Quantum dots (QDs) are regarded as a magic material in many optoelectronic applications, including light-emitting diodes (LEDs), (1,2) solar cells, and photodetectors, (3−7) primarily due to their tunable bandgap, which can be adjusted by changing their size, as well as their solution processability, which enables low-cost fabrication. Specifically, lead sulfide (PbS) QDs are particularly promising in short-wave infrared (SWIR) photodetectors due to the naturally narrow bandgap and suitable charge carrier diffusion length (<300 nm) of the solid films. (8,9) The limited charge carrier diffusion length of the QD solid films enables high-resolution QD imaging through monolithic QD deposition on CMOS chips, offering a significantly lower cost compared to the traditional InGaAs SWIR technique, which relies on a “flip-chip” process. (10−12)

Presently, to fabricate highly efficient photodiode devices, such as solar cells and SWIR photodetectors, thiol-treated QD layers are commonly used as hole transport layers (HTLs) due to their favorable band alignment and the formation of a “homojunction” with a lower energy barrier for charge carrier transfer between the QD active layer and the QD HTL. (13,14) The fabrication process of the QD HTL is known as a solid-state ligand exchange (SSLE) as described below. (15,16) Organic-capped QDs, such as those with long-chain ligands like oleic acid (OA), tend to form a superlattice due to their inherent propensity for long-range ordered stacking through self-assembly. A short-chain thiol precursor is consequently applied to the pristine QD film to initiate the SSLE, during which all long-chain ligands of the QDs are simultaneously substituted by thiol ligands. A 2- or 3-step rinsing process is then applied to remove any residual long-chain ligands, though this inevitably leads to crack formation due to the volume contraction of the QD ensemble. (17) The deposition process is repeated two or three times to increase the layer thickness and partially refill the crack space. (15) After several hours of storage in dry air, the QD solid film demonstrates p-type semiconductor behavior and can be used as the HTL in QD diode devices. (18,19) Many factors, including the light conditions, ambient humidity, etc. during oxidation doping, can significantly affect QD-HTL performance and device efficiency. (14,18) Therefore, refining the QD HTL treatment in the oxygen doping process is crucial for achieving high-performance QD devices, in addition to ensuring the quality of other functional layers and interfaces. Another challenge arises from structural modulation, as the final QD HTL often develops micrometer-scale cracks if not properly refilled during the SSLE process. Particularly monosize QDs exhibit a more pronounced crack formation, as indicated in Scheme 1a. These HTL cracks can further lead to unexpected contacts between the electrodes and the QD active layer, thus increasing the dark current density by introducing ohmic contacts. Many efforts in the literature have been implemented to reduce the crack formation of the QD HTLs. For instance, Wang et al. developed a QD ink technique via a solution-processed ligand exchange process for the HTL fabrication and deposition, effectively decreasing the volume contraction and thus reducing the crack generation. (20) Sharma et al. used a pretreatment of the OA-PbS QDs reducing the cracks of the final HTL by changing the surface condition of the QDs. (21) Based on a similar principle, Huang et al. used a hybrid ligand exchange method by using MPA and EDT synergic ligands in the SSLE process and successfully reduced the cracks from the volume contraction. (22) These efforts highlight two key factors in the formation of crack-free QD HTLs. First, achieving a crack-free QD HTL can improve device performance by reducing leakage current. Second, modulating the stacking dynamics of QDs during the “liquid-to-solid” transition plays a crucial role in crack-free QD film formation. (23−26) Specifically, establishing a long-range disordered film configuration, such as by reducing uniform surface conditions or altering the nanoparticle morphology (e.g., by partially fusing the QDs), is considered an effective method for achieving optimized QD HTLs with crack-free morphologies. (27) The fundamental mechanism behind the crack-free optimization of SSLE-based QD solids is hypothesized to involve short-range disorder inducing long-range disorder. This process can lead to self-compensation for volume contraction by alleviating or releasing strain forces generated during the ligand exchange-induced structural phase transition. Furthermore, short-range disorder in the QDs can be established through self-assembly of multimodal distribution QDs, (28,29) with the ligand exchange treatment causing an isotropic collapse of the QD arrangement, avoiding crack formation.

To prepare H-QDs, two different sizes of QDs are selected from their absorption features. OA-capped QDs with the first exciton peak at 850 nm are denoted as small-size QDs (S-QDs). OA-capped QDs with the first exciton peak at 1350 nm are denoted as large-size QDs (L-QDs). The fabrication process of the QDs is seen in the experimental section. The morphologies and the absorption spectra of S-QDs and L-QDs are given in Figures S1 and S2.

