Mingzhi Li, Zhikai Liu, Wang Yao, Chao Xu, Yangping Yu, Mei Yang*, Guangwen Chen

1 Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China

2 University of Chinese Academy of Sciences, Beijing 100049, China

Keywords:

ABSTRACT

1. Introduction

All-inorganic cesium lead halide perovskites(CsPbX3,X=Cl,Br,I) nanocrystals (NCs) possess many fascinating properties, which endow them with great potential in many fields [1,2]. As is well demonstrated, the physicochemical properties of CsPbX3NCs are significantly affected by the composition, morphology and size,etc. [3]. Therefore, great efforts have been made to synthesize high-quality CsPbX3NCs with controlled properties [4]. Hot injection is the most popular method, but inert atmosphere and high temperature are inevitably used in hot injection. This increases the cost and decreases the productivity. To address these issues,Li et al. [5] developed a novel strategy, namely ligand assisted reprecipitation (LARP), by transferring the perovskite precursors(i.e., CsX, PbX2) from a good solvent (i.e., N,N-dimethylformamide,dimethyl sulfoxide) to an antisolvent (i.e., toluene) at room temperature.Supersaturation is created once the good solvent is added into the antisolvent, triggering nucleation and further growth of CsPbX3NCs. Since the nucleation and growth rates of CsPbX3are ultrafast, the localized supersaturation level dramatically affects the nucleation and growth rate,and thus the morphology and size uniformity of CsPbX3NCs[6].Generally,the localized supersaturation is determined by the mixing degree of the good solvent and antisolvent. The good solvent must be quickly dispersed into the antisolvent to generate a spatially uniform supersaturation. In the study of Li et al. and subsequent studies, the synthesis of CsPbX3NCs via LARP was always carried out in a laboratoryscaled batch reactor (e.g., beaker or flask) under magnetic stirring.However,it is difficult for batch reactors to offer a uniform precipitation environment owing to the inferior transfer properties,leading to poor control over particle morphology and size and high batch-to-batch variation [7,8]. These issues become more severe when scaling up the laboratory-developed batch process due to the obvious scale-up problems of batch reactors [9]. Therefore, it is highly urgent to develop a novel reactor which can ensure the fast mixing of the good solvent and antisolvent on different scales to synthesize high-quality CsPbX3NCs.

In the past three decades, microreactor has emerged as one of the most promising process intensification technologies mainly due to its remarkable enhancement and precise control over the mixing and mass/heat transfer [10]. The continuous synthesis of nanocrystals in microreactors has attracted more and more attention. In particular, the continuous synthesis of CsPbX3NCs based on ionic metatheses reactions in the organic solvent at high temperature was achieved in droplet-based microfluidic platforms,and precise control over CsPbX3NCs was realized via the quick quench of high-temperature reactions [11,12]. For LARP-based synthesis of CsPbX3NCs,several protocols based on the microreactors with different structures were developed [13,14]. Although great progress has been made, two critical shortcomings limit the applicability of the microreactor in material synthesis. First, the mixing performance of the microreactor is highly dependent on the flow rate, which causes a large pressure drop and high cost for process screening and optimization [15,16]. Second, because the characteristic channel size is below 1 mm, the microreactor suffers from a high risk of fouling and blockage, which makes the long-term operation challenging. The droplet-based microfluidic process could effectively prevent the channel from being clogged,but the throughput was always very low (0.1–350 μl?min-1in the work of Lignos et al. [12]). Hence, it is of vital importance to minimize the dependency of mixing on the flow rate and address the fouling and blockage problem of the microreactor.

