Crystal phase-directed growth of rutile/anatase TiO₂ heterojunctions via in situ stepwise chemical bath deposition below 80 °C

Abstract

The fabrication of hierarchical TiO₂ phase junctions, such as rutile/anatase heterostructured films, via a low-temperature crystallization, has long been a challenge due to the lack of effective reaction strategies. In this study, we developed a temperature-gradient double-stage chemical bath deposition (CBD) method to produce rutile/anatase TiO₂ thin films.

---

The construction of titanium dioxide (TiO₂) heterojunctions with different crystal phases has been extensively studied, particularly for solar-driven energy conversion. Key targets include efficient charge separation in solar water-splitting and designs for improved electron transport in high-performance photovoltaic devices.^1,2

When depositing metal oxide thin films on conductive substrates, a format essential for many device architectures, CBD provides a distinct advantage over conventional spin- and spray-coating techniques. CBD enables a highly conformal coating over the substrates, which is regarded as a prerequisite for devices, such as perovskite solar cells.³ Moreover, the inherently slow and low-temperature reactions involved in CBD (typically below 100 °C) allow for nanoscale control over film growth, enabling performance optimization. However, TiO₂ typically requires an annealing process at temperatures in the range of 450–500 °C to ensure adequate charge transport via crystallization.⁴ Therefore, developing an annealing-free CBD process for forming TiO₂ heterojunctions has remained difficult due to the absence of suitable reaction strategies.

We previously reported the formation of the Cl-containing Ti(IV)–oxo cluster [Ti₃Cl₂(O)(OH)₉]⁻ in an aqueous solution at 27 ± 1 °C.⁵ In addition, at this low temperature, the Ti(IV)–oxo cluster promoted the subsequent polycondensation reaction (Ti–O–Ti formation) to form brush-shaped rutile nanoparticles. In this study, we explored a low-temperature CBD process using the Ti(IV)–oxo cluster solution under 80 °C that enables—for the first time—the in situ, stepwise hierarchical formation of rutile/anatase TiO₂ heterojunctions. This investigation is based on our important discovery that there is a critical temperature range between 60 and 70 °C that leads to the preferential crystal growth of rutile and anatase TiO₂ nanoparticles, respectively. Based on this insight, we developed a one-pot, two-step CBD protocol: (i) reaction at 60 °C to form a rutile TiO₂ bottom layer, followed by (ii) continued reaction at 70 °C to deposit anatase TiO₂ onto the rutile layer.

First, we examined the temperature-modulated polycondensation reactions of the Cl-containing Ti(IV)–oxo clusters (see Experimental in SI). The polycondensation reaction for obtaining TiO₂ nanoparticles is illustrated in Fig. 1a. Fig. 1b shows the X-ray diffraction (XRD) patterns of the powder samples after 24 h of reaction. Notably, a narrow temperature window between 60 and 70 °C enabled the unprecedented preferential growth of rutile and anatase TiO₂, respectively. Specifically, near single-phase anatase TiO₂ was precipitated at 70 °C (72 ± 2 °C), designated as TiO₂-70, by solely adjusting the reaction temperatures of the same precursor solution, while the rutile TiO₂ formed at 60 °C (60 ± 2 °C), designated as TiO₂-60. The XRD samples were further analyzed using Raman spectroscopy, and the spectra are shown in Fig. 1c. TiO₂-70 exhibited peaks for the Raman modes of E_g, B_1g, A_1g, and E_g at 155, 396, 514, and 635 cm⁻¹, respectively, corresponding to the anatase TiO₂ phase.⁶ TiO₂-60 was identified as the rutile TiO₂ phase based on peaks for the Raman modes of B_1g, E_g, and A_1g at 152, 435, and 602 cm⁻¹, respectively, and a multiphoton process at 249 cm⁻¹.⁷ Therefore, the crystal phases identified from the Raman spectra were consistent with those determined by XRD analysis.

