Direct photocatalytic patterning of colloidal emissive nanomaterials

Abstract

We present a universal direct photocatalytic patterning method that can completely preserve the optical properties of perovskite nanocrystals (PeNCs) and other emissive nanomaterials. Solubility change of PeNCs is achieved mainly by a photoinduced thiol-ene click reaction between specially tailored surface ligands and a dual-role photocatalytic reagent, pentaerythritol tetrakis(3-mercaptopropionate) (PTMP), where the thiol-ene reaction is enabled at a low light intensity dose (~ 30 millijoules per square centimeter) by the strong photocatalytic activity of PeNCs. The photochemical reaction mechanism was investigated using various analyses at each patterning step. The PTMP also acts as a defect passivation agent for the PeNCs and even enhances their photoluminescence quantum yield (by ~5%) and photostability. Multicolor patterns of cesium lead halide (CsPbX₃)PeNCs were fabricated with high resolution (<1 micrometer). Our method is widely applicable to other classes of nanomaterials including colloidal cadmium selenide–based and indium phosphide–based quantum dots and light-emitting polymers; this generality provides a nondestructive and simple way to pattern various functional materials and devices.

A universal photocatalytic method allows high-resolution patterning of various nanomaterials, preserving their optical properties.

INTRODUCTION

Colloidal metal halide perovskite nanocrystals (PeNCs) are being evaluated as solution-processable light-emitting materials promising for various displays and optoelectronic applications including color conversion layers, light-emitting diodes (LEDs), and photodetectors (1–3). PeNCs have a high photoluminescence (PL) quantum yield (PLQY) close to unity, facile and wide color tunability by halide anion exchange, and exceptional color purity owing to their narrow emission spectra with full width at half maximum of ~20 nm, which is much smaller than those of inorganic quantum dots (QDs) (25 to 40 nm) or organic emitters (≥40 nm) (4, 5). Because of their unique optical characteristics, PeNCs have been regarded as highly suitable light emitters for potential use in next-generation displays that provide immersive experiences featuring vivid and realistic contents (1–7).

To implement PeNCs in these realistic displays, red-green-blue (RGB) full-color pixel arrays that incorporate the PeNCs should be fabricated and integrated into the display panels using a patterning method tailored for each material. Therefore, development of patterning methods that can locate the light emitters precisely and uniformly in desired positions is imperative. Various methods [e.g., inkjet printing (8), transfer printing (9), and photolithography (10, 11)] have been developed to create high-quality PeNC patterns, but none of them satisfy all the conditions for an ideal pattern (text SA).

Direct optical lithography is an alternative patterning technique that can produce high-resolution and uniform patterns of colloidal inorganic nanocrystals with a reduced number of process steps while retaining all of the advantages of conventional photolithography (e.g., well-established manufacturing infrastructure) (12–15). In direct optical lithography, patterns are formed by photoinduced chemical reactions of photosensitive ligands or additives, which result in a solubility change of nanomaterials [e.g., metal oxide (16), PeNCs (17), and QDs (12)] at light-exposed regions without the need for a polymeric photoresist. Various photosensitive motifs and photochemical reactions such as azide (17–21), benzophenone (22), cinnamoyl (23), photo-acid generation (12), oxime sulfonate (13), thiol-ene (24–27), photo-amine generation (28), oxide bridging (16), photo-oxidation (29), and alkene cross-linking (30) have been explored for direct optical patterning of colloidal emissive nanocrystals. However, the lithography process often induces substantial surface damage to the deposited emissive nanomaterials, thereby degrading their optical properties such as PLQY (13, 17). In PeNCs, the loss in PLQY is especially problematic because of the weak ionic bonding nature and labile surface ligands (1, 4). Although posttreatment processes can partially mitigate this problem, additional processing is not desirable in industrial applications (13, 17). Overcoming these demerits requires the development of a photopatterning approach that provides high resolution and uniformity but does not sacrifice the optical properties of PeNCs.

Here, we report a universal direct photocatalytic patterning technique for PeNCs and other emissive nanomaterials using thiol-ene click chemistry. Our direct photocatalytic patterning uses a multifunctional click chemistry reagent, pentaerythritol tetrakis(3-mercaptopropionate) (PTMP) that has four thiol end groups (─SH), as both a ligand cross-linker and a defect-passivation agent. During light exposure, excitons generated from PeNCs photocatalytically assist the formation of thiyl radicals and thereby enable the thiol-ene click reaction between PTMP and the alkene group (C═C) of alkylammonium ligands at a low dose (~30 mJ cm⁻²) of a 275 nm light source. In addition, the deprotonated PTMP acts as a surface ligand for the PeNCs. Consequently, the sub–1 μm–scale high-resolution patterns of PeNCs were obtained, and their PLQY and photostability were even enhanced, as a result of effective cross-linking and defect passivation by PTMP without any additional posttreatment process. RGB PL patterns of PeNCs and electroluminescence (EL) patterns of PeNC LEDs were also demonstrated as a proof of concept to confirm the potential of our method toward practical applications. Moreover, direct photocatalytic patterning is a universal method that can be used to pattern various light emitters (e.g., CsPbX₃, CdSe QDs, InP QDs, and conjugated polymers) without degrading their optical properties. Our direct photocatalytic patterning presents a big step toward the successful implementation of a broad range of semiconducting materials for various optoelectronic devices and circuits.

