TiO2 nanotubes for dye‐sensitized solar cells—A review

Abstract

TiO₂ nanotubes (TNTs) are a potential candidate for the photoelectrode in dye‐sensitized solar cells (DSSCs). In this review, emphasis is given to the fabrication methods of the TNT photoelectrode, including the anodic oxidation method, the hydro/solvothermal method, and the template method. Modification of TNTs to improve the power conversion efficiency (PCE) and the long‐term stability of DSSCs is also covered. The active area of the DSSC strongly correlates with the PCE. Therefore, evaluating and comparing cell efficiencies with the same active area would be important. Reducing the material and manufacturing costs of TNT‐based DSSCs will be an important future target.

INTRODUCTION

Dye-sensitized solar cells (DSSCs) represent a relatively new potential photovoltaics technology due to the simple and low-cost processes for their manufacturing and wide range of applications.^1-3 The power conversion efficiency (PCE) and stability of DSSCs⁴ have improved over time, the highest PCE now being 14.3%.⁵ A typical DSSC is composed of a photoelectrode, a dye, an electrolyte, and a counter electrode (CE).⁶ The PCE very much depends on the electron transfer “efficiency” at the different photoelectrode/dye/electrolyte/counter electrode interfaces⁷ and the electron transport “efficiency” at the different cell components (PE, electrolyte, CE) as well as on the current collection ability of the photoelectrode. Therefore, the photoelectrode materials are important to the further development of the DSSCs,¹ in particular because of their importance to light absorption and electron transport.

Among the semiconducting photoelectrode materials, mesoporous TiO₂ is the most popular for DSSCs.^1,8 Often commercial TiO₂ nanoparticles (NPs) with a band gap of 3.0-3.2 eV⁹ are employed as the photoelectrode, that offers fast electron transfer and a large number of contact sites to adsorb the dye molecules, that is, a high specific surface area.¹⁰ However, grain boundaries may lead to electron recombination, which results in a photocurrent loss, and also to a loss of light absorption in the near-infrared region.¹¹ As an alternative to TiO₂ NPs, TiO₂ nanotubes (TNTs) have been considered as a promising photoelectrode option for DSSCs.^12,13 TNTs show improved light scattering, fast electron transport and low charge carrier recombination, and simple control geometry, compared with the NPs.^14,15 In addition, the tubular structure provides a relatively large contact surface area to decorate with metal oxide NPs and adsorb dye molecules on the photoelectrode.¹⁶

In 2002, Uchida et al¹⁷ for the first time reported of a DSSC with TNTs as the photoelectrode (TNT length 100 nm, diameter 8-10 nm, active cell area 25 mm²), which was produced by a hydrothermal method, showing an PCE of 2.9%. Later, anodic TNTs with a tube length of 500 nm and 2.5 μm were reported showing an PCE of 1.6% and 3.3%, respectively.¹⁸ The highest PCE reported is 10.2% with nanotwinned TNTs in DSSC using N719 dye with an active area of 20 mm².¹⁹ This value is still lower than that of a TiO₂ NP-based DSSC (14.7%).⁵

The literature on using TiO₂ TNTs in DSSCs is still limited albeit the potential of this technology.^20,21 Here, we will present a state-of-the art of this technology. The focus is on the preparation methods of TNTs for DSSC and on the modification methods of TNTs to improve the PCE of DSSCs. Key parameter values and critical factors limiting the performance of TNTs are summarized. The stability issues of TNT-based DSSCs are also briefly discussed, an important topic for DSSCs in general. Finally, possible future research areas for promoting the development of TNT-based DSSCs will be proposed.

PRINCIPLE AND STRUCTURE OF DYE-SENSITIZED SOLAR CELL

The structure and operational principle of DSSCs are shown in Figure 1.^6,22 A typical DSSC is composed of a photoelectrode (typically around 10 μm thick mesoporous TiO₂ film comprising of around 20 nm nanoparticles, on a transparent conducting oxide-coated glass, TCO glass, substrate), a dye (eg, ruthenium metal complex dye “N719” or an organic dye), electrolyte (eg, I⁻/I₃⁻ redox couple in an organic solvent), and counter electrode (typically platinum on TCO glass).⁶ In a DSSC, the dye absorbs photons from incident solar light to generate electrons.⁶ The photons need to match the energy level difference of the dye. From the excited energy level of the dye, the electrons are injected into the conduction band (CB) of the TiO₂. Then, the electrons diffuse through the TiO₂ and are transported to an external electrical circuit to provide electrical current. The oxidized dye is reduced by the electrolyte redox mediator ion (such as I⁻, or Co²⁺ metal complex ion). The platinum CE collects the electrons from the external circuit and reduces the oxidized electrolyte redox mediator, thus closing the electrical circuit.²²

The performance of a DSSC is described by the power conversion efficiency. The PCE of the cell depends on the short-circuit current (J_sc, mA/cm²), the open circuit voltage (V_oc, V), fill factor (FF), and the incident sunlight (P_in, normally 100 mW/cm², AM 1.5G spectrum) on the cell.²² The PCE can be written as: [Image Omitted. See PDF]where the fill factor is defined as [Image Omitted. See PDF]V_mpp and J_mpp correspond to the voltage and current values at the maximum power point (MPP).