Fig. 1 (a) Conceptual schematic of the polycondensation reaction of a Ti(IV)–oxo cluster to produce TiO₂ nanoparticles. (b) XRD patterns of TiO₂-60 and TiO₂-70 after 24 h of reaction. The temperature settings were 60 and 70 °C, respectively. (c) Raman spectra of the same samples as (b).

Transmission electron microscopy (TEM) analysis indicated that the 60 °C reaction yielded aggregated nanoparticles approximately 10 nm in size (Fig. 2a). The crystallite size estimated using the Scherrer equation for the 110 peak in the corresponding XRD pattern (Fig. 1b) was 9 nm. These results suggest that the individual particles of TiO₂-60 exhibited a single-crystal nature. The TEM image of TiO₂-70 (Fig. 2b) revealed a d-spacing of 0.35 nm for the (101) plane of anatase crystal phase, and the observed particles had a mean size of 3.5 ± 0.9 nm. Regarding rutile TiO₂ formation, HCl is likely eliminated during the polycondensation of the Cl-containing Ti(IV) clusters, leading to the formation of corner-sharing TiO₆ octahedra (Fig. 1a), which constitute the structural motif of the rutile phase TiO₂. This interpretation is consistent with our previous report on the synthesis of rutile TiO₂ synthesis at room temperature.⁵ Compared to the aggregated rutile TiO₂ particles obtained at the lower reaction temperature, the discrete particles observed for TiO₂-70 suggest that the anatase crystallites have a higher solubility. ATR-IR spectroscopy (Fig. S1) revealed that hydrogen-bonded water molecules at the surface of TiO₂ caused the broadening of the spectrum at approximately 2500–3500 cm⁻¹ for the 70 °C sample. This suggested that at the higher reaction temperature of 70 °C, a greater extent of Cl⁻ substitution by H₂O in the Ti(IV)–oxo cluster promoted two adjacent hydrolysis events on the same edge of a TiO₆ octahedron, resulting in the preferential formation of edge-sharing TiO₆ octahedra (Fig. 1a), which favor the formation of anatase-phase TiO₂.⁸ Notably, this represents an unusual case of low-temperature TiO₂ crystal growth, where a ligand in the Ti(IV) cluster plays a key structural role in temperature-modulated selective phase formation. Specifically, polycondensation at 60 °C preferentially yields rutile TiO₂, while a reaction temperature of 70 °C favors anatase TiO₂.

Fig. 2 TEM images and corresponding selected area diffraction (SAD) patterns of the two TiO₂ samples synthesized at (a) 60 and (b) 70 °C.