After deposition and the SSLE process, the surface morphologies of the QD HTLs made of different-size QDs are illustrated in Figures 1(a–c) and S3. The SEM images suggest the apparent crack generation of the QD HTL made of monosize QDs (S-QDs and L-QDs), while the QD solids made of H-QDs are significantly improved and appear “crack-free” at the observation scale. Monosize QDs undergo substantial superlattice transformation due to the short-range order configuration, (23) particularly in the ligand exchange process, (26) where anisotropic volume contraction induces strong strain force and leads to crack formation. As mentioned in the introduction, the H-QDs can easily release the strain force due to the short-range disorder configuration. The stacking configuration of the QDs in the solids made of monosize QDs and H-QDs are respectively studied by GISAXS and the 2D GISAXS patterns are shown in Figure 1(d–f). In the cases of monosize QD solids, prominent Bragg scattering features can be observed in the out-of-the-plane direction of the 2D GISAXS data in Figure 1(d,e), which suggests a uniformed interdot arrangement. No such structure configuration observed in H-QD solid from the GISAXS pattern in Figure 1(f). The corresponding azimuthal integrations to the GISAXS patterns of all QD solids are conducted according to previous literature, (30,31) and the results are shown in Figure 1(g–i). In reciprocal space, small QDs exhibit a larger q Bragg peak at q = 2.1 nm^–1 while the large QDs show the Bragg peak at a lower q position, q = 1.3 nm^–1 as seen in Figure 1(g,h). Interestingly, the scattering peak is smeared in the integration curve of the H-QD solid in Figure 1(i), indicating a successful short-range disorder configuration due to the homogeneous distribution of S-QDs and L-QDs, without the accumulation of monosize QDs. Notably, the H-QD solids exhibit similar GISAXS feature to QD solids made of “direct ink” with a broad QD size distribution. (32,33) The scattering similarity confirms our previous assumption about the peak smearing of interdot distance scattering of the direct ink-based QD solid. (34) A broadened size distribution (large particle radius standard deviation) is helpful to increase the packing density of the hard spheres model, (35,36) which can be used to explain the packing dynamics of the direct ink and the H-QDs. Notably, using H-QDs brings a very simplified method to modulate the packing dynamics of the QD HTL toward releasing the strains from monosized QDs and reducing film cracks. More importantly, previous efforts, including hybrid ligands, pretreatment, etc. could also be concluded in such principle, by partially introducing short-range disorder, such as different sizes, or different shapes, during the film fabrication process. (27,37) The as-formed long-range disorder can help boost the effective homogeneity and overall density of the QD HTLs.

Besides the structure study, we use photophysics methods to evaluate how the H-QDs influence the energy state distribution and the charge carrier dynamics of the QD HTLs, and the results are shown in Figure 2. The PL spectra of all compared QD HTLs are seen in Figure 2(a). Monosize QD solids exhibit featured PL peak positions from their band-edge radiative recombination at around 1150 nm for S-QDs and 1310 nm for L-QDs. Interestingly, H-QD solids exhibit enhanced PL intensities in different ratios in weight of S-QDs to L-QDs (S/L), in which the QD HTL with a ratio of S/L = 2:1 demonstrates the most prominent PL intensity as indicated by the pink curve. Two main factors contributing to the PL enhancement of the H-QD HTL are discussed below. The reduction of cracks in the QD HTLs by mixing QDs decreases the overall interface area and thus reduces the surface trap states, including the deep states for radiative recombination. Moreover, L-QDs surrounded by S-QDs become recombination centers for all nearby charge carriers, rather than facilitating charge carrier diffusion.

Figure 2(b) shows the TRPL data of all compared QD HTLs in a nonquench condition (on glass substrates). (38) We use a biexponential decay function to fit the data with parameters are seen in Table S1, in which τ₂ is assigned to be the factor relating to the surface trap states. We find that the QD HTL with S/L = 2:1 demonstrates the largest τ₂ value and the longest charge carrier lifetime, indicating the fewest trap density. Figure 2(c) shows the absorption spectra of the HTLs, where the first exciton absorption peak is clearly distinguishable in both monosize QD HTLs and H-QD HTLs. This suggests that the QDs maintain their absorption independence in mixed and strongly coupled QD stacking configurations.