RESULTS

Direct photocatalytic patterning: Concept and mechanism

Conventional direct optical patterning of PeNCs often damages their optical properties due to the concomitant ligand exchange and ultraviolet (UV)–induced photochemical degradation during patterning procedures (13, 17). One approach to mitigating the damage of PeNCs is to use photopolymerization and in situ crystallization upon annealing (24). However, the polymerization process requires a high intensity of light (~60 J cm⁻²) (24), and achieving precise size and morphology control of PeNCs can be challenging. Furthermore, PeNC-polymer composites may be unsuitable for EL devices due to the insulating nature of polymers, and a low emitter density within the composite may place a restriction on PL down-conversion applications.

To overcome these limitations, we developed a nondestructive patterning method that exploits the photocatalytic activity of PeNCs and thiol-ene click reaction between their surface ligands (Fig. 1A). The thiol-ene reaction is a useful click chemistry whereby a thiol group (─SH) interacts with an alkene group to form a single bond between a sulfur atom and a carbon atom (31). The reaction has various advantages including high yield, insensitivity to water and oxygen, and high mechanical strength by covalent bond formation and is therefore suitable for direct optical patterning of emissive materials. Instead of using the conventional method of thiol-ene photopolymerization, we used a ligand cross-linking strategy through PeNC-induced thiol-ene reactions, where the alkene functional group (C═C) is present in the conventional alkyl chain ligands [e.g., oleic acid (OA) and oleylamine (OLA)] of the PeNCs. As the PeNCs already contain an alkene moiety, a thiol-bearing moiety can simply be added to induce a solubility change through ligand cross-linking via UV exposure. The PeNCs serve as an efficient photocatalyst that effectively assists the generation of thiyl radicals from thiol groups, thereby enabling thiol-ene cross-linking reactions (32, 33).

Fig. 1. Schematic illustrations of the direct photocatalytic patterning process.

(A) Schematic of the direct photocatalytic patterning concept. A photo-patternable ink is produced by mixing perovskite nanocrystal (PeNC) and pentaerythritol tetrakis(3-mercaptopropionate) (PTMP), and upon exposure to UV light, PeNCs act as a photocatalyst, effectively generating thiyl radicals that initiate thiol-ene reactions with the alkene groups of oleylammonium ligands (red double line). This cross-linking reduces the solubility of PeNCs in nonpolar solvents (developer). PTMP can also act as a surface ligand and thereby passivate surface defects. (B) The direct photocatalytic patterning process and a zebra pattern of CsPbBr₃ PeNCs. Scale bar, 500 μm.

To efficiently generate patterns using a cross-linker, it is important to consider both the number of reactive functional groups and compatibility with colloidal PeNCs. Cross-linking adjacent PeNC ligands require at least two thiol functional groups located on opposite sides (26), and a higher number of reactive functional units (e.g., multiple azide groups) can increase cross-linking efficiency (19). In particular, PeNCs are susceptible to degradation in polar solvents (34), so the cross-linking agent should have high compatibility with nonpolar solvents. Considering these requirements, we identified PTMP as a suitable cross-linker for our direct photocatalytic patterning strategy. PTMP has four thiol end groups and dissolves well in nonpolar solvents.

PTMP has been used as a reagent for thiol-ene reactions in various applications, including QD-embedded composites (35), in situ crystallized perovskite-polymer composites (36), and thermal cross-linker in perovskite solar cells (37). The chemical structure of PTMP was confirmed using ¹H nuclear magnetic resonance analysis and Raman spectroscopy (fig. S1). UV-visible absorption spectra of two PTMP/toluene solutions (0.025 and 2.5 M) do not show distinct absorption characteristics for the wavelengths of the i-line (365 nm) and h-line (405 nm) (fig. S2). This result means that the thiol groups of PTMP were not transformed to thiyl radicals solely by UV irradiation, confirming the necessity of an appropriate photocatalyst (24, 31).

The PeNC pattern is produced in four steps: (i) mixing PeNC dispersion with an optimized amount of PTMP solution, (ii) spin-coating PeNC-PTMP mixture onto the substrate (glass or Si wafer), (iii) alignment of the photomask and light exposure, and (iv) removal of PeNCs in unexposed areas by a development process (Fig. 1B and details in Materials and Methods).