Building on the concept of crystal phase control, we investigated the direct TiO₂ thin-film CBD from an aqueous solution of Ti(IV)–oxo clusters. We primarily determined the TiO₂ phases based on HR-TEM images in addition to complementary grazing-incidence X-ray diffraction (GI-XRD). The phase-selective TiO₂ deposition was examined via a two-step CBD approach at specific reaction temperatures, as shown in Fig. 3a and Table S1. The resulting sample is hereafter referred to as TiO₂ (double). The initial temperature of the reaction solution was maintained at approximately 60 °C (the plateau at 62–65 °C) for 30 min, followed by an increase to above 70 °C (78–79 °C) for an extended duration of 250 min. HR-TEM imaging (Fig. 4) revealed co-crystallization of rutile and anatase TiO₂: approximately 3 nm anatase nanoparticles were deposited on a base layer of rutile particles, which were spherical to irregular in shape and ranged from 15 to 20 nm in crystallite size. Notably, crystal growth on fluorine-doped tin oxide (FTO) substrates proceeded more rapidly than in bulk powder synthesis, where similar-sized particles required 24 h of reaction. In addition, GI-XRD measurements of the film sample (Fig. S2) revealed weak diffraction peaks at approximately 42.6° and 49.5°, which were ascribed to brookite TiO₂ (ICDD PDF# 90-04140). However, one of the major peaks of the brookite phase, which corresponds to the (211) plane and is typically observed at approximately 30.8°, was not detected. Consistent with this observation, no prominent lattice fringes attributable to brookite TiO₂ were identified in the TEM images, suggesting that the formation of the brookite phase during the CBD process was minimal. The temperature profile exhibiting a maximum temperature plateau at 65 °C was performed for 6 h, denoted as TiO₂-65 (single) (Fig. 3b). Similarly, CBD performed at a maximum temperature of 78–79 °C (for 250 min out of a total reaction time of 6 h) is denoted as TiO₂-79 (single). The TiO₂ thin films on the FTO substrates obtained for TiO₂-65 (single) and TiO₂-79 (single) were biphasic rutile/brookite and triphasic rutile/brookite/anatase TiO₂, respectively, where brookite TiO₂ was determined to be a minor phase for both samples from TEM analysis (Fig. S3). TiO₂-65 (single) consists of uniformly formed rutile TiO₂ nanoparticles with sizes ranging from 5 to 10 nm that tend to aggregate. The TEM image of TiO₂-79 (single) shows the formation of tiny anatase nanoparticles (2–3 nm) along with larger rutile nanoparticles (approximately 15 nm). Several peaks (at approximately 30.8°, 32.7°, 42.6°, and 49.5°) were observed in the GI-XRD patterns of TiO₂-65 (single) and TiO₂-79 (single) (Fig. S4) that do not correspond to either rutile or anatase TiO₂, but were assigned to brookite TiO₂. Regarding synthesis, phase-pure brookite and biphase brookite/rutile TiO₂ have been reported to form under specific conditions, particularly in the presence of HCl, where reaction temperatures typically range from above 200 °C.⁹ The TiO₆ octahedra are incorporated into the two crystalline structures via a dissolution–precipitation mechanism. It is plausible that in our CBD employing Cl-containing Ti(IV)–oxo clusters (prepared using TiCl₄ as the Ti source), a biphase brookite/rutile TiO₂ structure is similarly formed under the applied conditions, despite the relatively low temperature of approximately 60 °C.

Fig. 3 Temperature profiles of the stepwise CBD approach: (a) two-step CBD to obtain TiO₂ (double) and (b) one-step CBD to obtain TiO₂-65 (single) and TiO₂-79 (single). Schematics illustrating the use of Ti(IV)–oxo cluster solution as a Ti(IV) source for rutile and anatase TiO₂ formation are shown in (a).

Fig. 4 HR-TEM image of the TiO₂ (double) thin film. The light green lines and yellow dotted lines demarcate lattice fringes of rutile TiO₂ and anatase TiO₂, respectively, with corresponding d-spacings.

Once the rutile phase of TiO₂ is formed, it is generally considered irreversible under standard conditions and does not revert to other polymorphs. In the first stages of the temperature profiles shown in Fig. 3, it is postulated that more dense rutile TiO₂ was formed for TiO₂ (double) than for TiO₂-79 (single) because the CBD time to reach 62 °C was 90 and 35 min for TiO₂ (double) and TiO₂-79 (single), respectively. As shown in Fig. 5a and b, for TiO₂ (double), TiO₂ seeds start to cover the FTO surface, uniformly depositing nanoparticles with dimensions of 15 nm after the 90 min reaction. It is reasonable to assume that the subsequent higher-temperature stage of over 70 °C promoted the hierarchical nucleation and growth of anatase TiO₂ on top of rutile TiO₂, resulting in a rough surface after 6 h of CBD (Fig. 5c). Banfield et al. demonstrated that anatase TiO₂ nanoparticles are more thermodynamically stable when their size is below approximately 14 nm because of their lower surface energy compared to other polymorphs.¹⁰ This supports the notion that the nanostructured surface of the TiO₂ (double) film—comprising 2–3 nm anatase particles—is more stable than that of TiO₂-65 (single). Furthermore, cross-sectional SEM images (Fig. 5d and Fig. S5) revealed that the TiO₂ particles were firmly attached onto the textured FTO surfaces for all samples with an approximate thickness of 100–150 nm. From the diffuse reflectance spectra of the TiO₂ thin films (Fig. S6) after 6 h of reaction, the bandgaps were estimated to be 3.36, 3.38, and 3.40 eV for TiO₂-65 (single), TiO₂-79 (single), and TiO₂ (double), respectively. The observed bandgaps are wider than those of bulk TiO₂, with the measured bandgaps being 0.3–0.4 and 0.1–0.2 eV larger than the values for rutile and anatase, respectively, which is due to the low dimensionality of the nanostructures.¹¹