To evaluate charge carrier transport in practical devices, we also use photophysics methods and investigate the charge carrier dynamics of QD HTLs on QD active layers, in which QD HTLs are respectively made of S-QDs, L-QDs, and H-QDs. The QD active layer is fabricated using the QD ink method, as described in the experimental section. The PL spectra of the compared active layers with different-size QD HTLs are shown in Figure 2(d). We find that charge carrier transfer from the S-QD HTL to the active layer is less efficient than that from the H-QD HTL, as indicated by the absence of the PL feature from the S-QDs. Nonetheless, the charge carrier transfer from S-QDs to the QD active layer is mostly prohibited from the apparent double PL peak in longer wavelengths in the case of H-QDs on the active layer. This means that the H-QD HTL improves hole transfer from the active layer to HTL, particularly to large QDs, which is confirmed by the TRPL data by tracking the PL lifetime of the active layer at around 1520 nm in Figure 2(e) and Table S2. The red shift of PL peak of QD active layer compared to the QDs in solvent is due to the partially aggregation. (39,40) Thus, all active layers were made in the same condition to ensure a uniformed PL position to reduce the influence from the intrinsic active layers. The H-QD configuration exhibits a shorter average charge carrier lifetime than the S-QD configuration. Consequently, we briefly describe the improvement schematic of the HTL made of the H-QDs, as shown in Figure 2(f), in which the L-QDs in H-QDs play an important role in efficiently extracting holes from the active layer.

Single charge carrier devices based on monosize QD and H-QDs are fabricated to evaluate the trap density of the QD HTLs. The device architecture is shown in Figure 3(a) and the derived J–V characters are shown in Figure 3(b). The trap-filled limit voltages (V_TFL) are further fitted from the J–V curves and the trap densities (n_t) are estimated according to eq 1 in which the ε and ε₀ are relative primitivity and vacuum primitivity respectively. e is the elementary charge, and the d is the distance of the electrode. The H-QD HTL exhibits the smallest V_TFL (0.54 V) indicating the lowest trap state density, in good agreement with the TRPL data of the QD HTLs in the nonquench condition, in Figure 2(b). The trap states are assumed distributed on the solid surfaces as illustrated in Figure 3(c), in which the H-QDs exhibit the smallest effective surface under noncrack conditions. Moreover, we assume the H-QDs are better passivated due to their denser packing and more homogeneous distribution, which reduces inner cracks compared to monosize QDs.

Further, we use the H-QDs to make HTLs in photodetector device fabrication to verify the benefits of H-QDs with different sizes. The photodetector device architecture is shown in Figure 4(a). It consists of a transport conductive oxide (TCO) layer as the bottom electrode, a ZnO electron transport layer (ETL), a QD solid active layer fabricated by solution-phase ligand exchange, and H-QDs as the HTL. A MoO_x buffer layer is added as an additional HTL, with a silver top electrode. The energy band landscapes of the function layers of the device with different HTLs are shown in Figure 4(b), based on ultraviolet photoelectron spectroscopy (UPS) analysis (Figure S4). The H-QD HTL exhibits an intermediate energy band structure compared to S-QDs and L-QDs. In working devices, the S-QD HTL reduces charge carrier extraction by forming an energy barrier due to Fermi level alignment, while the L-QD HTL exhibits weak electron blocking. Interestingly, the H-QD HTL can reduces the barrier while still blocking electron transport. The J–V characters of the devices from the same batch device fabrication in dark conditions and light conditions are shown in Figure 4(c), we find that the H-QD HTL device exhibits a better reverse saturation current characteristic with a reduced dark current density, compared to monosize QD HTL. This finding is mainly attributed to the synergic improvements from the energy band landscape and crack-free film morphology of the HTL. H-QD HTL thus reduces the leakage current from ohmic contact, which is studied by eq 2 Here J_D represents the dark current density of the device, V is the applied bias, and J₀ for the reverse saturation current density of the diode. R_S and R_SH denote the series resistance and shunt resistance of the device, respectively. n is the diode ideality factor, k is the defect level, and m represents the defect concentration. (3,14,41) Notably, the ideality factor n = 1.4 and power factor m = 2.4 are determined according to literature considering that we focused on separating the leaking current under reverse bias into three different components including (i) diode current, (ii) ohmic current, and (iii) nonohmic current contributions. (42,43) The J–V fitting results of H-QD device are shown in Figure 4(d) and the rest results are shown in Figure S6. The fitting parameters and results are provided in Tables S3 and S4. The device based on H-QD HTL with S/L = 2:1 of small QDs to large QDs exhibits the lowest dark current density of 392 nA/cm² in which the contribution from the ohmic contact (J_Ohm) is significantly depressed to 285 nA/cm². Notably, when the hybrid ratio of QDs is S/L = 1:1, the unexpected current contribution from the trap states is almost prohibited. Thus, the mixing QDs configuration effectively reduces the trap assistant tunneling from the J_non-Ohm comparison, (44,45) which agrees with the TRPL data and single-charge carrier device analysis. The results of frequency dependent current test are shown in Figure 4(e). The compared devices exhibit similar currents at f = 1 Hz and bias = 0 V.

