Tunable bandgap in hybrid perovskite CH₃NH₃Pb(Br_3−yX_y) single crystals and photodetector applications

The introduction of organometallic halide perovskites (especially CH₃NH₃PbX₃, MAPbX₃, X =Cl, Br or I) is triggering a quasi-revolution in the field of solar cells. Due to their high absorption, long balanced carrier diffusion length, tunable energy gap, and low cost, the organometallic halide perovskites have attracted many researchers’ attention.^1–4 Over the last decade, the organometallic halide perovskites based semiconductor devices, especially the perovskite solar cells, have developed dramatically. By introducing solid state hole transporting materials (HTM) and ions-doped HTM,^5,6 improving the morphology,^7–11 and optimizing energy level matching between different layers,^6,12 the power conversion efficiency (PCE) of perovskite solar cells keeps rising. The PCE of CH₃NH₃PbX₃ (MAPbX₃) perovskite solar cells has increased to over 20% only in 5 years.^{1,3,5,13–15} Besides, the combination of perovskites and mature silicon photovoltaic technology has also been researched extensively to fabricate solar cells with higher PCE.^16,17 Additionally, the organometallic halide perovskites also have been widely studied in other fields, such as light emitting diodes (LED),^18,19 laser²⁰ and photodetectors.^21,22 And these studies confirm the promising application of organometallic halide perovskites for semiconductor devices.

However, most of these high-performance devices were fabricated with perovskite polycrystalline thin films. Compared with polycrystalline films, single crystal counterparts have no grain boundaries and low defect density. Predictably, materials with highly ordered crystal lattice structure always behave good performances, such as low intrinsic carrier concentration, high carrier mobility, long carrier diffusion length and high quantum efficiency. As reported previously, the diffusion length of perovskites single crystals is around 175 ± 25 μm and hole mobility is derived to be 164 ± 25 cm²V⁻¹s⁻¹.²³ By contrast, the hole mobility is only 66 cm²V⁻¹s⁻¹ in annealed MAPbI₃ polycrystalline films.²⁴ Besides, M. Bakr and his co-workers demonstrated that the carrier diffusion lengths could exceed 10 micrometres (about 17 μm) in CH₃NH₃PbBr₃ single crystals.²⁵ Huang and colleagues also demonstrated a single crystal perovskite devices with almost 100% internal quantum efficiency.²³ These remarkable results demonstrate that the application of perovskite single crystals would dramatically improve the performances of semiconductor devices.

Cooling down high concentration precursor solution, evaporating precursor solution and diffusing the antisolvent into precursor solution are ordinary procedures in preparing perovskite single crystals. By cooling down the mixture aqueous HI-H₃PO₂ precursor (PbI₂/MAI was 1:1 by molar) solution, Kanatzidis and co-workers have prepared rhombic dodecahedral MAPbI₃ crystals.²⁴ And MAPbI₃ crystals with a wide absorption and a good thermal stability were also created with the method of cooling down high concentration hydroiodic acid precursor solution (MAI and Pb(CH₃COOH)₂ ⋅ 3H₂O, the molar ratio was 1:1).²⁶ Furthermore, Shi reported that diffusing the antisolvent vapour slowly into the precursors solution (PbX₂/MAX was 1:n by molar) is another way to grow MAPbX₃ crystals.²⁵ With this approach, they created large MAPbX₃ single crystals with low trap-state densities and carrier diffusion lengths of >10 μm. Equally important, Zhang kept MAPb(Br_3−xCl_x) DMF precursor solution at 50^∘C in a supersaturated state, and prepared mixed halide single crystals.²⁷ And further studies proved MAPbI₃ and MAPbBr₃ exhibit inverse solubility in γ-butyrolactone and DMF respectively, and by which they have also demonstrated a successful synthesis of perovskite single crystal.^28,29

In this work, we optimized the experimental procedures to heat the sealed precursor solution and increase the temperature stepwise. Via increasing temperature stepwise, we tried to maintain the precursor solution in a metastable supersaturation state to achieve a relative rapid crystal growth rate and avoid a synthesis of a mass of tiny crystals. And further experiments are conducted to investigate the influence of halogen content and preparation temperature on structural and optical properties of mixed-halide perovskite single crystals. Based on as-synthesized perovskite single crystals, we fabricated metal-semiconductor-metal (MSM) photodetectors, which have shown a high ON/OFF ratio and a high switch speed.