PREPARATION METHODS OF TNTS FOR DSSC

There are many methods to prepare TNTs, such as the anodic oxidation method,^23-30 the hydro/solvothermal synthesis method,^31-37 and the template method.^7,11-15 In the following, the preparation methods of TNTs for DSSCs are explained in more detail.

Anodic oxidation method

Most publications have reported fabricating TNTs by the anodic oxidation method, and there are 12 328 publications on the topic in the Web of Science Core Collection (keywords: TiO₂ nanotubes, and anodic oxidation, or anodization, time span:1945-2020).

The anodic oxidation method of the TNT preparation is carried out with a DC power supply at a constant voltage or current in a two-electrode system, where titanium (Ti) metal functions as the working electrode and typically Pt as the counter electrode. The setup is shown in Figure 2. It is worthwhile to mention that the Ti metal is cheaper than FTO glass, also having good flexibility, and a low sheet resistance, and it would be suitable for high temperature treatment.^38-41

Yun et al¹⁶ adopted the anodic oxidation method to prepare the photoelectrode TNTs/Ti at a constant voltage of 60 V in an electrolyte consisting of ethylene glycol (EG) containing 0.5 wt. % NaF and 5 wt. % H₂O. They varied the reaction time (1, 5 and 15 hours) to produce TNTs with the lengths of 3, 10, and 22 μm, yielding PCEs of 0.55%, 1.45%, and 1.88%, respectively, under a 60 minutes 1 sunlight soaking treatment. Luo et al²⁴ prepared vertically oriented TNTs on a Ti mesh by the anodic oxidation method at an constant voltage of 30 V with reaction times of 10-30 hours to adjust the tube length from 9 μm to near 20 μm. It showed the highest PCE of 2.66% when the tube length is ~18 μm. Yi et al²³ investigated the effects of the electrolyte water content and the anodizing time on the TNT-based DSSCs. It was found that when the water content was 2 vol %, the cells showed the best performance, due to the increased TNT diameter resulting in increased specific surface area of the tube surface and dye loading ability. In order to match the energy levels with the N719 dye, Xie et al⁴ prepared the TiO₂ nanotube photonic crystals (NTPCs) with a ~150 nm lattice by a two stage anodic oxidation method. The first step was the anodic oxidation method, and in the second step, they adopted a high periodic alternating current (anodic voltage ~60 V, time 30 seconds) and low current (0 A, 90 seconds) pulses to prepare the NTPCs by controlling the pulse number (N = 15, 30, 45 and 60) to obtain lattices with different thicknesses, and the details and results showed in Table 1, N = 5. Ji et al²⁵ realized bamboo-type double-walled TNTs (DBN) on Ti metal substrates. The benefit of the structure was improved dye loading and PCE. The dye loading of DBN-1 (double-walled, bamboo-type nanotubes, grown under AV conditions, with a sequence of 2 minutes at 120 V and 12 minutes at 40 V for 8 hours anodization) and DBN-2 (double-walled bamboo-type tubes, grown under AV conditions, with a sequence of 4 minutes at 120 V and 12 minutes at 40 V for 6 hours anodization) was 3.2 × 10⁻⁸ mol/cm² and 2.4 × 10⁻⁸ mol/cm², which was higher than the dye adsorption ability of smooth-walled TNTs (1.9 × 10⁻⁸ mol/cm²).

TABLE 1 Summary of TiO₂ nanotubes-based photoanode in DSSCs, preparation methods, cell size, dye, counter electrode, electrolyte, and corresponding photovoltaic parameters, under simulated AM-1.5 illumination (power 100 mW/cm²)