One of the most promising solutions is the introduction of power ultrasound (20–200 kHz) into the microreactor [17,18].When power ultrasound propagates into a liquid medium with the power higher than a threshold, ultrasonic cavitation occurs.Microbubbles generated via ultrasonic cavitation oscillate violently in the liquid medium, inducing strong shear force, cavitation microstreaming and shock waves. These phenomena are known as the mechanical effects of ultrasonic cavitation [19]. The mechanical effects can not only enhance mixing and mass transfer,but also break up aggregates of particles, and detach particles deposited on channel walls [20,21]. Therefore, the introduction of power ultrasound can smartly solve the aforementioned two problems associated with microreactors, and has already been used to intensify the synthesis of different nanocrystals [22]. As discussed above,the key step for the synthesis of CsPbX3NCs via LARP is the rapid mixing of the good solvent and antisolvent,where the ultrasonic microreactor (USMR) has a significant advantage. Therefore,it can be expected that the synthesis of CsPbX3NCs could be intensified by the USMR.Ng et al.[23]used a high-frequency ultrasound(1.06 MHz) actuated mixer to actuate acoustic streaming and achieve millisecond mixing. The continuous synthesis of highquality CsPbBr3perovskite NCs was realized. However, acoustic streaming-enhanced micromixers are always operated at low flow rates,e.g.,1.16 ml?min-1in the study of Ng et al.This might hinder the large-scale production. Unlike acoustic streaming, process intensification based on ultrasonic cavitation actuated by power ultrasound (20–200 kHz) can be operated at higher flow rates,being beneficial for process scale-up.Hence,we developed a robust USMR by directly coupling glass capillaries with a Langevin-type ultrasonic transducer to actuate ultrasound cavitation in this study.The USMR was used to continuously synthesize cesium lead halide perovskite NCs by LARP. The effects of ultrasonic power,ultrasonic treatment time, total flow rate, water additive, and reprecipitation temperature were systematically investigated. As compared to the batch method, the continuous method based on the USMR showed an obvious advantage in photoluminescence(PL) intensity and experimental reproducibility

2. Experimental

2.1. Materials

Lead bromide (PbBr2, AR, 99.0%, Aladdin, China), cesium bromide (CsBr, metals basis, 99.5%, Aladdin), lead chloride (PbCl2,AR,Aladdin),cesium chloride(CsCl,AR,≥99%,Aladdin),lead iodide(PbI2, AR, 98%, Aladdin), cesium iodide (CsI, metals basis, 99.9%,Aladdin), oleic acid (OA, AR, Sinopharm Chemical Reagent Co.,Ltd., China), oleylamine (OLA, C18: 80%–90%, Macklin, China), N,N-dimethylformamide (DMF, AR, ≥99.5%, Sinopharm Chemical Reagent Co., Ltd.), dimethyl sulfoxide (DMSO, AR, ≥99.0%, Sinopharm Chemical Reagent Co., Ltd.), toluene (AR, ≥99.5%, Sinopharm Chemical Reagent Co., Ltd.), Rhodamine B (AR, Aladdin)were purchased. All reagents were used as received except DMF,DMSO and toluene.DMF,DMSO and toluene were dried by adding 0.4 nm molecular sieves(Dalian Molecular Sieve Plant,China)prior to use. In order to guarantee the number of gas nucleus under ultrasound, the dried toluene was purged with N2to obtain the N2-saturated state.

2.2. Fabrication of USMR

As shown in Fig.S1(in Supplementary Material),the USMR was fabricated by directly coupling glass capillaries with a Langevintype ultrasonic transducer (ZFHN-8020, Baoding Zhengjie Electronic Technology Co., Ltd., China) via epoxy glue (ENDFEST300,UHU, Germany). The working frequency of the transducer was around 21 kHz, and the ultrasonic power could be tuned in the range of 5–35 W. A fluorinated ethylene propylene (FEP) capillary(outer diameter (OD) = 2 mm, inter diameter (ID) = 1 mm) was inserted into the glass capillary. A stainless-steel capillary(OD = 0.7 mm, ID = 0.5 mm) was inserted into the FEP capillary and then went into the glass capillary to organize the inlet structure. The detailed geometry information of these capillaries is exhibited in Fig. S1(a). The ultrasonic treatment time was tuned by varying the number of the glass capillary,and the corresponding USMR was denoted as xUSMR,where x represented the number of the glass capillary. Different glass capillaries were connected by FEP capillaries with an OD of 2 mm, an ID of 1 mm and a length of 3 cm. The USMR was excited by a signal generator (AFG2112,GW InSTEK, Taiwan, China), whose signal was further amplified using an RF power amplifier (AG 1016, T&C Power Conversion,USA).

2.3. Mixing behavior in USMR

The mixing behavior under ultrasound was investigated by a dye dilution-based characterization method. The experimental setup is shown in Fig. S2. To clearly observe the mixing behavior in the glass capillary, a LED light source and a high-speed camera were placed on both sides of the reactor. Therefore, 1USMR was used to eliminate the interference between different glass capillaries when capturing a mixing picture. Other experimental details could be found in the Supplementary Material.

2.4. Continuous synthesis of CsPbBr3 NCs in the USMR

Fig.1. (a)The experimental setup for the continuous synthesis of CsPbBr3 NCs in the USMR.(b)The optical image of the USMR under irradiation of UV light(365 nm)during the synthesis process of CsPbBr3 NCs. (c) The schematic diagram of the recrystallization process of CsPbBr3 NCs under ultrasonic irradiation.