The chemical reactions between the surface ligands of PeNCs and PTMP during the direct photocatalytic patterning process were investigated using Raman spectroscopy, Fourier transform infrared (FTIR) spectroscopy, x-ray photoelectron spectroscopy (XPS), and electron paramagnetic resonance (EPR) spectroscopy (Fig. 2). In Raman spectra, the presence of PeNCs is indicated by the strong band at 1664 cm⁻¹ [stretching of alkene group (C═C) in oleylammonium ligands], whereas the presence of PTMP is indicated by a band at 2574 cm⁻¹ (stretching mode of thiol group S─H) and another at 1737 cm⁻¹ [stretching mode of ketone group (C═O)] (Fig. 2A and fig. S3). After UV irradiation and development, the intensity of thiol and alkene stretching modes decreased notably; this change confirms that the thiol-ene reaction has proceeded. The decreased intensities of thiol and ketone stretching modes upon patterning imply that almost all unreacted PTMP molecules were removed because of their high solubility in toluene (developer). The weak ketone stretching mode remaining after development was ascribed to the residual PTMP molecules acting as cross-linkers and surface ligands (fig. S3A). FTIR analysis detected the similar tendencies in the stretching modes of the alkene group (C═C), thiol group (S─H), and ketone group (C═O) (fig. S3, B to D). These consistent results demonstrate the occurrence of the thiol-ene reaction and further confirm the effective removal of unreacted PTMP molecules after development (fig. S3).

Fig. 2. Reaction mechanism of direct photocatalytic patterning.

(A) Raman spectra of a CsPbBr₃ perovskite nanocrystal (PeNC)–pentaerythritol tetrakis(3-mercaptopropionate) (PTMP) film before UV exposure (red) and after patterning (green). (B) X-ray photoelectron spectroscopy (XPS) S 2p spectra of a pristine CsPbBr₃ PeNC film and PeNC-PTMP films at each patterning step. a.u., arbitrary units. (C) X-band electron paramagnetic resonance (EPR) spectra of a PeNC-PTMP solution before and after UV irradiation. 5,5-dimethyl-1-pyrroline N-oxide (DMPO) was used as a spin trap. To increase the intensity of the measured values, an excess amount of PTMP (100 wt %) was added, and the measurements were carried out in a nitrogen atmosphere. (D) Schematic energy level diagram (left) and proposed thiol-ene reaction scheme (right) during the direct photocatalytic patterning process. CBM, conduction band minimum; VBM, valence band minimum.

XPS S 2p spectra (Fig. 2B) determined the change in the form of the thiol group of PTMP. As expected, the S 2p spectrum showed no signal in the pristine PeNC film but appeared in PeNC-PTMP mixture films due to the presence of PTMP. After deconvolution of each S 2p peak, the PeNC-PTMP mixture exhibited peaks at 164.53 and 163.25 eV (S 2p_1/2 and S 2p_3/2, respectively), indicating the presence of C─S bonds of PTMP. Upon UV exposure, the thiol-ene reaction resulted in the formation of C─S─C bonds, resulting in a slight shift of the deconvolution peak to lower binding energies (164.42 and 163.15 eV for S 2p_1/2 and S 2p_3/2, respectively). During the development process, unreacted PTMP was removed, resulting in a reduction in the intensity of the S 2p peak. Because the PTMP involved in cross-linking remained, the binding energy substantially decreased (163.75 and 162.82 eV for S 2p_1/2 and S 2p_3/2, respectively) (38). These results also provide evidence that PTMP was retained as the cross-linker in patterned films.

EPR analysis was conducted to clarify the function of PeNCs as a photocatalyst. Radical-forming reactions are highly unstable, so 5,5-dimethyl-1-pyrroline N-oxide (DMPO) was used as a spin trap. The samples were prepared under N₂ atmosphere, and a light source of 365 nm with a lamp intensity of 40 mW cm⁻² was used. Whereas the EPR spectrum of the PeNC-PTMP mixture under dark condition showed a low intensity, strong radical signals were generated after UV exposure; comparison with the simulated spectrum confirmed the origin of the signals to be thiyl radical (Fig. 2C) (32). These results indicate that the presence of PeNC itself does not simply act as a catalyst but that light-stimulated formation of excitons in PeNC notably contributes to the formation of radicals (32, 33).