Number	Methods	Cell size (mm2)	Parameters of TNTs Length = L (μm) Pore diameter = D (nm)			Dye	CE	Electrolyte	Jsc (mA/cm2)	Voc (V)	FF	PCE (%)	Ref.
1	Anodic oxidation	36	L = 3.32 ± 0.06 D = 50.01 ± 2.35			Porphyrin (GD2)	Pt/FTO	A mixture of 0.1 mol/L LiI, 0.6 mol/L 1, 2dimethyl-3propylimidazoliumiodide (DMPII), 0.03 mol/L I2, and 0.5 mol/L t-butylpyridine (TBP) in acetonitrile	1.36	0.49	0.56	0.37	16
			L = 10.45 ± 0.53 D = 78.61 ± 5.55						2.33	0.48	0.53	0.59
			L = 22.32 ± 0.85 D = 98.67 ± 4.87						2.69	0.48	0.66	0.70
2	Anodic oxidation	100	L = 14.3			N719	SputteredPt/Ti foil	A mixture of 0.1 mol/L LiI, 0.6 mol/L DMPII, 0.05 mol/L I2, and 0.5 mol/L TBP in acetonitrile	4.3	0.7	0.62-0.66	2.04	24
			L = 17.3						x	x		2.27
			L = 18.3						x	x		2.32
			L = 19.2						6.5	0.67		2.30
3	Anodic oxidation	49	L = 5			N719	Coated Pt/FTO	A mixture of 0.3 mol/L LiI, 0.6 mol/L DMPII, 0.03 mol/L I2, 0.1 mol/L GuNCS, and 0.5 mol/L TBP in acetonitrile	2.02	0.6	0.42	0.50	23
			L = 10						3.02	0.69	0.49	1.03
			L = 15						3.95	0.71	0.55	1.54
			L = 20						5.2	0.70	0.54	1.96
4	Anodic oxidation	16	L = 15 D = 100			0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.1 mol/L LiI, 0.6 mol/L DMPII, 0.01 mol/L I2, 0.1 mol/L GuNCS, and 0.5 mol/L TBP in acetonitrile	10.26	0.790	0.723	5.9	119
			L = 22 D = 100						11.68	0.783	0.701	6.4
			L = 35 D = 100						13.12	0.762	0.661	6.6
			L = 30 D = 100						12.84	0.779	0.703	7.0
			L = 30 D = 100					0 wt % PMMA-EA	12.95	0.771	0.707	7.1
								4 wt % PMMA-EA	11.82	0.782	0.687	6.4
								7 wt % PMMA-EA	12.68	0.791	0.684	6.9
								10 wt % PMMA-EA	12.05	0.786	0.655	6.2
								13 wt % PMMA-EA	11.08	0.766	0.649	5.5
5	Anodic oxidation	16	10 μm NP			0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.1 mol/L LiI, 1.0 mol/L DMPD, 0.12 mol/L I2, 0.5 mol/L TBP, and 3-methoxy propionitrile (purchased electrolyte)	11.09	0.706	0.598	4.68	4
			10 μm NP +15c NTPC		L = 2.3				15.57	0.705	0.618	6.78
			~10 μm NP +30c NTPC		L = 4.5				16.10	0.706	0.625	7.1
			~10 μm NP +45c NTPC		L = 6.7				14.69	0.697	0.604	6.19
			~10 μm NP +60c NTPC		L = 9				13.22	0.693	0.613	5.62
6	Anodic oxidation	20	Cylinder NT (Notreatment)		L = 13	0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.60 mol/L BMIM-I, 0.03 mol/L I2, and 0.10 mol/L GTC in acetonitrile and valeronitrile (85:15 vol.) (purchased electrolyte)	10.18	0.76	0.553	4.28	93
			2 times TiCl4 -Cylinder NT						13.31	0.76	0.493	5.04
			Cone NT (Notreatment)						6.69	0.84	0.560	3.15
			5 times TiCl4 -Cone NT						15.63	0.79	0.655	8.09
7	Anodic oxidation	20	NT		L = 12 D > 500	0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.60 mol/L BMIM-I, 0.03 mol/L I2, and 0.10 mol/L GTC in acetonitrile and valeronitrile (85:15 vol.) (purchased electrolyte)	7.50	0.710	0.52	2.78	89
			NT + NP						12.90	0.754	0.60	5.84
			NT + NP + SiO2						14.70	0.754	0.61	6.71
			NT + NP + Al2O3						14.35	0.754	0.59	6.35
8	Anodic oxidation	50	SNT		L = 10 D = 100	0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.1 mol/L LiI, 0.05 mol/L I2, and 0.5 mol/L TBP in acetonitrile	5.4	0.68	0.57	2.01	25
			DBN-1		L = 10 Inner d = 70 D = 230				7.4	0.72	0.63	3.46
9	Anodic oxidation	12	S		L ≈ 10.5	0.2 mmol/L N719	Coated Pt/FTO	A mixture of 0.60 mol/L BMIM-I, 0.03 mol/L I2, 0.10 mol/L GTC, and 0.50 mol/L TBP in acetonitrile and valeronitrile (85:15 vol.)	11.74	0.72	0.51	4.3	94
			S-H						14.31	0.79	0.63	7.12
			S-T						12.77	0.77	0.67	6.54
			S-H-O2						15.64	0.77	0.62	7.75
10	Anodic oxidation	x	Reference NT		L = 15 D = 120	0.3 mmol/L N719	x	x	11.03	0.76	0.484	4.06	98
			Ta 0.03 at%						14.62	0.70	0.401	4.44
			Ta 0.1 at%						14.87	0.72	0.475	5.09
			Ta 0.4 at%						15.68	0.71	0.438	4.88
11	Anodic oxidation	20	Air-450			0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.60 mol/L BMIM-I, 0.03 mol/L I2, and 0.10 mol/L GTC in acetonitrile and valeronitrile (85:15 vol.) (purchased electrolyte)	15.42	0.78	0.6306	7.58	19
			Air-550						15.88	0.8	0.6344	8.06
			Air-650						16.77	0.82	0.6297	8.66
			Air-750						17.04	0.78	0.6199	8.24
			O2-450						15.83	0.79	0.6285	7.86
			O2-550						16.25	0.834	0.6220	8.48
			O2-650						17.10	0.883	0.6225	9.4