The response speed test results are seen in Figure 4(f) and the analysis are shown in Figure S8 and Table S6. H-QD HTL boosts the photodetector’s response speed, which is attributed to a faster charge carrier mobility of the H-QD HTL than the monosize QD HTL. Specifically, the short-range disorder induced by H-QDs can enhance charge carrier mobility, consistent with previous studies, (46,47) which attribute this effect primarily to the closer packing density observed through GISAXS. However, according to the Bässler equation, (48,49) charge carrier mobility is expected to decrease in H-QDs due to energy state disorder, as charge carriers tend to hop to sites with lower energy states and reduced interdot distance. Nonetheless, the increased packing density and reduced trap states in H-QDs can also enhance overall charge carrier mobility. This synergistic effect not only increases charge carrier diffusivity but may also create new transport pathways within the QD solid. (50−52) Briefly, in our case, the H-QDs-based structure disorder overwhelms the energetic disorder and brings a positive influence to the charge carrier mobility.

The spectra for responsivity (R), detectivity (D*) and EQE are shown in Figure 4(g–i). The responsivity of the device is calculated based on eq 3 in which J_ph and L_light respectively represent the light-generated light current density and the incident light power density. I_light and I_dark are the corresponding currents of the device in different test conditions, in specific light illumination, and under dark. P_hν is the light power of the incidence light with a frequency of ν on the work area of the device. We confirm that H-QD HTL brings the device a very high responsivity of R = 0.65 A/W at 1240 nm, which is much higher than that of devices with monosize QD HTLs, specifically, R = 0.46 A/W for the S-QDs and R = 0.35 A/W for the L-QDs, at 1200 nm. Besides, the detectivity (D*) of the device is calculated based on eq 4 in which the A stands for the working area. f is the test current frequency of 1 Hz. R_λ is the responsivity at specific wavelength λ calculated from eq 2, and i_n is the noise current obtained from the noise spectral current in Figure 4(e). The device based on H-QD HTL exhibits an improved D* = 2.4 × 10¹² jones (at 1240 nm, the peak position) compared to devices based on monosize QD HTL with the highest D* = 5.0 × 10¹¹ jones at 1200 nm.

The external quantum efficiency (EQE) of the devices is calculated based on eq 5 in which h stands for the plank constant, c for the light speed, and λ for the wavelength respectively. The H-QD HTL SWIR photodetector exhibits over 65% EQE at 1240 nm, which is much higher than devices based on monosize QDs (48% for S-QD HTL device, 37% for L-QD HTL device at 1200 nm). In addition, a device comparison table is prepared as seen in Table S9 which indicates our champion device based on H-QD HTL outperforms most recent devices with state-of-the-art in literature.

Figure 5(a–c) show the J–V characters of 30 compared devices with optimized ZnO ETL layers in a dark condition. Figure 5(d) illustrates the statistics of the dark current density of the devices at bias = −0.5 V. The QD-PD devices with H-QD HTLs exhibit apparently depressed dark current density (minimum J_D = 388 nA/cm²) compared to the devices with monosize QD HTLs, minimum J_D = 620 nA/cm² for S-QD devices and minimum J_D = 2175 nA/cm² for L-QD devices. Besides, the statistics of the EQE of the compared devices as shown in Figure 5(e) indicate the priority of the devices with H-QD HTLs. Specifically, the average EQE = 54% for the H-QD devices, 41% for the S-QD devices and 33% for the L-QD devices. Besides, the linear dynamic ranges (LDRs) of the champion devices for comparison are shown in Figure 5(f). The device with the H-QD HTL demonstrates a broader LDR than the rest champion devices with monosize QD HTLs.