The high reactivity of radicals can cause unintended side reactions, so control experiments were conducted to identify the exact patterning mechanism. It has been reported that PeNCs can also be used as a photocatalyst in the formation of disulfide bonds (33). To identify the exact reaction pathways involved in the patterning process, we investigated the patterning capability of PeNCs capped with various ligands (4-dodecylbenzenesulfonic acid, tetraoctylammonium bromide, octadecylphosphonic acid, didodecyldimethylammonium bromide, and octanoic acid), which do not have alkene groups that are essential for the thiol-ene reaction (fig. S4). After synthesizing the PeNCs with these alkene-free ligands, each PeNC dispersion was mixed with 10 and 100 weight % (wt %) of PTMP and then patterned with strong light exposure (365 nm, 400 mJ cm⁻²). While the alkene-bearing oleylammonium-capped PeNCs with 10 wt % of PTMP yielded clear, high-resolution patterns without background residue (fig. S4A), the PeNCs with alkene-free ligands could not form any patterns with 10 wt % of PTMP and showed only obscure patterns at best with 100 wt % (fig. S4, B and C). This suggests that the primary mechanism for our photocatalytic patterning strategy is the thiol-ene reaction between the alkene groups in the ligands and the thiol groups in the PTMP, rather than the formation of disulfide bonds. The thiol-ene mechanism was further supported by the changes in FTIR spectra. Upon the addition of PTMP and subsequent UV exposure, the alkene (C═C) stretching mode of oleylammonium ligands disappeared because of thiol-ene reaction. In contrast, the FTIR spectra of alkene-free ligand-capped PeNCs exhibited no notable changes except the emergence of the C═O stretching mode from PTMP (fig. S4D). This implies that the alkene-free ligands did not participate in the photochemical reactions during the patterning process. On the basis of these observations, we propose that the patterning mechanism for alkene-free ligand-capped PeNCs involves the formation of disulfide bonds between PTMP molecules. The limited patternability observed in fig. S4 (B and C) can be attributed to the following reasons: (i) The kinetics of disulfide bond formation is slower than that of thiol-ene reaction (39), and (ii) the disulfide formation between PTMP molecules appears to create a physical molecular network where PeNCs are simply physically entangled. Consequently, we conclude that the thiol-ene reaction is the primary mechanism governing this photocatalytic patterning method.

On the basis of the experimental results, we propose the following model for the photocatalytic thiol-ene reaction (Fig. 2D). When a PeNC is irradiated by photons of energy greater than its bandgap, excitons or electron-hole pairs are formed. If the redox potential of PTMP is energetically lower than the valence band maximum of PeNC, then the holes can move to the thiol group of PTMP, leaving reactive thiyl radicals. Note that the redox potential of typical thiol compounds is higher than the valence band maximum of PeNC (32, 40). This model suggests that other emissive nanomaterials that satisfy the photocatalytic reaction conditions can also be patterned; this possibility will be confirmed in subsequent sections of this paper.

Strategies to preserve optical properties of PeNCs upon patterning

For practical applications, the patterning process must not degrade the optical properties of PeNCs and achieve high resolution. However, direct optical patterning dominantly depends on ligand chemistry, so detachment of ligands during the patterning process often generates defects that adversely affect the passivation of PeNCs and thereby decrease the PLQY of the final patterns.

We solved this problem by introducing multifunctional PTMP additive molecules. The PTMP not only acted as a cross-linker enabling patternability but also served as an effective defect-passivation agent, providing improved PLQY and colloidal/photostability of PeNCs. However, the thiol moiety does not always act as a defect passivation agent in PeNCs and can sometimes have a detrimental effect depending on the surface of the PeNCs. To prevent such a detrimental effect, we precisely controlled the surface of PeNCs and optimized the patterning process to regulate the extent of ligand exchange with PTMP. As a result, the formation of surface defects during the patterning process was effectively suppressed, and, thereby, the optical properties of PeNC patterns were completely preserved or even enhanced.

High-stability colloidal PeNCs were synthesized by incorporation of ZnBr₂ during the synthesis. As the zinc ion preferentially binds to oleate, the resulting PeNCs contain alkylammonium bromide, which has a high binding affinity, as a predominant ligand (fig. S5) (41, 42). The strong binding of ammonium ligands greatly increased the PLQY of CsPbBr₃ PeNCs to 70 to 80% (Fig. 3B), compared to conventional PeNCs capped with OA and OLA, which have a PLQY of 40 to 50%. These results are consistent with previous reports in the literature (41, 42). The specially designed oleylammonium-capped CsPbBr₃ PeNCs were used as both the photocatalyst and patterning substance throughout the patterning process.

Fig. 3. Preservation of optical properties of CsPbBr₃ perovskite nanocrystals (PeNCs) upon the patterning process.

(A) Absorption and photoluminescence (PL) spectra of a CsPbBr₃ PeNC solution (black) and PeNC–pentaerythritol tetrakis(3-mercaptopropionate) (PTMP) solution (red). Curves offset vertically for clarity. (B) Absolute PL quantum yield (PLQY) of a pristine PeNC film and a PeNC-PTMP film at each patterning step. (C) PL lifetime curves and (D) X-ray photoelectron spectroscopy (XPS) Pb 4f spectra of pristine PeNC films and PeNC-PTMP films at each patterning step. All samples were prepared under the same condition: 10 wt % of PTMP and UV (365 nm) dose of 400 mJ cm⁻². (E) Schematic diagram showing the effect of initial surface ligands on the passivation of PeNCs with PTMP ligands. (F) Photostability of pristine and patterned PeNC films.

