Interfacial engineering of conformal titanium oxide nanofilms on porous carbon supercapacitor electrodes via atomic layer deposition

Interfacial engineering of conformal titanium oxide nanofilms on porous carbon supercapacitor electrodes via atomic layer deposition Remuel Isaac M. Vitto a,* , Hamed Pourkheirollah a , Jari Keskinen a , Steffen Vindt b , Andrew Cook c , L¯ ıga Gr¯ ınberga d , L¯ ıga Ignat¯ ane d , Gints Kuˇ cinskis d , Aleksandrs Volperts e , Amit Tewari a , Donald Lupo a , Paul R. Berger a,f , Matti M¨ antysalo a a Faculty of Information Technology and Communication Sciences, Tampere University, Finland b InnoCell ApS, Svendborg, Denmark c Beneq Oy, Espoo, Finland d Institute of Solid State Physics, University of Latvia, Latvia e Latvian State Institute of Wood Chemistry, Riga, Latvia f Department of Electrical and Computer Engineering, Ohio State University, USA HIGHLIGHTS GRAPHICAL ABSTRACT •Interfacial decoration of TiO 2 nanofilms on AC supercapacitors via ALD technique. •Highly uniform and conformal TiO 2 nanofilms observed in SEM/crosssectional TEM. •The TiO 2 /AC demonstrate higher capacitance during extensive chargedischarge cycles. •New insights on next generation supercapacitors based TiO 2 ALD-coated AC electrodes. ARTICLE INFO Keywords: Atomic layer deposition Titanium dioxide Growth per cycle Activated carbon Supercapacitors Printed electronics Specific capacitance Equivalent series resistance ABSTRACT We report the development of an atomic layer deposition (ALD) process for depositing titanium dioxide (TiO 2 ) nanofilms onto porous activated carbon (AC) electrodes to enhance the electrochemical performance of supercapacitors. This study investigates previously unexplored aspects of the ALD process, including the influence of precursor pulse duration and film thickness on the deposition behavior within complex porous AC structures. The deposited TiO 2 nanofilms were amorphous and exhibited excellent uniformity and conformality across the AC surface. Electrochemical measurements revealed a combination of surface redox and intercalation-type pseudocapacitance, with surface redox reactions identified as the dominant energy storage mechanism. An optimal balance between accessible surface area and film thickness was achieved at 60 ALD cycles, corresponding to a TiO 2 film thickness of approximately 2.3 nm. Under these conditions, the specific capacitance of TiO 2 -coated supercapacitors exceeded that of bare AC electrodes, owing to the additional pseudocapacitance contributed by * Corresponding author. E-mail address: [email protected] (R.I.M. Vitto). Contents lists available at ScienceDirect Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour https://doi.org/10.1016/j.jpowsour.2025.238691 Received 16 April 2025; Received in revised form 22 October 2025; Accepted 24 October 2025 Journal of Power Sources 662 (2026) 238691 Available online 30 October 2025 0378-7753/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). the TiO 2 layer. These findings demonstrate an effective strategy for utilizing ALD to engineer advanced electrode materials for high-performance supercapacitors. 1. Introduction Supercapacitor (SC) devices represent a highly promising energy storage technology capable of meeting the growing power demands of modern society. Nevertheless, their adoption in industrial applications remains limited due to their relatively low energy density compared to commercially available lithium-ion batteries (LiBs). Conventional market-ready SCs, typically based on carbonaceous materials such as activated carbon (AC), exhibit energy densities in the range of approximately 2.3–8 Wh kg −1 [1], whereas LiBs can achieve values as high as 280 Wh kg −1 [2,3]. Despite this shortcoming, SCs offer distinct advantages, notably their ability to deliver energy at much higher power outputs than batteries. Increasing the energy density of SCs to levels comparable with LiBs would address this major limitation and enable the development of next-generation energy storage systems capable of meeting the requirements of advanced technologies. It is important throughout this quest for superior energy dense SCs to employ non-toxic materials and maintain their eco-friendliness and noncarcinogenic, mutagenic, and toxic for reproduction (CMR) qualities. This consideration makes ACs derived from waste biomass materials a highly appealing baseline candidate electrode. The energy storage mechanisms for AC-based SCs are primarily through the formation of electric-double layer capacitance (EDLC) between the electrode surface and the electrolyte ions. A common strategy to enhance the energy density of AC SCs is to increase the overall capacitance by combining the EDLC with surface chemistry of nanostructures such as Transitional Metal Oxides (TMOs) that undergo faradaic redox reactions known as pseudocapacitance (PC) 0.