The experimental setup for the continuous synthesis of CsPbBr3NCs via LARP was shown in Fig. 1(a). A FEP sampling tubing(OD = 2 mm, ID = 1 mm) was inserted into the glass capillary to prolong the residence time. The precursor solution was prepared by dissolving CsBr (0.8 mmol), PbBr2(0.8 mmol), OLA (1 ml), and OA(2 ml)with 20 ml dried DMF.When studying the effect of water additive, 0–200 μl deionized water was added into the precursor solution. In other cases, an optimal amount of water additive of 100 μl was added into the precursor solution. Subsequently, the precursor solution was injected into the USMR through the stainless-steel capillary with the volumetric flow rate ranging from 0.25–2 ml?min-1, while the dried toluene went into the USMR through the FEP capillary with the volumetric flow rate between 2.5 and 20 ml?min-1. The flow rate ratio of the precursor solution to the dried toluene was fixed at 0.1.The reprecipitation temperature (20–40 °C) was controlled by immersing two FEP preheating tubing(OD=1.6 mm,ID=0.6 mm,length=150 cm)and the sampling tubing(OD=2 mm,ID=1 mm,length=130–155 cm)into a water bath. The as-obtained colloidal solution of CsPbBr3NCs flowed into a 20 ml sample bottle and was left to stand for 1 h to remove the extremely large nanoparticles. The precipitate on the bottom of the sample bottle was discarded, and the supernatant was kept for further characterization. Because we aimed to study the effects of process parameters(e.g.,ultrasonic power,ultrasonic treatment time, total flow rate) on the particle size distribution of CsPbBr3NCs, we did not use the common purification process involving high-speed centrifugation [5], which could remove the particles with larger size thoroughly and thus hide the effects of different process parameters on the particle size distribution to some extent. Fig. 1(b) shows the optical image of the USMR under irradiation of UV light (365 nm) during the synthesis process. It could be seen that the PL emission experienced a color evolution from blue to green along the microchannel, indicating the formation and growth of the CsPbBr3NCs along the microchannel.By virtue of the mechanical effect of ultrasonic cavitation, the mixing between DMF and toluene was greatly enhanced, triggering the recrystallization of CsPbBr3NCs occurred in a uniform environment (Fig. 1(c)). The ultrasonic power, volumetric flow rates of the precursor solution and dried toluene,amount of water additive,reprecipitation temperature and ultrasonic treatment time were denoted as P, QDMF, Qtoluene, Vwater, T and τultrasound, respectively. It should be pointed out that when studying the effect of the ultrasonic treatment time,USMR with different numbers of glass capillaries(0.21 s for 1USMR,0.43 s for 2USMR,0.64 s for 3USMR,0.86 s for 4USMR) were used. The residence time was kept at 6.85 s by adjusting the length of the sampling tubing for the total flow rate of 11 ml?min-1. In other cases, 4USMR was used for the synthesis of CsPbBr3NCs.

2.5. Synthesis of CsPbBr3 NCs in the batch reactor

The precursor solution containing 50 and 100 μl of water additive and dried toluene were prepared following the same procedure described above, respectively. First, 10 ml of dried toluene was added into a flask immersed in a water bath at 30 °C. Then,1 ml of the precursor solution was quickly injected into the flask under vigorous magnetic stirring. A strong green light emission was observed immediately after injection. The as-obtained colloidal solution of CsPbBr3NCs was left to stand for 1 h to remove the extremely large nanoparticles.

2.6. Characterization

The powder X-ray diffraction (XRD) pattern was recorded using a PANalytical X’pert Pro diffractometer (Netherlands) at a scanning rate of 8 (°)?min-1. The UV–Vis absorption spectrum was measured by a UV-2700 UV–Vis spectrophotometer (Shimadzu, Japan). The PL emission spectrum was recorded by an Ocean InSIGHT FLAME-SUV–Vis–ES spectrometer (Ocean Optics, USA) with a LED laser excitation (365 nm). The PL quantum yield was measured by Quanta Master TM 40 fluorometer (PTI, USA) with an integrated sphere which was excited by a monochromatic 450 nm lamp. The as-obtained colloidal solution of CsPbBr3NCs was diluted five times with dried toluene before measurement of UV–Vis absorption spectrum, PL emission spectrum and PL quantum yield. The transmission electron microscopy (TEM) image was obtained by a JEM-2100 transmission electron microscope (JEOL, Japan).