To ensure that the PTMP does not degrade the optical properties of PeNCs, we obtained steady-state PL and UV-visible spectra of colloidal oleylammonium-capped CsPbBr₃ PeNCs (Fig. 3A). The peak positions of absorption and emission spectra (500 and 508 nm, respectively) did not shift after addition of PTMP. In addition, the x-ray diffraction patterns of CsPbBr₃ PeNC films and the transmission electron microscopy images of CsPbBr₃ PeNCs showed that the patterning process did not change the crystal structure and average size of PeNCs (~8.2 nm) (figs. S6 and S7).

The absolute PLQY of PeNC films was completely preserved throughout the entire patterning process, and it even increased with the addition of PTMP (Fig. 3B). The initial PLQY of pristine PeNC dispersion was ~80%, which is consistent with previous research (42), while the PLQY of pristine PeNC films was measured to be ~67%. During the formulation, UV exposure, and development stages, the PLQY of the PeNC films increased by 7, 9, and 5%, respectively, compared to the pristine PeNC film.

To further support this observation, time-resolved PL analysis was conducted at each patterning step. The average lifetime trend exhibited a strong correlation with the PLQY results (Fig. 3C). The average PL lifetime increased from 3.5 to 6.3 ns upon the inclusion of PTMP as the proportion of nonradiative recombination decreased (table S1). These results provide evidence that the addition of PTMP contributed to improved surface passivation and enhanced optical properties.

The enhancement in optical properties can be attributed to the strong affinity between sulfur and lead (43), which can be observed not only in the film state but also in the solution (dispersion) state. By adding PTMP under various conditions (e.g., shelf time after formulation, UV exposure duration, and PTMP concentration), we observed that the optical properties of the PeNC dispersion were enhanced (fig. S8), probably because of the surface passivation of PTMP.

To confirm the presence of PTMP functioning as a ligand, we investigated the variations in surface properties of the PeNCs. Upon addition of PTMP, the XPS Pb 4f peak shifted by ~0.1 eV to lower binding energy (43), and an S 2p peak appeared; these changes indicate passivation of undercoordinated surface Pb atoms by thiolate ligands (Fig. 3D). In addition, the zeta-potential of colloidal PeNCs gradually shifted to a lower value after PTMP addition and UV exposure (fig. S9). Note that the initial surface charge was positive because of the oleylammonium ligands. This apparent decrease in surface charge provides additional evidence supporting the role of thiolate as a ligand, along with its electron-donating ability.

The improvement of optical properties upon the addition of PTMP and the patterning process was only observed in oleylammonium-capped PeNCs, whereas the optical properties were degraded in OA-OLA–capped PeNCs. To elucidate these observations, we have formulated the following hypothesis: In pristine OA-OLA PeNCs, PTMP acts as an acid through deprotonation, facilitating ligand detachment and exchange. Consequently, upon the addition of PTMP, the PL intensity initially decreased substantially because of the detachment of OA; however, over time, we observed a partial recovery of the optical properties, which was attributed to the slow attachment and exchange process of PTMP ligands. However, because PTMP is a relatively short-chain ligand with more binding sites compared to OA or OLA, the PL and colloidal stability were reduced when the PTMP was added (44). In addition, excessive ligand exchange into PTMP caused the PeNCs to agglomerate and altered their solubility, resulting in a significant amount of substantial residual PeNCs in the background after the patterning process (fig. S10). On the other hand, oleylammonium-capped PeNCs have strong ligand binding (41, 42), and because of the limited availability of ligands capable of being protonated by PTMP even when it acts as an acid, the degree of ligand exchange by PTMP is reduced. In this case, PTMP primarily acts as an additional passivation agent for surface defects (most probably undercoordinated Pb atoms), leading to the enhanced PLQY (Fig. 3E).

The effective surface passivation by dual-function PTMP also increased the colloidal stability and the photostability of patterned PeNC film (text SB and fig. S11). When irradiated with high-power UV (365 nm, 40 mW cm⁻²), the patterned PeNC film retained PLQY better than the pristine PeNC film (Fig. 3F). The high stability of the patterned PeNCs shows that this nondestructive and efficient patterning method is suitable for various optoelectronic devices.

PL and EL patterns of RGB PeNCs

The direct photocatalytic patterning strategy was used to fabricate various high-resolution patterns of PeNCs with high uniformity (Fig. 4, A and B). Uniform 1-μm-wide line patterns were successfully demonstrated using the oleylammonium-capped CsPbBr₃ PeNCs with an optimized 10 wt % of PTMP (Fig. 4C). The conditions for high-resolution line patterns were further optimized by varying the parameters, including PTMP concentration, UV dose, and excitation wavelength (text SC and fig. S12). With optimization, we achieved a line pattern of PeNCs with a feature size as small as 560 nm. These high-resolution patterns correspond to a pixel density of 12,700 pixel per inch (ppi), which is highly suitable for augmented reality applications.