The mixed halide MAPb(Br_3−yX_y) single crystals were synthesized from lead halide (PbX₂) and methylammonium halide (MAX) DMF solution. The precursor solutions were prepared by mixing PbX₂ and MAX with different ratios of Br/Cl or Br/I, and dissolved in a suitable amount of DMF. The molar ratio of PbX₂/MAX was 1:1 constantly. Then the precursor solutions were filtered with PTFE syringe filters (pore size 0.22 μm) after dissolving adequately. And 2 milliliters of filtered precursor solution with different halogen ratio were sealed in vials respectively. Using oil bath, these vials were heated up with one appropriate heating rate and a holding time. As temperature rose from room temperature to 95^∘C, single crystals formed successively under proper temperature. With this method, we prepared the MAPb(Br_3−yCly) and MAPb(Br_3−yI_y) single crystals as expected. The composition of precursor solutions and experiment results were showed as Tables I and Table II. We fabricated MSM photodetectors by coating patterned electrodes on as-prepared perovskite single crystal surface with an annealing of 100^∘C/30 mins afterwards. The electrodes of 2290nm (70nm Cr/1700nm Al/20nm Ti/500nm Au) were prepared by electron beam evaporation with a shadow mask.

In the processes of crystal growth, an appropriate heating rate results in bulk perovskite crystals. By contrast, over-rapid heating makes the precursor solution to be in an unstable supersaturation state which results in a plenty of small crystals in a short time (Fig. 1(a)). Fortunately, this misoperation could be redressed by dissolving at a lower temperature. Thus, a relatively low heating rate is necessary to prepare large crystals in a rapid growth rate.

Table I shows the experimental details and optical images of perovskite single crystals with different Br/Cl ratios. The sizes of these crystals were about 7mm×7mm×2mm. Obviously, the color of MAPb(Br_3−yCl_y) single crystals vary from orange-red to yellow with increasing chloride content. When the MABr: PbBr₂ molar ratio is 1:1 without chlorine-mixing, and the concentration of MAPb(Br_3−yCl_y) DMF solution is 1.1mol/L, the crystals nucleate at around 50^∘C. While the PbBr₂: MABr: MACl ratios are 10:7:3 (the Br/Cl ratio is 9:1) and 10:2.5:7.5 (the Br/Cl ratio is 3:1), the crystals nucleate at about 46^∘C and 39^∘C, respectively. As the chlorine content increases, the crystal growth temperature decreases gradually as following.

Table II shows the experimentals and optical images of perovskite single crystals with different Br/I ratios. The sizes of these single crystals are about 4mm×4mm×1.3mm. The color of MAPb(Br_3−yI_y) single crystals change from orange-red to dark red with increasing iodide content. In contrast to chlorine-mixed solution, the iodine introduction increases the temperature of nucleation. When the MABr: PbBr₂: PbI₂ ratio is 10:8:2 (the Br: I ratio is 13:2) and the concentration of MAPb(Br_3−yI_y) DMF solution is 1.1mol/L, the single crystals would nucleate at about 65^∘C. As the I: Br ratio increases to 1:2, the nucleation temperature increases to 79^∘C approximately under the same concentration. As expected, the higher concentration makes lower crystallization temperature with the same Br: I ratio. With the Br: I ratio of 13:2, the crystals form at 69^∘C and 63^∘C in 1.1mol/L and 1.3mol/L precursor solution respectively. Although iodine-mixing makes it easier to prepare bulk crystals, the growth rate of crystals is much lower comparing with chlorine-mixed crystals.