3. Results and Discussion

3.1. Mixing behavior in USMR

3.1.1. Effect of ultrasonic power

Since the mixing performance in USMR is strongly affected by the reactor structure, properties of the working fluids and operation conditions [24,25], the mixing behavior between DMF and toluene in the USMR developed in this study was studied first.Fig. 2(a) shows the optical images of fluid mixing under the irradiation of ultrasound at different powers. The volumetric flow rates of DMF and toluene were fixed at 1 and 10 ml?min-1,respectively.When no ultrasound was exerted, the reactor could be seen as a coaxial jet mixer. In the coaxial jet mixer, mixing could be enhanced by the shear-force arising from the different velocities of the inner and outer fluids[26].Previous study classified the flow patterns in the coaxial jet mixer into laminar,transition,vortex and turbulence, and turbulent jet regimes, according to the flow rate ratio and the Reynolds(Re)number[27].Under the present condition that the velocities of DMF and toluene being 0.08 and 0.41 m?s-1, respectively, and the Re number being 350, the flow exhibited a turbulent jet behavior. A clear interface between DMF and toluene could be observed at the confluence point, indicating poor mixing. As the fluid moved forward, the interface became unclear, implying DMF was gradually dispersed into toluene. When the ultrasound of 5 W was applied, the interface at the confluence point immediately became disordered. The concentration field of Rhodamine B at 5 W was more uniform than that at 0 W. Meanwhile, several cavitation bubbles (marked with blue circles) appeared in the channel. Under the irradiation of ultrasound, these bubbles formed, oscillated, grew, and collapsed,interrupting the fluid interface, and thus improving the mixing degree. Therefore, the mechanism of enhanced mixing could be ascribed to the mechanical effect arising from ultrasonic cavitation.[17,28]. As the ultrasonic power increased from 5 to 35 W, more cavitation bubbles were generated. Under the vigorous stirring of these bubbles, a more homogenous distribution of Rhodamine B was obtained,implying that higher ultrasonic power was beneficial for mixing.

Fig. 2. (a) Optical images of fluid mixing, (b) mixing index along the channel length, and (c) mixing time at different ultrasonic powers. QDMF = 1 ml?min-1,Qtoluene = 10 ml?min-1.

To quantitatively demonstrate the effect of ultrasonic power on the mixing degree, mixing index (MI) and mixing time (tmix) were calculated (see details in the Supplementary Material). According to Eq.(S1),the higher the mixing index,the better the mixing.Mixing was considered to be relatively homogeneous when the mixing index reached 0.9[29].Fig.2(b)shows the mixing index along the channel length at different ultrasonic powers. When applying ultrasound, the mixing index immediately increased, implying the introduction of ultrasound could improve the mixing between DMF and toluene significantly.When the ultrasonic power reached 35 W,a completely homogeneous mixing could be obtained at the point 0.875 mm away from the confluence point. Mixing time is another key parameter to evaluate the mixing degree of two fluids,and represents the time required for achieving the mixing index of 0.9. Because the mixing index at 0 and 5 W in the photo area did not reach 0.9, the mixing times under these conditions could not be calculated. As shown in Fig. 2(c), the mixing time decreased from 11.2 to 3.8 ms when the ultrasonic power increased from 10 to 35 W.

3.1.2. Effect of total flow rate

The mixing behavior between DMF and toluene at different total flow rates (Re = 80–700) with and without ultrasound were also investigated. As discussed above, two factors mainly affected the mixing performance of USMR developed in this study. One was the shear force between the inner and outer fluids, and the other was the mechanical effect caused by ultrasonic cavitation.When the total flow rate was varied, these two factors both changed, codetermining the mixing degree. Fig. 3(a) shows the optical images of fluid mixing at different total flow rates without ultrasound. It could be seen that when the total flow rate was 2.75 ml?min-1, the flow exhibited a laminar behavior. As the total flow rate reached 5.5 ml?min-1, the flow pattern transferred from laminar flow to turbulent jet flow, and thus better mixing was obtained.Particularly,a homogenous mixing was quickly achieved at 22 ml?min-1. This was because the shear force between the inner and outer fluids increased with the increase in the total flow rate. When 35 W of ultrasound was applied, cavitation bubbles were found in the whole total flow rate range studied in this section (Fig. 3(b)). However, the increasing total flow rate showed an adverse effect on the activity of ultrasonic cavitation, which was determined by the number and oscillation intensity of cavitation bubbles. As shown in Fig. 3(b), when the total flow rate increased, the number of cavitation bubbles decreased. According to the study of Dong et al. [30], there was a build-up time after which cavitation bubbles were formed. This build-up time was caused by the inception and growth of the initial cavitation nuclei.The increase in the total flow rate resulted in a decreasing ultrasonic treatment time. At higher total flow rate, some cavitation nuclei were flushed away from the channel and had not enough time to grow up.The decrease in the number of cavitation bubbles weakened the mechanical effect of ultrasonic cavitation. The mixing times at different total flow rates were also listed in Fig.3.The mixing time monotonously decreased in the range of 2.75–11 ml?min-1.This could be attributed to the synergistic effect of the shear force and ultrasonic cavitation. As the total flow rate increased to 22 ml?min-1, the mixing time slightly increased to 5.1 ms. This was because the highest total flow rate resulted in the strongest shear force but the weakest mechanical effect arising from ultrasonic cavitation.