Fig. 4. Optical microscope (OM)and atomic force microscopy (AFM) images of perovskite nanocrystal (PeNC) patterns.

(A) Fluorescence OM images and (B) Photoluminescence (PL) intensity profiles of green PeNC patterns with 254 and 2540 ppi resolution. (C) A uniform 1-μm-wide and 560-nm-wide (inset) line pattern of PeNCs. This corresponds to 12,700 ppi resolution. (D) Pattern arrays of PeNCs including heart, donut, circle, and triangle shapes. Scale bars, 100 μm. (E) AFM height profile (top), three-dimensional profile (middle), and area profile of yellow dashed area (bottom) of a 10-μm-wide line pattern of PeNCs. Cross patterns of (F) red CsPbI₃, (G) green CsPbBr₃, and (H) blue CsPbBr₃ PeNCs.

Arrays of various shapes could also be demonstrated (e.g., hearts, rings, dots, and triangles; Fig. 4D); these results demonstrate the versatility of our photocatalytic patterning method. Atomic force microscopy (AFM) analysis was conducted on 10-μm-wide line patterns to reveal the morphology of the fabricated patterns (Fig. 4E). The 50-nm-thick line patterns were well defined with low surface roughness. Under consistent patterning conditions, negligible variation in the thickness of the fabricated pattern was observed, regardless of differences in pattern width (fig. S13, A and B). The thickness of the PeNC patterns was able to be controlled in the range of 10 to 95 nm by adjusting the concentration of the PeNC dispersion (fig. S13, C to F).

In addition, we examined the surface morphology of the developed areas (not subjected to UV irradiation) using AFM and scanning electron microscopy (SEM) analyses to ensure the complete removal of any residue after the development process. The SEM image of patterned area obviously indicates the full retention of PeNCs (fig. S14A). In contrast, in the developed region, the majority of the PeNCs were removed during the development process, and only substrate surface was observed (fig. S14, B to D). These results conclusively demonstrate that our direct photocatalytic patterning method effectively removes any residual substances from the developed area. The CsPbBr₃ PeNC patterns fabricated through direct photocatalytic patterning exhibit high uniformity and fidelity in their structures, as evidenced by parameters such as line edge roughness (LER) and line width roughness (LWR) (fig. S15). The statistically calculated LWR and LER of the fabricated 10-μm-width pattern were measured to be 0.6 and 0.22 μm, respectively. The LWR and LER values were determined as three times the SD (3σ). The taper angle was observed to be 22.6°, which is consistent with the LER value (fig. S15B). Colors emitted by the patterns could be controlled by adjusting the chemical composition of the perovskites: CsPbI₃ emitted red (Fig. 4F), CsPbBr₃ emitted green (Fig. 4G), and CsPbBr₃ emitted blue (Fig. 4H) (all alkene-bearing ligand-capped).

To systematically determine the optimal UV exposure dose and minimal amount of PTMP additive necessary for successful patterning, we measured the film retention ratios under different patterning conditions (fig. S16). For determining the minimum UV exposure dose, we kept the amount of PTMP constant at 10 wt % and investigated the film retention ratio using 275 and 365 nm light sources (fig. S16A). Our investigation indicated that a minimum dose of 30 mJ cm⁻² (at 275 nm) was necessary to produce distinct patterns (i.e., film retention ratio > 70%). Subsequently, to ascertain the minimum quantity of PTMP, we measured the film retention ratio under fixed UV exposure doses (fig. S16B). Our results revealed that a PTMP concentration as low as ~1 wt % was sufficient for achieving high film retention ratios at a UV exposure dose of 600 mJ cm⁻² (at 275 nm). This observation suggests that a lesser amount of PTMP (1 to 2 wt %) can still generate high-resolution patterns under relatively high UV exposure doses. Furthermore, through FTIR analysis, we confirmed that the formation of distinct patterns correlates with a thiol conversion exceeding ~20%.

To demonstrate the potential of our method for EL applications, we fabricated perovskite-LEDs (PeLEDs) incorporating patterned CsPbBr₃ PeNC emission layers. The device structure was indium tin oxide (145 nm)/poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (40 nm)/poly(N,N′-bis-4-butylphenyl-N,N′-bisphenyl)benzidine (25 nm)/poly[9,9-bis(3′-(N,N-dimethyl)-N-ethylammoinium-propyl-2,7-fluorene)-alt-2,7-(9,9-dioctylfluorene)]dibromide (5 nm)/PeNC (30 nm)/2,2′,2″-(1,3,5-benzinetriyl)-tris(1-phenyl-1-H-benzimidazole) (50 nm)/LiF (1 nm)/Al (100 nm) (fig. S18A). A clear EL pattern of 3-μm-wide line was demonstrated from the patterned PeLED (fig. S18B). In addition, the device characteristics were not adversely affected by PTMP addition and the subsequent patterning process (fig. S18, C and D). We attribute this successful EL demonstration to the minimal concentration of PTMP used (2.5 wt %), which allows for the preservation of the intrinsic charge transporting properties of the pristine PeNC layers. This indicates that our direct photocatalytic patterning process does not detrimentally influence the optical and electrical properties of the patterned PeNC layers, thus ensuring their suitability for EL applications.