According to the expression $ΔG=RT⋅ln C 0 C$ and $Δ G * = 4 3 π σ 3 /Δ g v 2$⁠, we deduce that there are two main factors for that different halogen compositions lead to varied phenomena. Firstly, chlorine can lower the critical nucleation energy in MAPb(Br_3−yCl_y) precursor solution, and iodine increases the critical nucleation energy in MAPb(Br_3−yI_y) precursor solution. Secondly, halogen have different solubility and affinity in precursor solution. The priorities of halogen solubility in precursor solution are shown below: I > Br > Cl. For these reasons, the replacement of bromine by chlorine can maintain the driving force in nucleation and growth processes under lower supersaturation. This would help explain why crystallization occurs under lower temperature with chlorine mixing.

This is in accordance with O. M. Bakr’s work, in which they showed that the solubility of MAPbBr₃ and MAPbI₃ in certain solvents dropped markedly with temperature increasing.²⁹ The perovskite solubility curve is shown schematically in Fig. 1(b). The solubility curve can be separated in three regions: the unsaturated region, the metastable supersaturated region and the unstable supersaturated region.³⁰ According to the solubility curve, heating rate is one of the most critical factors in crystallization. The smaller metastable supersaturation region leads to lower heating rate to avoid masses of nucleations. The experiments demonstrate that chlorine content shrinks the area of metastable supersaturation region, while iodine content expands the area of metastable supersaturation region. To obtain bulk single crystals, the heating rate should be precisely controlled to make sure that the precursor solution is metastable supersaturated.

Fig. 2 shows the surface and cross-section SEM images of MAPb(Br_3−yCl_y) crystals. The MAPbBr₃ crystals with regular form (Fig. 2(a)) and MAPb(Br_3−yCl_y) crystals with flat cleavage plane (in Fig. 2(b), y=0.2 and in Fig. 2(c), y=0.5) indicate a good crystallization which is consistent with the XRD results below. However, the nanosized hole in a MAPb(Br_2.5Cl_0.5) crystal (Fig. 2(c)) suggests that an over-rapid crystal growth results in some defects. Moreover, the crystals growth is a dynamic process, in which redissolution and crystallization occur simultaneously. When the crystallization prevails over redissolution, the grains grow up. In our experiments, we kept the precursor solution at about 60^∘C (a higher temperature than before) for more than 2 hours. In the early stage, the crystallization overcame redissolution, and the crystals grew up with regular parallelepiped shape. However, in the later stage, with decreased supersaturation, the redissolution balanced crystallization, some of the crystals turned to irregular margins and surfaces due to the dynamically and selectively recrystallizing at the defect locations.

The surface morphologies of the perovskite crystals prepared under different heating rates are shown in Fig. 3. A low heating rate, in theory, leads to low supersaturation which result in little nucleation driving force, and vice versa. In Fig. 3(a), the stepped surfaces and stepped edges reveal the Frank-van der Merwe (FM: layer-by-layer) growth mode under low supersaturation condition. While in Fig. 3(b), the stepped surfaces, island structures and inverse pyramid structures show the Stranski-Krastanov (SK: layer-plus-island) growth mode under high supersaturation condition. The stepped surfaces resulted from the growth in a layer-by-layer fashion on the perovskite crystals surfaces. As a result of the three-dimensional nucleations and coalescence with the stepped surfaces, the island structures formed under high supersaturation condition. As the islands grew up and coalesced together, some spaces between these islands developed into inverse pyramid structures.

Fig. 4(a) shows the DC XRD results of the MAPb(Br_3−yCl_y) (y=0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.75) crystals. The sharp peaks of (100), (200) and (300) in each curve confirm the pure phase of perovskite crystals with different compositions. And no peaks from impurity phases are observed. The XRD peaks of (100) crystal planes of the MAPb(Br_3−yCl_y) (y=0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.75) crystals are observed at 14.98°, 15.00°, 15.05°, 15.06°, 15.08°, 15.11° and 15.24°, respectively. It is obvious that the characteristic peaks of MAPb(Br_3−yCl_y) perovskite shift to larger diffraction angles with the increasing of chlorine content. Calculated by the equation d=λ/(2sinθ), lattice constants almost linearly reduce from 5.9082Å to 5.8095 Å (Fig. 4(c)). And the trend is consistent with Vegard’s law (Relationship to lattice parameter shown as below) with certain margin of errors. The lattice constant tends to decrease due to the mixing of chlorine with smaller atomic radius.