3.2. Continuous synthesis of CsPbBr3 NCs in USMR

3.2.1. Effect of ultrasonic power

Since the mixing degree between DMF and toluene was greatly influenced by the ultrasonic power, the effect of ultrasonic power was studied first under the condition of QDMF= 1 ml?min-1,Qtoluene=10 ml?min-1,T=30°C,Vwater=100 μl and τultrasound=0.86 ms in the USMR. UV–Vis absorption and PL emission spectroscopy were performed to characterize the optical properties of the colloidal solutions of CsPbBr3NCs synthesized at different ultrasonic powers.Fig.4(a)displays the UV–Vis absorption spectra of the colloidal solutions of CsPbBr3NCs. The absorption spectrum of the sample synthesized at 0 W showed a weak absorption over the wavelength range of 450–520 nm. In contrast, the other samples synthesized under ultrasonic irradiation exhibited strong absorption, which lighted up at ca. 520 nm. As shown in Fig. 4(b), the sample synthesized at 0 W exhibited a weak and broad PL emission peak with the maximum emission at 507 nm.When an ultrasound of 5 W was exerted, the sample showed a more obvious PL emission peak at 516 nm. As the ultrasonic power increased from 5 to 35 W, the intensity of the PL emission peak gradually increased while the peak position was maintained at ca. 516–517 nm. The PL emission peak of CsPbBr3NCs with smaller particle size always located at lower wavelength [31]. Hence, the CsPbBr3NCs synthesized at 0 W had a smaller average particle size than CsPbBr3synthesized under the irradiation of ultrasound. Under UV light with excitation at 365 nm, the colloidal solution synthesized at 0 W exhibited a blue-green photoluminescence, while the other samples emitted bright green light.This observation was in agreement with the PL emission spectra. As is well known, the particle size distribution (PSD) of CsPbBr3can be deduced by the full width at half maximum (FWHM) [31,32]. Generally, the larger the FWHM,the wider the PSD.As shown in Fig.4(c),the FWHM decreased significantly from 39 to 25 nm with the increasing ultrasonic power,inferring that increasing ultrasonic power could narrow the PSD of CsPbBr3NCs. This phenomenon could be ascribed to the diverse mixing degree at different ultrasonic powers. As discussed above,the mixing between DMF and toluene was worst at 0 W, which led to the formation of CsPbBr3NCs with the widest particle size distribution. The largest number of CsPbBr3NCs with large size was settled to the bottom of the sample bottle and discarded at 0 W(observed by naked eyes).This accounted for the lowest intensity of the UV–Vis absorption and PL emission peaks at 0 W.Higher ultrasound power could induce more vigorous cavitation,resulting in a faster mixing process. Thus, a more uniform supersaturation throughout the channel was created, which caused a more homogeneous environment for the nucleation and subsequent growth of CsPbBr3NCs, and a narrower PSD. Subsequently, the optimized sample synthesized at 35 W was characterized by XRD and TEM.As shown in Fig.4(d),all the diffraction peaks in the XRD spectrum could be assigned to the characteristic diffraction peaks of the orthorhombic perovskite phase (JCPDS No. 18–0364), which was in good accordance with the literature [33]. As shown in Fig. 4(e),the TEM image showed that CsPbBr3NCs synthesized at 35 W were composed of nanocubes with an average diameter of (7.99 ± 1.54)nm. The quantum yield of this sample was measured to be 97%.

Fig. 3. The optical images of fluid mixing at different total flow rates at: (a) 0 W and (b) 35 W. QDMF/Qtoluene = 0.1.