Fig. 5 Top-view SEM images of (a) the FTO surface and (b) TiO₂ (double) on FTO after 90 min to reach 62 °C with a magnified inset, and (c) TiO₂ (double) on FTO after 6 h with a magnified inset. (d) Cross-sectional SEM image of the sample shown in (c). The arrows indicate regions with a thickness of approximately 100 nm.

We also investigated the electrical conductivity and oxygen vacancy formation on the surface of TiO₂ thin films deposited on FTO substrates (Table 1). As shown in Fig. S7, the TiO₂ (double) exhibits the highest conductivity, confirming its superior charge transport properties compared to the other TiO₂ thin films. The TiO₂ (double) formation involves a slower rutile TiO₂ formation process than TiO₂-79 (single), as illustrated in Fig. 3b, affording more densely packed rutile TiO₂, which may lead to increased conductivity. Given that anatase TiO₂ has a lower effective mass and higher electron mobility than rutile,¹² a rutile/anatase heterojunction is favorable for improving electron transport. Notably, the conductivity values obtained here are comparable to those reported for high-temperature-annealed TiO₂ compact layers used as electron transport layers in perovskite solar cells.¹³

Table 1 Summary of crystal phases and electrical conductivity of TiO₂ thin films

TiO2 films	Crystal phases^a	Conductivity (mS cm^−1)
a A minor brookite phase is also included.
TiO2-65 (single)	Rutile + brookite	3.3 × 10^−3
TiO2-79 (single)	Rutile + anatase + brookite	4.3 × 10^−3
TiO2 (double)	Rutile + anatase + brookite	5.9 × 10^−3

---

X-ray photoelectron spectroscopy analysis provided information regarding the chemical state of the TiO₂ thin films.¹⁴ The deconvoluted O 1s spectra shown in Fig. S8 exhibit two distinct peaks, where a primary peak at approximately 530.7 eV was attributed to lattice oxygen in TiO₂, and a secondary peak at approximately 532.2 eV was associated with oxygen vacancies and surface-adsorbed species. A reduced intensity of the 532.2 eV peak for TiO₂ (double) indicates a decrease in oxygen-related defects, which may reduce interfacial energy losses in photoenergy conversion systems. In addition, the intensity of the Ti 2p_3/2 peak was higher in TiO₂ (double), suggesting an improved surface crystallinity or increased titanium content. These findings highlight the superior electronic characteristics of the TiO₂ thin film obtained in the two-step CBD. Therefore, the second stage CBD at 70 °C likely plays a decisive role in controlling the nucleation and growth of anatase TiO₂ crystallites, enhancing electron transport.

Electrochemical measurements were also performed to investigate the band alignment of the TiO₂ materials. The onset potentials derived from the current–voltage curves under intermittent illumination (Fig. S9) indicated that the conduction band levels of all three TiO₂ substrates were comparable (−4.1 eV vs. vacuum). This suggests the absence of an energy cascade in the TiO₂-based system. Such a band engineering for the energy cascade could be achieved by a suitable metal-doping to elevate the conduction band level in the second step of CBD over 70 °C, which is in progress in our laboratory. Consequently, less-oxygen defective TiO₂ phase junction with anatase TiO₂-modification (that is, TiO₂ (double)) will have a major advantage for applications in photoenergy conversion.