As shown in Fig. 4(b), the photoluminescence (PL) peaks are at 544.3nm, 535.0nm, 526.5nm, 520.5nm, 519.2nm, 514.0nm and 489.0nm for MAPb(Br_3−yCl_y) (y=0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.75) perovskite crystals respectively. The PL peaks shift to shorter wavelengths with the increasing of chloride content. The series PL spectra of MAPb(Br_3−yCl_y) crystals demonstrate a continuously tunable band gap. By calculating, the bandgap is increased gradually from 2.28ev to 2.53ev with increasing chlorine content, which conforms to a linear relationship (Fig. 4(c)). This result is consistent with Fang’s work.³¹ The shoulder peaks at long wavelengths which broaden the PL peaks probably result from crystal defects and surface states.

The DC XRD results of the MAPb(Br_3−yI_y) (y=0, 0.2, 0.4, 0.6, 0.8) crystals are shown in Fig. 5(a). The XRD peaks shift to smaller diffraction angles with the increasing of iodine content. The lattice constants of MAPb(Br_3−yI_y) perovskite crystals increase from 5.9082 Å for the MAPbBr₃ crystals, to 5.9360 Å and 5.9786 Å for the MAPbBr_2.6I_0.4 and MAPbBr_2.2I_0.8 crystals respectively. Obviously, the increase of iodine content results in a little bigger crystal lattices. However, the perovskite crystals decomposed as the laser shines on it duo to its high energy density. The bandgaps were calculated based on the absorption spectrum (Fig. 5(b)) and Vegard’s law (Relationship to band gaps in semiconductors are shown below). The bandgap decreased from 2.28ev to about 1.92ev dramatically.

To investigate the photoelectric properties of these perovskite single crystals, we fabricated MSM photodetectors. The device structure is shown in Fig. 6(a). The temporal photo-response (Fig. 6(b)) of the MAPbBr₃ single-crystal photodetectors suggests that the current ON-OFF ratio is about 100 with stable dark-current and photocurrent output and a high response speed. The I–V curve of the same photodetectors is illustrated in Fig. 6(c), which demonstrates linear I–V behavior at low voltage. The dark current is as low as 2.58×10⁻⁸A at 1V, while the photocurrent is 6.65×10⁻⁷A. The linear fitting equation is I=9.57×10⁻⁵+5.83×10⁻⁴ V. This result is consistent with the linear portion of the I-V curve at low voltages in Fang’s work,³¹ There remains works to be done to improve the device performance by optimizing device structure and fabricating processes. And further studies will be published elsewhere.

In conclusion, we obtained MAPb(Br_3−yX_y) perovskite single crystals with different halogen content by precisely raising the temperature of precursor solution stepwise. The crystallization was attribute to the negative temperature coefficient of MAPbX₃ perovskite solubility in DMF which was reported before. Further studies demonstrated the growth mechanisms of perovskite crystals. The MAPb(Br_3−yX_y) crystallization shows Frank-van der Merwe mode under low supersaturation condition and Stranski-Krastanov mode under high supersaturation condition. It is also observed that MAPb(Br_3−yCl_y) crystals nucleate at lower supersaturation condition compared to MAPb(Br_3−yI_y) crystals, which is attributed to the low critical nucleation energy and the low solubility of chlorine. And the bulk MAPb(Br_3−yCl_y) crystals should be prepared with relative low heating rate, while the iodine-mixing counterpart could be prepared with a high heating rate. Based on these bandgap tunable perovskite single crystals, we fabricated photodetectors with stable and high performance. These results demonstrate promising development of perovskite single crystals for semiconductor device application.

This work was supported by National Natural Science Foundation of China (NOs.51472229, 61422405, 61474109, 61274008), “100 talent program” of Chinese Academy of Sciences, and State Key Lab of Silicon Materials Opening funding Nr. SKL2014–4.