Fig.4. (a)UV–Vis absorption spectra,(b) PL emission spectroscopy,and(c) FWHM of the CsPbBr3 NCs synthesized at different ultrasonic powers.(d)XRD spectrum and(e)TEM image of the CsPbBr3 NCs synthesized at 35 W. QDMF = 1 ml?min-1, Qtoluene = 10 ml?min-1, Vwater = 100 μl, T = 30 °C,τultrasound = 0.86 s.

3.2.2. Effect of the ultrasonic treatment time

The effect of the ultrasonic treatment time on the synthesis of CsPbBr3NCs was studied. USMRs with different numbers of glass capillaries (0.21 s for 1USMR, 0.43 s for 2USMR, 0.64 s for 3USMR,and 0.86 s for 4USMR) were used. The residence time was kept unchanged by varying the length of the sampling tubing.As shown in Fig.5(a),as the ultrasonic treatment time decreased from 0.86 to 0.21 s, the PL emission peak showed a slight red shift from 517 to 519 nm.Considering the quantum size effect,the slight red shift of the PL emission peak implied an increase in the size of CsPbBr3NCs.Meanwhile,the PL emission peak of CsPbBr3NCs synthesized at shorter ultrasonic treatment time became more asymmetric.As shown in Fig.5(b),the FHWM showed no obvious change with the decrease in the ultrasonic treatment time. This was because that DMF and toluene were mixed homogenously at the point 0.875 mm away from the confluence point under the condition of P=35 W,QDMF=1 ml?min-1and Qtoluene=10 ml?min-1.Further increase in the ultrasonic treatment time showed no effect on the mixing between DMF and toluene. However, as displayed from Fig. 2(b), the color evolution in the second, third and fourth glass capillary could be clearly observed, indicating the growth of CsPbBr3was not completed within 0.86 s. Hence, whether the growth of CsPbBr3NCs occurred under ultrasound showed a slight effect on the PL properties of CsPbBr3NCs. Based on the experimental results, we deduced that the growth of CsPbBr3NCs under ultrasound was beneficial to prepare CsPbBr3NCs with a slightly smaller particle size.

3.2.3. Effect of the total flow rate

The effect of the total flow rate was studied in the USMR with the ultrasonic power of 35 W. For comparison, CsPbBr3NCs were also synthesized in the same reactor without ultrasound irradiation. Fig. 6(a) shows the PL emission spectra of the colloidal solutions of CsPbBr3NCs synthesized at different total flow rates with and without ultrasound. The samples obtained at 2.75 and 5.5 ml?min-1without ultrasound were not diluted with toluene because of the low product yield caused by very poor mixing between the reactants. When no ultrasound was exerted, the PL emission peaks experienced a small red-shift with the increase of the total flow rate,while the corresponding FWHM of PL emission decreased significantly from 52 to 25 nm (Fig. 6(b)). This was because increasing the total flow rate improved the mixing between DMF and toluene (already discussed above), and in turn the uniformity of CsPbBr3NCs. When 35 W of ultrasound was applied, the FWHM of the CsPbBr3NCs synthesized at 2.75–11 ml?min-1all decreased significantly as compared to those without ultrasound, demonstrating the advantage of USMR. At 22 ml?min-1, the effect of ultrasound could be neglected, likely because of the good mixing between DMF and toluene even without ultrasound. As shown in Fig. 6(b), the FWHM of the CsPbBr3NCs synthesized at 35 W showed no obvious change with the total flow rate increasing from 2.75 to 22 ml?min-1.

Fig. 5. (a) The PL emission spectroscopy and (b) FWHM of the CsPbBr3 NCs synthesized at different ultrasonic treatment time. P = 35 W, QDMF = 1 ml?min-1, Qtoluene = 10 ml?min-1, Vwater = 100 μl, T = 30 °C.

Fig.6. (a)The PL emission spectroscopy and(b)FWHM of the CsPbBr3 NCs synthesized at different total flow rates with 35 W of ultrasound(solid)and without ultrasound(dash). QDMF/Qtoluene = 0.1, Vwater = 100 μl, T = 30 °C,τultrasound = 0.86 s.

Fig. 7. (a) The PL emission spectroscopy and (b) FWHM of the CsPbBr3 NCs synthesized with different water additive. P = 35 W, QDMF = 1 ml?min-1, Qtoluene = 10 ml?min-1,T = 30 °C,τultrasound = 0.86 s.