			O2-750						17.33	0.864	0.6112	9.15
			1TiCl4-O2-650						17.34	0.897	0.6382	9.94
			2TiCl4-O2-650						17.71	0.891	0.6483	10.23
			3TiCl4-O2-650						18.08	0.873	0.6076	9.59
12	Anodic oxidation	16	TNTA/FTO		D = 126	0.5 mmol/L N719	Sputtered Pt/FTO	A mixture of 0.5 mol/L LiI, 0.3 mol/L HMII, 0.05 mol/L I2, 0.3 mol/L NMB, and 0.5 MTBP in 3-ethoxypropionitrile	10.15	0.745	0.62	4.69	86
			Au-TNTA		D ~ 126				10.86	0.722	0.66	5.18
			MH-TNTA-3		L = 15.2 D = 66				16.25	0.765	0.69	8.54
			H-TNTA-3		L = 15.2 D = 66				15.31	0.771	0.67	7.91
			MH-TNTA-3 (TiCl4treatmen)		D ~ 66				16.85	0.785	0.68	8.93
13	Anodic oxidation	x	TNT-DSSC		L = 33 D = 120	0.3 mmol/L N719	Sputtered Pt/FTO	A mixture of 0.60 mol/L BMII, 0.03 mol/L I2, 0.10 mol/L CuSCN, and 0.5 mol/L TBP in acetonitrile and valeronitrile	12.27	0.62	0.6301	4.76	85
			TiCl4-DSSC		L = 33 D ~ 100				21.50	0.66	0.6237	8.90
			TNT-TNA-DSSC		L = 33 D ~ 60				24.78	0.71	0.5596	9.86
14	Anodic oxidation	x	Untreated		L = 19.3 D = 120	0.3 mmol/L N719	Sputtered Pt/FTO	A mixture of 0.60 mol/L BMII, 0.03 mol/L I2, 0.10 mol/L CuSCN, and 0.5 mol/L TBP in acetonitrile and valeronitrile	11.56	0.707	0.6042	4.94	116
			0.025 mol/L HCl						12.80	0.739	0.6400	6.05
			0.05 mol/L HCl						14.12	0.735	0.6334	6.57
			0.1 mol/L HCl						16.81	0.728	0.6896	8.44
			0.1 mol/L HCl						15.21	0.724	0.6184	6.81
15	Anodic oxidation	260-300	TNT		L = 33	0.3 mmol/L N719	Sputtered Pt/FTO	A mixture of 0.60 mol/L BMII, 0.03 mol/L I2, 0.10 mol/L CuSCN, and 0.5 mol/L TBP in acetonitrile and valeronitrile	12.39	0.62	0.545	4.12	71
		13	TNT		L = 33				11.47	0.63	0.668	4.86
			TNT-TiCl4 treatment		L = 33				22.76	0.65	0.614	9.02
16	Anodic oxidation	25	NPs coated on TNTs/Ti		L ~ 5 D ~ 150	0.3 mmol/L N719	Coated Pt/FTO	A mixture of 0.60 mol/L BMII, 0.03 mol/L I2, 0.10 mol/L GTC, and 0.5 mol/L TBP in acetonitrile and valeronitrile (85:15 vol.)	9.36	0.79	0.66	4.82	13
		100							9.69	0.75	0.61	4.50
		400							9.68	0.73	0.61	4.35
		900							9.72	0.67	0.45	2.92
		25	NPs/FTO						17.90	0.81	0.55	7.92
		100							14.70	0.82	0.54	6.56
		400							13.20	0.80	0.28	2.96
		900							5.99	0.77	0.25	1.18
17	Hydrothermal	20	P25			0.5 mmol/L N719	Pt/FTO	A mixture of 0.1 mol/L LiI, 0.6 mol/L PMII, 0.05 mol/L I2, and 0.5 mol/L TBP in a mixture of acetonitrile and valeronitrile (85:15 vol.)	13.715	0.758	0.74	7.7	31
			P25 + G2						15.925	0.752	0.75	9.036
			P25 + C-TNT						15.709	0.741	0.759	8.856
			P25 + C-TNT Comp.						16.319	0.737	0.764	9.201
			P25 + HF-TNT						17.275	0.738	0.764	9.743
			P25 + HF-TNT Comp.						18.135	0.732	0.766	10.168
18	Hydrothermal	28.3	T0P1			0.3 mmol/L N719	Sputtered Pt/FTO	Oligo-PEGDME electrolyte	6.93	0.60	0.48	1.99	60
			T0.1P1						7.80	0.59	0.46	2.14
			T1P1						2.97	0.58	0.57	0.96
			T3P1						0.60	0.6	0.59	0.21
			T10P1						0.2	0.56	0.52	0.06
			T10P1					DMPII ionic liquid electrolyte	20.72	0.7	0.71	10.27
19	Hydrothermal	14-18	TiO2 NPs			0.3 mmol/L N3	Coated Pt/FTO	Commercially electrolyte AN-50 (Solaronix)	Best cell				7
									14.0	0.70	0.61	6.0
			Nb-doped TiO2 NPs						14.3	0.68	0.67	6.5
			TiO2 NTs						14.6	0.73	0.66	7.1
			Nb-doped TiO2 NTs						15.0	0.72	0.75	8.1
20	ZnO template	36	TiO2 90 min	x		0.5 mmol/L N719	Coated Pt/FTO	A mixture of 0.1 mol/L LiI, 0.60 mol/L DMPII, 0.05 mol/L I2, 0.10 mol/L GTC, and 0.5 mol/L TBP in acetonitrile and valeronitrile (85:15 vol.)	0.76	0.76	0.65	0.37	65
			TiO2 1.5 h	L = 1.6 ± 0.1 Do = 87 ± 9 Di = 50 ± 6					1.41	0.77	0.67	0.73
			TiO2 4.5 h	L = 1.7 ± 0.1 Do = 130 ± 10 Di = 40 ± 5					2.47	0.75	0.59	1.10
			TiO2 6 h	L = 1.8 ± 0.1 Do = 200 ± 12 Di = 40 ± 5					3.42	0.76	0.68	1.77
			TiO2 15 h	L = 2.7 ± 0.4 Do = 850 ± 20 Di = 235 ± 15					5.06	0.76	0.64	2.46
21	ZnO template	25	TNT	L = 20		0.5 mmol/L N719	sputtered Pt/FTO	A mixture of 0.60 mol/L BMII, 0.03 mol/L I2, 0.10 mol/L CuSCN, and 0.5 mol/L TBP in acetonitrile and valeronitrile	7.0	0.84	0.75	4.4	64
22	AAO template ALD	x	CE-DSSC closed	L = 8 D = 70		0.5 mmol/L N719	FTO	A mixture of 0.60 mol/L BMII, 0.03 mol/L I2, 0.10 mol/L GTC, and 0.5 mol/L TBP in acetonitrile and valeronitrile (85:15 vol.)	2.32	0.71	0.38	0.63	67
			OE-DSSC open						3.81	0.75	0.43	1.17