In summary, we demonstrated the in situ formation of TiO₂ phase junctions via CBD using an aqueous solution of Ti(IV)–oxo cluster precursors. To the best of our knowledge, this is the first report of hierarchical rutile/anatase TiO₂ thin film formation via CBD. In particular, TiO₂ crystal growth on top of FTO substrates in CBD could be achieved in shorter reaction periods compared to powder synthesis. Our findings present new opportunities for applying TiO₂ heterojunctions in technologies, such as solar cells and artificial photosynthetic systems that harness diverse light-harvesting materials. For instance, the TiO₂ heterojunction film is expected to improve the performance of flexible perovskite solar cells that require low-temperature processing.¹⁵

K. M. acknowledges financial support from the Tokyo Ohka Foundation for the Promotion of Science and Technology, the Tatematsu Foundation, the Ogawa Science and Technology Foundation, and the General Incorporated Foundation International Club. This work was also supported by JSPS KAKENHI Grant (JP24K08492). The authors thank Dr Norimitsu Yoshida and Dr Michiyuki Yoshida for their support with the gold evaporation experiments and XRD measurements, respectively.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the SI.

ATR-IR spectra of TiO₂ samples; GI-XRD pattern of TiO₂ films; TEM and SEM images of TiO₂ films; Tauc plots of TiO₂ films; conductivity measurements; XPS analysis; photovoltammograms of TiO₂ electrodes. See DOI: https://doi.org/10.1039/d5cc03571f

Notes and references

1. P. Wang, W. Song and W. Li, ChemCatChem, 2024, 16, e202301489.
2. Y. Rana, H.-C. Chiu, D. Strandell, Z. Wang, S. Bessette, R. Gauvin, P. Kambhampati and G. P. Demopoulos, Nanoscale, 2023, 15, 2152–2161.
3. Y. Lu, M.-C. Shih, S. Tan, M. J. Grotevent, L. Wang, H. Zhu, R. Zhang, J.-H. Lee, J.-W. Lee, V. Bulovi and M. G. Bawendi, Adv. Mater., 2023, 35, 2304168.
4. J. Suo, B. Yang, E. Mosconi, H.-S. Choi, Y. Kim, S. M. Sakeeruddin, F.-D. Angelis, M. Grätzel, H.-S. Kim and A. Hagfeldt, Adv. Funct. Mater., 2021, 31, 2102902.
5. K. Manseki, K. Saka, M. Matsui, S. Vafaei and T. Sugiura, CrystEngComm, 2017, 19, 5844–5848.
6. H. C. Choi, Y. M. Jung and S. B. Kim, Vib. Spectrosc., 2005, 37, 33–38.
7. H. L. Ma, J. Y. Yang, Y. Dai, Y. B. Zhang, B. Lu and G. H. Ma, Appl. Surf. Sci., 2007, 253, 7497–7500.
8. N. M. Kinsinger, A. Wong, D. Li, F. Villalobos and D. Kisailus, Cryst. Growth Des., 2010, 10, 5254–5261.
9. M. Manzoli, F. S. Freyria, N. Blangetti and B. Bonelli, RSC Adv., 2022, 12, 3322–3334.
10. H. Zhang and J. F. Banfield, J. Mater. Chem., 1998, 8, 2073–2076.
11. A. Alberti, E. Smecca, S. Sanzaro, C. Bongiorno, F. Giannazzo, G. Mannino, A. L. Magna, M. Liu, P. Vivo, A. Listorti, E. Calabrò, F. Matteocci and A. D. Carlo, ACS Appl. Energy Mater., 2019, 2, 6218–6229.
12. J. Zhang, P. Zhou, J. Liu and J. Yu, Phys. Chem. Chem. Phys., 2014, 16, 20382–20386.
13. J. Peng, D. Walter, Y. Ren, M. Tebyetekerwa, Y. Wu, T. Duong, Q. Lin, J. Li, T. Lu, M. A. Mahmud, O. L. C. Lem, S. Zhao, W. Liu, Y. Liu, H. Shen, L. Li, F. Kremer, H. T. Nguyen, D.-Y. Choi, K. J. Weber, K. R. Catchpole and T. P. White, Science, 2021, 371, 390–395.
14. Y. Ko, T. Kim, C. Lee, C. Lee, Y. J. Yun and Y. Jun, Energy Environ. Mater., 2023, 6, e12347.
15. L. Yang, Q. Xiong, Y. Li, P. Gao, B. Xu, H. Lin, X. Li and T. Miyasaka, J. Mater. Chem. A, 2021, 9, 1574–1582.

---