3.2.4. Effect of water additive

It has been widely found that the morphology and particle size of CsPbBr3NCs were greatly affected by the sorption–desorption equilibrium of the ligands, which could be tuned by the introduction of polar solvents[34,35].For example,Udayabhaskararao et al.[36] found that the addition of ethanol resulted in the fast formation of ultrathin long nanowires(hundreds of nm)along with large nanocubes (tens of nm). Therefore, the effect of water which was the most common polar solvent was studied on the synthesis of CsPbBr3NCs in the USMR. The ultrasonic power was fixed at 35 W to minimize the effect of uneven mixing.Other experimental conditions were set as QDMF= 1 ml?min-1, Qtoluene= 10 ml?min-1,T=30°C,and τultrasound=0.86 s.As shown in Fig.7,the PL emission spectrum of the colloidal solution synthesized without water additive exhibited a broad peak with the maximum emission located at ca. 511 nm and a FHWM of 39 nm. When the amount of water additive increased from 0 to 200 μl, the PL emission spectra experienced a gradual red-shift from 511 to 518 nm, while the corresponding FWHM decreased from 39 nm to 22 nm (Fig. 7(b)),indicating a narrower PSD of CsPbBr3NCs. Fig. S3 showed the TEM images of CsPbBr3NCs synthesized with different amounts of water additive. As shown in Fig. S3(a) and (b), the CsPbBr3NCs synthesized under waterless condition exhibited a combination of amorphous nanoparticles (marked with red circles), nanocubes((6.97± 1.38)nm in edge length) and extremely large-sized nanocubes (larger than 100 nm). The CsPbBr3NCs synthesized with 50 μl of water additive were also composed of nanospheres ((2.2 5 ± 0.69) nm in size, Fig. S3(c)), nanocubes ((7.09 ± 1.47) nm in edge length, Fig. S3(d)) and extremely large-sized nanocubes (larger than 100 nm, Fig. S3(e)). When the amount of water additive increased to 100–200 μl,all the as-prepared CsPbBr3NCs exhibited a cubic morphology (Fig. S3(f)–(h)). The average edge lengths of CsPbBr3NCs synthesized with 100, 150, and 200 μl of water additive were(7.99±1.54),(8.91±1.63),and(10.12±1.98)nm,respectively.In addition,a small number of nanocubes(Fig.S3(i))with an average edge length of(21.73±6.21)nm was present in the 200 μl sample. When not considering the extremely large-sized nanocubes,the increase in water additives led to an increase in the size of CsPbBr3NCs. This was the reason why the PL emission peak showed a redshift with the increasing amount of water additive.

When water was added into the reaction system, two changes occurred. First, the surface coverage of OA–(OLA+) on CsPbBr3NCs decreased. As is well known, OA and OLA are the most frequently used ligands, but are easily desorbed from the surface of CsPbBr3NCs by the polar solvents [35]. When water was introduced, water molecules themselves could produce H3O+and OH–.H3O+(or OH–) could destabilize OA–(or OLA+) on the surface of CsPbBr3NCs [37]. It could be speculated that the surface coverage of OA–(or OLA+) decreased with the increasing amount of water additive, which could accelerate the oriented attachment of CsPbBr3NCs.Second,the introduction of water increased the solubility of CsPbBr3,causing a lower degree of supersaturation in comparison with the waterless condition. Low supersaturation always resulted in the decrease of nucleation rate,and in turn large crystal size.With respect to the formation mechanism of extremely largesized nanocubes under waterless and 50 μl conditions, a possible explanation could not be proposed at present. The related study is underway.

3.2.5. Effect of reprecipitation temperature

Reprecipitation temperature significantly influences the nucleation and growth rate of CsPbBr3NCs [38]. Therefore, the effect of reprecipitation temperature was investigated. As shown in Fig. S4, the temperature had no obvious effect on the mixing performance. Fig. 8 show the PL emission spectra and corresponding FWHM of the colloid solutions of CsPbBr3NCs synthesized at different temperatures. It was clear that there was a gradual redshift of the PL emission peak with the increase of reprecipitation temperature, while the FHWM decreased. Fig. S5(a) and (b)showed the TEM images of CsPbBr3NCs synthesized at 20 and 40 °C, respectively. For CsPbBr3NCs synthesized at 20 °C, regular nanocubes ((7.35 ± 2.30) nm in edge length) and nanoplatelets((4.15 ± 0.88) nm in width, (17.62 ± 2.65) nm in length) coexisted, which accounted for the broader PL emission peak.According to the literature, OA and OLA can also serve as the soft template for the formation of nanoplatelets besides using as the capping agent [36]. At low temperatures, energy is not large enough to dissociate more ligands from the nanoparticles, triggering the formation of the low dimensional structure. Hence, nanoplatelets were observed in the sample synthesized at 20 °C. The CsPbBr3synthesized at 40 °C was mainly composed of regular nanocubes with (8.76 ± 1.70) nm in edge length. With respect to the nanocubes, the increase in the reprecipitation temperature induced a larger size.This could explain why the PL emission peak showed a redshift. When the reprecipitation temperature increased, the solubility of CsPbBr3increased, reducing the degree of supersaturation for nucleation. Therefore, smaller number of nuclei were generated, causing more monomers consumed in the growth process to produce larger NCs. In addition, the mobility of monomers was larger at high temperatures, which was beneficial for the synthesis of larger NCs. What is more, the surface ligands could be more dynamic, and the surface coverage was decreased with the increase of precipitation temperature, which could accelerate the growth of CsPbBr3NCs.