Note: x: the data are not shown in the related publications.

Employing the anodic oxidation method to prepare TNTs directly on Ti metal is appealing not only because Ti is an ideal candidate for a DSSC substrate due to its high bending ability under external force,²⁴ but also because the resulting TNTs are aligned in a highly ordered manner. They have a high length-to-diameter ratio (Figure 3) and excellent electron transporting performance, they can be fabricated conformally over large areas, the Ti metal and TNTs/Ti acts as an electrode in electrochemical devices, and they are feasible for several practical applications.^24,42-54 Some publications reported sputtered Ti metal (sputtered Ti thickness 1-2 μm) on transparent conductive oxide (TCO) glass substrate to prepare TNTs by the anodic oxidation method,^55-57 to avoid the losses caused by the necessarily employed back illumination decreasing the light-harvesting capability, when the TNTs are prepared on the opaque Ti metal.

Hydro/solvothermal method

The hydrothermal method is an effective and cost-effective way to prepare TNTs,^31-37 and there are 3903 publications on the topic in the Web of Science Core Collection (keywords: TiO₂ nanotubes, hydrothermal, solvothermal, time span:1945-2020).

The typical hydrothermal method to prepare TNTs consists of mixing commercial P25 TiO₂ powder and an aqueous NaOH solution together in a Teflon autoclave at a temperature of 100-150°C,⁵⁸ which is also suitable for the preparation of TNTs/TiO₂ NP composites used for DSSC photoelectrode fabrication (SEM is shown in Fig. 4).^34,59-62

Qadir et al³¹ prepared highly functional TNTs by a hydrothermal method, which showed 68% higher dye loading ability than conventional TNTs and also 50% higher photocatalytic degradation rate. The PCE of the highly functional TNT-based DSSCs was 10.1%. First, a solution was prepared with 12 mol/L NaOH aqueous solution [160 mL, in deionized water (DI) water] and 3 g of new precursor SG-T0200 TiO₂ nanoparticles with magnetic stirring for 1 hour. SG-T0200 particles (Sukgyung AT) are much larger than the conventional precursor nanoparticles (P25), resulted in a 68% enhancement of dye loading. After 15 minutes of ultrasonication, the solution was heated at 130°C with stirring for 48 hours. When the solution was cooled down to room temperature, 0.05 mol/L HCl (aq.) was added to maintain its pH at 1.0. After a DI water washing and a centrifugation process, the prepared slurry was placed in the hydrothermal instrument again, with another heating and stirring process at 200°C for 12 hours. In the end, it was annealed at 430°C for 4 hours to obtain the anatase TNTs. Tsvetkov et al⁷ reported Nb-doped TNTs by using Nb-doped TiO₂ NPs as a starting material, which were first dissolved in 100 mL of 10 mol/L NaOH aqueous solution and autoclaved for 12 hours at 120°C. Then, 150 mL of 0.1 mol/L HCl (aq.) was added in the solution, followed by centrifugation in water and ethanol to obtain Nb-doped TNTs. The ratio of TNTs and TiO₂ NPs can affect the charge transfer and electron lifetime, further impacting the PCE in DSSCs⁶⁰ (Table 1, number 18). In order to prepare well-aligned hierarchical TiO₂ nanotubes (HTNTs) in a simple and cost-effective way, Chen et al³² adopted a one-step hydrothermal method with potassium titanium oxalate, ethanol and H₂O. In the beginning, TiO₂ collosol was spin coated on fluorine-doped tin oxide (FTO) glass substrate to form a seed layer. After mixing potassium titanium oxalate, ethanol, and H₂O in a Teflon liner and stirring for 1 hour, an FTO glass with the seed layer was immersed in the solution and heated in the oven at 200°C with varying times of 3, 6, 9, 12, and 15 hours to produce hierarchical TNTs with the tube lengths of 12, 16, 18, 20, and 22 μm, the corresponding PCE of the cell were 3.43%,7.45%, 9.30%, 9.89%, and 8.36%. Not only well-aligned hierarchical TNTs applied in the assemble of DSSCs, but also pine tree-like TNTs were synthesized by a similar one-step hydrothermal method in a Teflon-lined stainless steel autoclave with the chemicals of PTO, water, and diethylene glycol at 200°C for 11 hours.^32,63 The pine tree-like TNTs give a potential to improve the adsorption ability of dye on the photoelectrode to enhance the PCE of DSSCs.