Universality and multicolor patterning

Our direct photocatalytic patterning method can be readily applicable to other classes of emissive materials as long as they have photocatalytic activity. To confirm the broad applicability, we also applied the same direct photocatalytic patterning strategy to other types of materials: CdSe-based core/shell QDs, InP-based core/shell QDs, and even light-emitting conjugated polymers (Fig. 5). With slight modifications of patterning conditions, we successfully fabricated a green-emitting “KAIST” pattern with CdSe-based QDs (Fig. 5A), high-resolution line patterns (feature size of 4 μm) using InP-based QDs (Fig. 5D), and a whale pattern with a yellow-emitting conjugated polymer (Fig. 5G), demonstrating the universality of our method. These results indicate that these materials have high photocatalytic activities and effectively generate thiyl radicals upon irradiation, since their valence band maxima are lower than the reduction potential of PTMP (45). In addition, the successful formation of patterns was attributed to the presence of alkene groups (C═C) in their molecular structure (for the conjugated polymer) or surface ligands (for QDs), allowing thiol-ene cross-linking.

Fig. 5. Universality of direct photocatalytic patterning.

(A and D) Fluorescent optical microscope (OM) images of (A) CdSe-based and (D) InP-based core/shell quantum dots (QDs). (B and E) Absorption (black) and photoluminescence (PL) (red) spectra of a QD solution (solid line) and a QD/pentaerythritol tetrakis(3-mercaptopropionate) (PTMP) mixture solution (dashed line): (B) CdSe-based and (E) InP-based core/shell QDs. (C and F) PL quantum yield (PLQY) of (C) CdSe-based and (F) InP-based QD-PTMP films at each patterning step (relative to the pristine film, which does not contain PTMP). (G) A whale pattern of conjugated polymer (super yellow). (H and I) Dual-color butterfly patterns composed of (H) blue and green perovskite nanocrystal (PeNC) regions and (I) green and red PeNC regions. The cyanish color of the green PeNC pattern in (H) is ascribed to the use of blue-cyan region bandpass filter.

The patterning process did not degrade the optical properties of QD films (Fig. 5, B, C, E, and F). There were no observable shifts in the PL peaks, and the PLQY was preserved or even improved, particularly for the InP-based core/shell QDs. This improvement is most likely due to the additional surface passivation provided by PTMP ligands, as we revealed for PeNCs.

To produce high-resolution displays with subpixels of different colors, the patterning process must be orthogonal; i.e., the predeposited emitter patterns should remain intact during subsequent deposition and patterning processes. We have demonstrated that our direct photocatalytic patterning method enables orthogonal processing by creating dual-color butterfly patterns through consecutive patterning processes. Specifically, we achieved a green-and-blue pattern by consecutively patterning green and blue PeNCs and a red-and-green pattern by patterning green and red InP QDs, without compromising the underlying layer (Fig. 5H and fig. S19A). Moreover, red-and-green and RGB multicolor pattern using PeNCs was demonstrated without any undesired color contamination with controlled patterning conditions (Fig. 5I and fig. S19B).

DISCUSSION

We have developed a simple and universal method to precisely pattern colloidal PeNCs and other emissive nanomaterials without damaging their optical and structural properties. This direct photocatalytic patterning method provided high-resolution patterns of RGB PeNCs with complete preservation of the PLQY at a very low light dose (down to ~30 mJ cm⁻²) with a 275 nm light source by exploiting two approaches: photocatalytic patterning strategy and a multifunctional click-chemistry reagent. The yield of the thiol-ene reaction was substantially increased by the photocatalytic activity of PeNCs; the increase enabled cross-linking between PeNCs and consequent pattern formation without the need for an additional photoinitiator. PTMP, which has four thiol end groups, acted as an efficient cross-linker for direct photocatalytic patterning and also as a thiolate ligand for the PeNCs, thereby enhancing the surface passivation and consequently increasing the PLQY and photostability of the PeNC patterns. As a result, high-resolution patterns of PeNCs (≥560 nm feature size and 2 μm pitch) with various shapes and colors were fabricated using a small proportion of PTMP. This corresponds to a resolution of >12,000 ppi for single-color pixel arrays. Furthermore, we found that the patterned sample stored under ambient conditions for over 150 days maintained a similar quality to that of the sample right after fabrication (fig. S20). The underlying patterning mechanism was identified by investigating the photochemical transformations of surface ligands and PTMP, particularly the formation of thiyl radicals, which are responsible for initiating the thiol-ene reaction.

