/blue, visible, and near‐infrared regions. By optimizing the dopant concentrations, the total photoluminescence quantum yield (PL QY) of the codoped NCs can reach ≈125.3% due to quantum cutting effects. Mechanism studies reveal the efficient energy transfer processes from host NCs to Mn2+ and Yb3+ dopant ions, as well as a possible inter‐dopant energy transfer from Mn2+ to Yb3+ ion centers. Owing to the high PL QYs and minimal reabsorption loss, the codoped perovskite NCs are demonstrated to be used as efficient emitters in luminescent solar concentrators, with greatly enhanced external optical efficiency compared to that of using solely Mn2+ doped CsPbCl3 NCs. This study presents a new model system for enriching doping chemistry studies and future applications of perovskite NCs.

mixture was stirred and kept under vacuum for 1h to form a homogeneous clear solution without bubbles inside. Then the mixture solution was transferred into curing molds with different dimensions. The sample was cured at 75°C for 12h in an oven. The final Mn 2+ /Yb 3+ codoped NCs LSCs were removed from the mold after curing.
General characterization NC samples were dispersed in toluene for the absorption, PL, PLE, and pseudocolor map of excitation-dependent PL spectra measurements and in hexane for PL QY measurements. UV-Vis absorption spectra were measured using an Agilent Technologies Cary 5000 UV-Vis-NIR Spectrophotometer. Photoluminescence (PL) and photoluminescence excitation (PLE) spectra were measured by an Edinburgh Instruments Fluorescence Spectrometer FS5 utilizing a Xe lamp for excitation. Emission spectra were corrected for intensity and wavelength from factory provided correction files using calibrated and traceable lamps. The detectors used for measuring the PL include a visible range signal detector (a UV enhanced silicon photodiode) with spectral coverage from 230 nm to 870 nm and a NIR signal detector (a thermoelectric cooled InGaAs photodiode) with spectral coverage from 850 nm to 1650 nm. The full PL spectra were obtained by connecting two spectra collected by the visible signal detector and NIR signal detector with a scaling factor using CuInS2/ZnS core/shell nanocrystal (emitting at 737 nm with a peak width of ~ 145 nm) solution as a calibration sample for the responsivity of the two detectors under the same measurement conditions. The obtained signal scaling factor was further validated by the organic dye IR 806 (emitting at 830 nm with a peak width of ~ 30 nm). The bandgap (BG) and Mn emission PL QYs were measured directly using integrating sphere based on the equation shown below: Where "I" is the intensity of the emitted light after correction, "E" is the intensity of excitation light after correction, "sample" means the measurements for the NC solution samples, and "reference" means the measurements for the reference sample (hexane only solution) in a quartz cuvette. The Yb PL QYs and overall PL QYs were calculated based on the obtained scaling factor and integrated PL peak areas. All the spectra are corrected in real time for intensity variances as a function of detector and grating efficiency based on the provided correction files.
X-band electron paramagnetic resonance (EPR) spectra were obtained by a Bruker EMX Premium-X EPR Spectrometer. Measurements were taken at room temperature with a 9.86 GHz microwave frequency, 4 G modulation amplitude and a power of 2 mW.
Transmission electron microscopy (TEM) characterization was performed on a JEOL 2100F operated at 200 kV. The NC sample dissolved in a toluene solution (~10 μL) was drop-casted onto a 300-mesh copper TEM grid and dried in ambient conditions. X-ray diffraction (XRD) patterns were obtained on a Bruker D8 Discovery 2D X-ray Diffractometer equipped with a Vantec 500 2D area detector operating with Cu Kα ( = 1.541 Å) radiation. The NC samples were drop-casted on the glass slides and evaporated under mild heating (~ 60 ºC).
X-ray photoelectron spectroscopy (XPS) measurements were performed on a Thermo Scientific K-Alpha instrument operating on Al Kα=1486.6 eV radiation with a spot size of ~ 200 μm. The NC samples were drop-casted on the silicon wafers and evaporated under mild heating (~ 60 ºC).
For inductively coupled plasma-atomic emission spectroscopy (ICP−AES) analysis, the NC solution was dried and then digested in nitric acid (∼70 °C, 6 hours) to ensure complete dissolution . The solution was then diluted with 2% HNO3 solution to suitable concentrations.  The measurements were carried out on a Thermo Scientific iCAP 7400 DUO ICP−AES  equipped with a Teledyne ASX-560 240 position autosampler. Transmission spectra were measured using an Agilent Technologies Cary 5000 UV-Vis-NIR Spectrophotometer. The LSC devices were placed onto a film holder accessory for the transmission measurement.