Template-assisted method

There are 1350 publications on the topic of template method to prepare TNT in the Web of Science Core Collection (keywords: TiO₂ nanotubes and template, time span:1945-2020).

Using ZnO nanowires as a template to prepare TNTs is a common method to manufacture TNT-based DSSCs.^64-66 It is carried out by immersing a ZnO nanowire template in a TiO₂ sol, ethanol and water for 30 seconds and repeating the procedure 20 times to form a TiO₂ shell with a thickness of 20-40 nm on the ZnO nanowires. Next, the obtained core/shell structure is annealed immediately at 350°C, followed by an etching process with 10 mmol/L TiCl₄ solution at room temperature, for the transformation of the core/shell structure into TNTs (Fig. 5).⁶⁴ Also Zhang et al⁶⁵ reported a template method to prepare TNTs by immersing a ZnO nanowire template in an aqueous solution of 0.075 mol/L (NH₄)₂TiF₆ and 0.2 mol/L H₃BO₃ at room temperature for 90 minutes to 15 hours. Lee et al⁶⁷ employed anodic aluminum oxide (AAO) as a template for an atomic layer deposition technique to fabricate open-end and close-end TNTs. Using titanium tetraisopropoxide (Ti[OCH(CH₃)₂]₄) and water as precursors to produce the TiO₂-deposited AAO, the sample was annealed at 450°C for 3 hours to form the anatase crystal structure. After removing the AAO template, crystalline anatase TNTs were obtained.

Summary of performance of TNT photoelectrodes in DSSCs

A summary of TNT-based DSSCs using different manufacturing methods is given in Table 1 with key parameters shown. A detailed analysis of the TNT photoelectrode is given in the following Section.

MODIFICATION OF TNT PHOTOELECTRODE TO IMPROVE THE EFFICIENCY OF DSSC

The specific surface area of TNTs is lower than that of the conventional TiO₂ NPs, which limits the amount of the adsorbed dye molecules on the photoelectrodes and therefore decreases the solar cells' light-harvesting efficiency. Thus, many modification methods have been investigated to minimize the aforementioned issues in TNTs in order to enhance the overall power conversion efficiency of TNT-based DSSCs.^1,68 Some results of the modifications of TNTs to fabricate DSSCs are summarized in Table 1. A lot of work has also been carried out to tune the energetic band levels of TNTs to better match those of the dyes⁵⁸ and to enhance the electron transfer in TNTs.¹⁶ The most reported modification methods are explained in the following.

Geometry of the TNTs and the resulting films

Larger surface area of the TNT films can be obtained by increasing the tube length and tuning the pore size of the tubes. With increasing the tube length, the electron lifetime and the diffusion length can be increased, followed by a significantly increased photocurrent density (J_sc).^{16,24,38,66,69-76} 3, 10, and 22 μm long TNTs yielded solar cell PCEs of 0.37%, 0.59%, and 0.70%, respectively.¹⁶ Joseph et al⁷⁷ found out that the counter electrode material in the electrochemical preparation of TNTs has an effect on the tube length and also the PCE in DSSCs. Platinum, titanium, iron, graphite pencil, and charcoal rod CEs yielded TNT lengths of 102, 130, 38, 55, and 88 μm, respectively, with a 24 hours anodization. Nyein et al⁷⁸ prepared TNTs in the electrolyte of KOH/fluoride/EG, LiOH/fluoride/EG, and H₂O/fluoride/EG at 60 V for 1 hour with the DC device and obtained TNTs with lengths of 18, 15, and 10 μm, respectively, and the corresponding PCEs in DSSCs were 3.0%, 2.7%, and 1.3%. The pore size of the TNTs can affect the dye loading and the light absorption ability. Qadir et al³¹ prepared highly functional TNTs, with large pore diameters, resulting in a 68% improved dye loading and enhanced light absorption. Xie et al^4,79,80 adopted a periodic alternating high current and low current pulses to prepare TiO₂ NTPCs with a ~150 nm lattice constant, which had well matched band gap with the light absorption range of N719 dye under AM 1.5G.