Consecutive orthogonal patterning of emissive nanomaterials was also demonstrated with various multicolor patterns. This direct photocatalytic patterning provides enhanced PLQY and photostability of PeNCs after patterning, without requiring any posttreatment. Moreover, the method represents the most advanced achievement in terms of high-resolution, broad range of irradiation sources, and minimum light dose (table S2).

The general applicability of direct photocatalytic patterning makes it highly promising for use with other types of solution-processable semiconducting materials, such as other colloidal inorganic nanocrystals (e.g., II-VI and III-V), metal oxide nanoparticles, carbon dots, graphene QDs, conducting polymers, and even organic small molecules, as long as they can function as photocatalysts with photogenerated excitons and have alkene groups (C═C) in their structures or surface ligands. We have demonstrated that various patterns of CdSe- and InP-based core/shell QDs and a conjugated polymer can be fabricated using direct photocatalytic patterning while retaining their original photophysical properties. The universality of this patterning method shows the potential to enable high-resolution patterning of a wide range of materials, electronic devices, and their integrated circuits in a nondestructive manner.

To expedite the practical industrial application of direct photocatalytic patterning, the following future research directions should be considered. First, it is necessary to identify and evaluate more photocatalytic click-chemistry reactions that can produce patterns at low doses while maintaining or enhancing their optical properties. Next, the chemical structure of click-chemistry reagents should be optimized to maximize the reaction yield without compromising the electronic properties of the resulting patterns. Moreover, it is crucial to comprehend the precise patterning mechanisms of various target materials with different chemical or electronic structures and photocatalytic activities. We envision that this direct photocatalytic patterning method will be able to have widespread applications in next-generation optoelectronic devices such as augmented reality headsets, stretchable displays, hyperspectral image sensors, and smart healthcare devices.

MATERIALS AND METHODS

Synthesis of CsPbBr₃ PeNCs

CsPbBr₃ PeNCs were synthesized by following the article with some modifications (46). In a 100-ml three-necked round-bottom flask, 700 mg of PbBr₂, 1400 mg of ZnBr₂, 7 ml of OA, 7 ml of OLA, and 40 ml of 1-octadecene (ODE)were mixed. The lead bromide precursor solution was degassed at 120°C for 1 hour. Then, the lead bromide precursor solution was kept in a nitrogen atmosphere and increased the temperature to 180°C. When the reaction temperature (for green CsPbBr₃, 180°C) was reached, swiftly injected 3 ml of the Cs-oleate solution. After 30 s of reaction time, the CsPbBr₃ PeNC dispersion was cooled down through an ice bath. Thirty milliliters of methyl acetate was injected into the CsPbBr₃ for purification. The CsPbBr₃ PeNC dispersion was centrifuged at 7800 rpm for 5 min to get monodisperse CsPbBr₃ PeNCs. After the first purification, only the precipitate was collected and dispersed in 6 ml of toluene. To achieve additional purification, 18 ml of methyl acetate [toluene:methyl acetate = 1:3 (v/v)] is added to the PeNC dispersion and centrifuged at 7800 rpm for 5 min. The precipitate is collected, and residual materials such as ligands and unreacted materials are removed. This purification step must be carried out more than twice. After purification, PeNCs dispersed in toluene. After a day of benchtop, the transparent and bright CsPbBr₃ PeNC dispersion was collected. OA-OLA CsPbBr₃ PeNC was synthesized with the same procedure only without ZnBr₂. Synthesis of red-emitting PeNCs (47), blue-emitting PeNCs (48), PeNCs with alkene-free ligands (49, 50), and QDs were described in the Supplementary Materials.

Patterning process

PTMP solution was prepared by mixing tetra-thiol in toluene (concentration = 0.025 M). Before formulating patternable ink, we washed the CsPbBr₃ PeNC dispersion with methyl acetate at least once to remove residual materials. We mixed the CsPbBr₃ PeNC dispersion (~20 mg/ml) and 0.025 M tetra-thiol/toluene solution (10 wt % of PeNC). To remove some aggregated particles, we filtered the mixed ink once with polyvinylidene difluoride 0.45 μm filter to make a clear patternable ink. Fifty to 100 μl (depending on the size of the substrate) of the patternable ink was spin-coated (2000 rpm, 30 s) on a glass substrate or a Si substrate. The fabricated PeNC-PTMP film was aligned with a premade photomasks. Then, UV light was irradiated to the PeNC-PTMP film through the manufactured optical device with 40 mW cm⁻² of power density. After UV exposure, we developed the unexposed PeNCs through toluene to produce a pattern. Other emissive nanomaterials (CdSe-based QD, InP-based QD, and conjugated organic emitter) can also be patterned using the same method.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S20

Tables S1 and S2