Solution PL lifetime measurements
For BG-PL and Mn-PL lifetime, the measurements were conducted using an Edinburgh Instruments Fluorescence Spectrometer FS5 equipped with time correlated single photon counting method (TCSPC) and an EPLED-360 light source or a microsecond xenon flashlamp. The samples were dispersed in toluene within a quartz cuvette and excited at 360 nm. For Yb-PL lifetime measurement, 401 nm coherent laser cube is operated at pulsed mode, 300 Hz repetition rate, 100 ns pulse width (2.5 ns on/off switching time). Sample is illuminated under 2000 mW/cm 2 power density when laser is on. PL signals then pass through a bandpass filter (Semrock FF01-1001/234-25) and detected with single-photon avalanche photodiode (PicoQuant, τ-SPAD). Multichannel Averager (Stanford Research Systems, SR430) is used for recording time correlated single photon counting data. The decay curve was fitted with the triexponential decay expressed below: where I(t) is the observed ensemble PL intensity at time t, and Ik and k are the intensity and lifetime of an arbitrary excited state of k, respectively. We performed the fitting using the FAST software package (Edinburgh Instruments) and the PL intensity decays were found to be wellfitted by three exponential decay curves (k =1, 2, 3), with a goodness-of-fit (χR 2 ) in the range of 1.0 to 1.35.
External optical efficiency (ηext), internal optical efficiency (ηint), and concentration factor (C factor) measurements for LSCs Internal optical efficiency (ηint) is defined as the ratio of the LSC edge-collected photons to the total LSC absorbed photons from incident light, which can be obtained by measuring the LSC edge PL QY. External optical efficiency (ηext) is the ratio of edge emitted photons to the total incident photons, which can be calculated by the following equation: Where, ηabs is the absorption ratio of LSC device, which represents the portion of incident photons captured by the LSC device, described as: where T is the transmission of the overall LSC device, is the absorption coefficient of the LSC device, R is the reflection coefficient of LSC surface. The C factor can be obtained by multiplying the ηext with the geometry gain factor (G, the ratio of the areas of the top/bottom surface to edge regions). The overall parameter to describe LSC performances [2] is expressed as:

LSC photovoltaic (PV) integrated device characterizations
Polyscrystalline silicon (c-Si) AMX3d Micro Mini Solar Cells coupled with the edge regions of LSCs were fabricated for the devcie characterizatons. The black tape was applied to cover the excess area of the solar cell, the rest of LSC edge region and LSC bottom surface. Then, the current voltage measurements (J-V curves) were conducted for the whole system. The J-V characteristics of the solar cells, Mn-LSC and Mn/Yb-LSC were obtained using the 2400 SourceMeter (Keithley, USA) under the simulated 1-sun AM1.5G 100 mA•cm -2 intensity (Sol3A Class AAA, Oriel, Newport; USA) at ambient conditions, with a step size of 0.05 V and a delay time of 100 ms. The spectrally resolved external quantum efficiency (EQE) measurements were conducted using a quantum efficiency measurement system (IQE 200B; Oriel; USA) consisting of a xenon lamp, amonochromator, a lock-in amplifier, and a calibrated silicon photodetector, with a step size of 20 nm.

Monte Carlo ray-tracing simulation (MC simulation)
As a widely applied simulation approach for LSC system, MC simulation can track a number of photon traces in a LSC device. [3][4][5][6] Therefore, we can predict an overall trend of photon propagation and further evaluate the device performances. The working principle of the MCmode is described in the following paragraphs and shown in the logic flow chart below.
Simulation inputs include: the total number of photons (i.e., 30,000) in the MC simulation, the dimension parameter (1.26 cm × 1.26 cm × 0.1 cm) of the LSC, the refractive index of the polymer matrix material (n = 1.4 for PDMS), the emission and absorption spectra and the PL QY of the emitters. Next, according to absorption data and Fresnel Law, we calculate how many of the incident random photons will be absorbed by the embedded NCs. For each captured photon, we assign random x and y coordinates for its initial position. We then determine if the absorbed photon will be emitted by the NC emitters (according to the emission spectrum and the PL QY of the perovskite NCs). If not, the photon is lost through the non-radiative decay channels of the NCs. Otherwise, we determine if the photon will hit the surface of LSC before re-absorbed by another NC. If not, the photon will be considered back to the previous emission determination step. Otherwise, the photon will be updated with a new set of coordination vectors then proceed to the next step. Based on the new photon location coordination, we can determine which surface the photon hits (top/bottom or edge of the LSC).
According to the location vectors of the photon that hits the surface and refractive index of the polymer matrix, we can determine whether the photon will be reflected through the total internal reflection (TIR). If not, the photon will transmit through the surface and close the photon tracing loop. Otherwise, the simulation procedure moves back to the initial emission determination step and surface hitting step until the loop is closed. After that, the next random photon will be released into the main loop to repeat the ray-tracing procedures. The main loop will be repeated until all the photons are traced. If the photon travels through the edge surfaces, we consider it as a collected photon; if the photon goes to the top/bottom surfaces, we consider it as a lost photon.

Analytical mode simulation
Based on Weber and Lambe analysis, the LSC efficiency function is expressed as below.

, (S4)
Here, is the probability of the first-time reabsorption loss, is a geometry correction parameter (set by the equation (S12)). Then, the collection efficiency ( ) of the first generation of emitted photon can be represented as: is the PL QY of emitter, is the light trapping efficiency which is defined as: Where, n is the LSC's refractive index, θesc is the escape cone angle, which is defined by Snell's law as arcsin . To account for the collection efficiency ( ) of the secondgeneration of re-emitted photons, the function is expressed as follow: Thus, each re-emission process term represents a member of a geometric progression. The total collection efficiency ( ) can be expressed as a sum of contributions due to all photon generations.
However, the Weber and Lambe analysis do not consider any scattering effect. This can generate considerable errors for the re-absorption free systems like the LSC in this case. Therefore, by considering the scattering and re-absorption free effect, the ηint and ηext calculations can be expressed as follows: , , (S10) Where, S2 is the scattering coefficient at the PL wavelength, L is the edge length of LSC. <α > is the average absorption coefficient among the incident light wavelength range represented as: α , (S11) The obtained value can be applied for the α in the equation (S9). α is the absorption coefficient of the emitter. λis the wavelength. d is the thickness of the LCS. is the bandgap of the emitter. is the spectral shape of incident light.
For β value determination, the initial value can be calculated based on the following equation:
c) The PL QY of 200% is for simulation as the theoretical limit.  Table S10 below.   a) The reference [16] and [19], only lateral dimensions of the LSC devices are provided.