Fig. 8. (a) The PL emission spectroscopy and (b) FWHM of the CsPbBr3 NCs synthesized at different temperatures. P = 35 W, QDMF = 1 ml?min-1, Qtoluene = 10 ml?min-1,Vwater = 100 μl,τultrasound = 0.86 s.

3.2.6. Comparison between the USMR and batch reactor and PL adjustability of USMR-based method

The performances of the USMR and batch reactor in the synthesis of CsPbBr3NCs were compared in terms of the peak position,FWHM, PL intensity, and repeatability. The experimental procedure was repeated five times. As shown in Table 1, the average peak position of CsPbBr3NCs synthesized in the batch reactor was smaller than that of CsPbBr3NCs synthesized at 35 W, while the FWHM of CsPbBr3NCs synthesized in the batch reactor was comparable with that of CsPbBr3NCs synthesized at 35 W.The volume ratio of DMF to toluene was varied as the dropwise addition proceeded for the batch reactor, while it was kept constant in the USMR.This might be the reason for the difference in the peak position between the samples synthesized in the batch reactor and USMR. The PL intensity of CsPbBr3NCs synthesized in the batch reactor was lower than that of CsPbBr3NCs synthesized at 35 W,which may be caused by the lower product yield or PL quantum yield. Furthermore, the standard deviations for the peak position and FWHM of the CsPbBr3NCs synthesized in the USMR were smaller than those of CsPbBr3synthesized in the batch reactor,indicating the repeatability of the USMR was better than that of the batch reactor. This could be attributed to the excellent mixing performance of USMR,which could offer a more uniform reprecipitation environment.In addition,the USMR can flexibly control the PL color by modulating the composition of cesium lead halide perovskite NCs.As shown in Fig.9,cesium lead halide perovskite NCs with different halide compositions which covered a wide visible spectrum(426–661 nm)were successfully synthesized.The experimental details were listed in the Supplementary Material.

Table 1 The PL peak position,FWHM,peak intensity and corresponding standard deviations of CsPbBr3 NCs synthesized in the USMR (35 W) and batch reactor at 30 °C

Fig. 9. The Optical images under a 365 nm UV lamp and PL spectra of cesium lead halide perovskite NCs with different halide compositions: (1) CsPbCl3, (2)CsPbClBr2, (3) CsPbBr3, (4) CsPbBr2I, (5) CsPbBr1.5I1.5 and (6) CsPbBrI2.

4. Conclusions

In summary, a USMR was developed to realize the continuous synthesis of CsPbBr3NCs via LARP. The mixing performance in USMR was co-determined by the shear force between the inner and outer fluids,and mechanical effect caused by ultrasonic cavitation. By virtue of the excellent mixing performance of the USMR,CsPbBr3NCs with good photoluminescence properties were synthesized. The influences of ultrasonic power, ultrasonic treatment time, total flow rate, water additive, and reprecipitation temperature were studied in detail. The higher ultrasound power could induce a faster mixing process, resulting in a more homogeneous environment for the nucleation and subsequent growth of CsPbBr3NCs, and thus a narrower PSD. When 35 W of ultrasound was applied, the ultrasonic treatment time and total flow rate showed no obvious effect on the synthesis of CsPbBr3NCs. Increasing the amount of water additive and reprecipitation temperature could induce a red-shift of PL emission peak and changes in the size and morphology of CsPbBr3NCs. The optimized sample synthesized at 35 W exhibited a PL emission peak position at ca. 517 nm with a full width at half-maximum of 25 nm and a quantum yield of 97%.Compared to the batch method,the continuous method based on USMR showed an obvious advantage in experimental repeatability.

Data Availability

Data will be made available on request.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We would like to acknowledge the financial supports from National Natural Science Foundation of China (22178336 and 21991103).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2022.12.001.