Decorating TNTs with nanoparticles

Decorating the TNTs with nanoparticles (NPs) is a common modification method to increase the active area of TNTs to improve the PCE of TNT-based DSSCs,⁸¹ which is due to the enhanced dye absorption ability and promoted charge transfer.^67,82,83 TiO₂ NPs are the most used NPs to enhance the PCE in DSSCs.^70,81,84-87 Zhang et al⁸⁵ employed a sol-gel method to introduce TiO₂ NPs to decorate TNTs, which yielded the highest DSSC PCE of 9.86% (pristine TNT-DSSC, PCE = 4.76%) due to the increased specific surface are for enhanced dye absorption.

Depositing a thin layer of Al₂O₃^88,89 or SiO₂⁸⁹ on the surface of TNTs improved the PCE too, because the recombination rate of the charge carriers was reduced. ZnO^90-92 layer was utilized to decorate the TNTs to enhance the DSSC performance, resulting in lower recombination rate and charge transfer resistance.

TiCl₄ treatment

The TiCl₄ treatment is not only a common modification method of TiO₂ NP-based in DSSCs, but also one of the most used modification methods to increase the specific surface area and improve the electron collection efficiency of TNT films in DSSCs. Schmuki's group^2,89,93 as well as other groups^{43,44,64,71,72,84,86,88,94-97} modified TNTs with the TiCl₄ treatment (different times TiCl₄ treatment) to improve the electron collection and increase the surface area for enhancing the dye loading ability. The TiCl₄ treatment lead a back-side illuminated TNT-based DSSCs to yield a considerable PCE close to 8%, due to resulted in a multiple layered decoration to give a further improvement of the electronic properties.²

Doping

Doping is a mature and effective way to modify the electronic structure of TNTs. Ta-doped TNTs layers were prepared on TiTa alloy foils (Ta 0.03 at. %, 0.1 at. %, and 0.4 at. %) = by the anodic oxidation method. The PCE of DSSCs with Ta-doped TNTs was improved by 9.77%, 124.8%, and 121.1% when compared with pure TNTs.⁹⁸ Also C,⁹⁹ N,¹⁰⁰ Cr,⁵⁸ Cu,³ Nb,⁷ and Li¹⁰¹ doping of NTs to fabricate DSSCs has been investigated with PCEs of 6.59%, 7.91%, 8.69% (Cr/Ti atomic percentage is 7.50%), 0.30% (5.2% Cu doped), 8.1%, and 6.4% (c(Li⁺) = 50 mmol/L). Introducing noble metal Au^86,102,103 and Ag NPs^{83,87,104-107} into TNTs can improve the PCE of the DSSCs, due to the enhanced light harvesting via surface plasmon resonance. The electrophoretic deposition technology was applied to deposit reduced graphene oxide,¹⁰⁸ TiO₂ NPs,^96,109 Au NPs,^99,102 and Ag NPs¹⁰⁵ to modify TNTs to enhance the PCE of DSSCs. Both a hydrothermal process and an O₂ plasma exposure improved the dye absorption on the TNT photoelectrode⁹⁴ by creating rough surfaces and hydroxyl groups on the TiO₂. Fu et al⁸⁶ prepared multi-hierarchical TNTs by introducing Au clusters on the walls of the TNTs by the photoreduction approach and filling TiO₂ NPs into TNTs, yielding a PCE of 8.93%, which is 190.4% better than that of the pristine TNT-based DSSCs. The Au NPs directly injected hot electrons into the semiconductor to accelerate electron transfer and improve the light harvesting in the DSSCs.⁷⁹

Open-end TNTs

Open-end TNT-based DSSCs are often reported to have a higher PCE than closed-end TNT-based DSSCs.^67,110,111 Lin et al¹¹² obtained a DSSC PCE of 9.1% with open-end TNTs, which was explained by an enhanced light harvesting and electron collection ability due to the open-end TNTs helping the redox electrolyte easily reaching the NP-TiO₂ underlayer, which was printed on the FTO to fabricated the open-end TNT-based DSSC. Zhu et al⁴⁴ employed abrasive paper (3000 mesh) to abrade the fixed membrane or adopted manual grinding polisher (~100 rpm) to remove the closed bottom caps of anodic TNTs to achieve open-end TNTs, which resulted in a DSSC PCE of 7.7%, a 66% enhancement compared with the pristine TNTs based DSSCs.

Annealing conditions

The annealing temperature^{19,100-101,113-115} and atmosphere¹⁹ affect the crystallization of the TNTs, further impacting the electron transport in the resulting films. However, the thermal treatment in high temperature (>750°C) usually worsens the dye loading ability. Therefore, there is a trade-off between the annealing temperature, the dye loading, and the PCE of DSSCs. So et al¹⁹ reported that both increasing the annealing temperature from 450°C to 650°C and changing the annealing atmosphere from air to oxygen improved the PCE of TNT-based DSSCs due to the formation of a crystalline anatase structure.

Other modifications

Yun et al¹⁶ exposed the DSSCs to a light soaking in simulated sunlight (100 mW/cm², AM1.5G spectrum) for 5-60 minutes and obtained the highest efficiency enhancement of 168% on TNT15 sample (anodization time was 15 hours, tube length was 22 μm). The use of hydrochloric acid (HCl) to introduce hydroxyl groups on the surface of TNTs improved the dye loading from 137.8 nmol/cm² (untreated TNTs) to 230.0 nmol/cm² (0.1 mol/L HCl-treated TNTs).¹¹⁶ The effect of the Ti metal substrate surface roughness on the PCE of TNT-based DSSCs has also been investigated. Combining mechanical polishing by sandpaper and electropolishing to give a very smooth Ti surface, yielded a higher TNT-DSSC PCE (1.61%) than that where the TNTs were grown on a rough substrate (0.72%).¹¹⁷

SCALE EFFECTS AND STABILITY OF TNT-BASED DSSCs

For industrialization of DSSCs, large-area modules will be necessary.¹¹⁸ However, typically the DSSC PCE decreases with increasing cell size.¹³ TNT photoelectrodes on the other hand often ease the area dependent deterioration of the DSSC performance because of the good electron transporting abilities of the TNTs.¹³

Table 1^13,71 shows that increasing the cell size decreases the PCE in both TiO₂ NP-based and TNT-based DSSCs. When the cell size was 25 mm², the PCE of the TiO₂ NP-based and TNT-based DSSCs were 7.92% and 4.82%, respectively. When the cell size was increased to 900 mm², the PCE of the TiO₂ NP-based and TNT-based DSSCs decreased to 1.18% and 2.92%, respectively. The drop was clearly less with the TNT structure, which implies that the TNT-based DSSCs could have a potential for large-area application.

The highest PCE achieved with TNT-based DSSCs is over 10%.^19,31 However, for commercial applications of TNT-based DSSCs, long-term stability is another important factor. Unfortunately, the data available on this topic are very limited: only two studies related to the long-term stability of TNT-based DSSCs were found. Seidalilir et al¹¹⁹ tested the stability TNT-based DSSCs with a liquid electrolyte and a polymer-based gel electrolyte containing poly (methyl methacrylate-co-ethyl acrylate) (7 wt.%) at 1 Sun illumination intensity and at 50°C for 1000 hours. The cells with liquid electrolyte degraded significantly after light soaking for 500 hours. The cell with the gel electrolyte retained 90% of the PCE after 1000 hours of light soaking. Hou et al¹¹⁰ reported the PCE and degradation rates of TNT-based DSSCs placed in air for 0, 1, 2, 7, and 14 days. They showed that the PCE of the best performing cell with (initial PCE of 5.01%) decreased by 17.4% after 14 days light soaking under 100 mW/cm² (AM 1.5G).

CONCLUSIONS

TiO₂ nanotubes are promising as photoelectrode material for dye-sensitized solar cells. Here, we have reviewed relevant literature on TNT-based DSSCs. The review identified three main methods to prepare TNTs for DSSCs: the anodic oxidation, the hydro/solvothermal, and the template method. The anodic oxidation method is the most used one, as it is simple and conformally controllable fabrication method yielding highly ordered nanotubes perpendicular to the substrate having a high length-to-diameter ratio.

The PCE of TNT-based DSSCs can be improved by modifying the TNTs through several ways, such as controlling the reaction time to adjust the geometry of the TNT in the preparation process, depositing metal oxide nanoparticles on the TNTs, TiCl₄ treatment, doping the TNTs, creating open the end of the tubes, and using different annealing conditions. Also, the relationship of the DSSC active area and the PCE was discussed. By increasing the cell size, the PCE of the DSSCs typically decreases, but for TNT-based DSSCs this is less dramatic than for the TiO₂ NP-based DSSCs.

It can be concluded that good electron transport and the possibility of large-area use are important positive attributes to the TNT-based DSSCs. However, the PCE of DSSCs still needs to be improved for commercial applications.¹²⁰ The highest efficiency of TNT-based DSSCs is presently 10.2%.¹⁹ For the fabrication of TNT-based DSSCs, mostly N719 dye, thermally coated or sputtered Pt counter electrode and I₃⁻/I⁻ redox mediator-based electrolyte are often used. A more systematic research of TNT-based DSSCs using different modification methods, dyes, electrolytes, and even other counter electrode structures/materials would be useful in order to find ways to improve the PCE. In addition to improving the efficiency, the question of the long-term stability of TNT-based DSSCs will need more scientific efforts in the future, as currently very little research on their stability is reported. It is also important to reduce the manufacturing costs of TNT-based DSSCs, by, for example, using cheaper materials.^121-126

ACKNOWLEDGMENTS

This work has been supported by the China Scholarship Council (CSC), No. 201706250038, the Academy of Finland Flagship Programme, Photonics Research and Innovation (PREIN), No. 320167, and the Jane and Aatos Erkko Foundation ASPIRE project (Finland).