[4,5,6] However, it is crucial that the TMO nanostructure decorations retain the unique and intricate porous structures of the substrates. This characteristic property of ACs is responsible for their extremely high surface area that allows more active sites in the formation of electric-double layers for higher capacitance. Several physical and chemical coating techniques of TMOs onto ACs and other carbon-based electrodes were developed over the years including hydrothermal [7], precipitation [8], calcination [9], electrodeposition [10], sol-gel [11], and chemical vapor deposition [12]. However, these material engineering techniques face several challenges regarding the preservation of porous structures including relatively large granules and agglomeration, reduced surface area due to the pore blockage, uncontrolled growth thickness, and lack of coating uniformity. In recent years, coating technology known as atomic layer deposition (ALD) has attracted the attention of researchers for its powerful ability to deposit films with excellent uniformity and conformality in the nanoscale. Furthermore, ALD involves cost-effective processes that parallel the sustainability objectives of ACs thanks to its precise atomic layer-by-layer growth of nanostructured films that significantly reduces the waste from excess elements [13]. The ALD technique has come a long way since its invention by Suntola and Antson in mid 1970s [14] and it is now making its way into the realm of electrochemical energy storage devices: LiBs, sodium-ion batteries (NiBs), and SCs [15–17]. Among the selection of various TMOs, titanium dioxide (TiO 2 ) is one of the most sought-after materials in electrochemical applications for its low-cost and excellent chemical stability [18]. In addition, electroactive materials such as TiO 2 possess remarkable advantages in their electrochemical properties as they reach the scale of nanoparticles. However, the carcinogenicity and mutagen toxicity of TiO 2 remains unclear [19, 20]. Hence, further evaluations are required to clarify its potential hazards. Wang et al. [21] fabricated electrodes based on TiO 2 nanofilms (7–30 nm) on glassy carbon substrates and determined their pseudocapacitive behavior as a factor of particle size. The voltammetric measurements showed increasing amounts of overall stored charge and significantly faster charge/discharge cycles as the particle size decreased below 10 nm. These findings further reinforce the benefits of utilizing the ALD technique to deposit TiO 2 nanofilms on AC electrodes to enhance its overall SC performance. Moreover, the three-dimensional (3D) interconnected porous structures and high surface area of ACs offer an effective solution to fabricate small enough TiO 2 nanoparticles. To counter the hindered capacitive abilities of bulk TiO 2 particles due to slow molecular transport of ions in the network, Brezesinski et al. [22] studied templated nanocrystal-based porous TiO 2 films and observed that both the mesoporous morphology and nanocrystals templates to be key in having high levels of capacitive charge storage due to shorter diffusion path lengths for ion transport. Moreover, the surface properties of the substrates for the ALD process must be considered as it heavily dictates the growth mechanism of the deposited TiO 2 films. In a study conducted by Wang et al. [23], the influence of surface energy on the film quality of ALD-coated TiO 2 films was explored on pristine carbon nanotubes (CNTs) and those with intermediate layers of aluminum oxide (Al 2 O 3 ) and zinc oxide (ZnO). The surface morphology characterization revealed more uniform and conformal TiO 2 films on the Al 2 O 3 /ZnO coated CNT substrates compared to the pristine due to their higher surface energy which facilitates the layer-by-layer film growth at the initial stages. Considering that each carbon allotrope possess unique surface properties to each other, it suggests a difference in the ALD growth mechanism of TiO 2 films between CNTs and ACs. Although previous works on TiO 2 ALD-coated ACs have been conducted by Tan et al. [24] and Bai et al. [25], both study lacks certain methodological frameworks and characterizations to provide insights in the growth mechanism. In particular, the ALD parameters and thickness of TiO 2 film were left unexplored and the film quality were assessed inadequately to show whether the surface morphology are individual bulk particles or a uniform coating. In addition, Tan et al. utilized titanium (IV) isopropoxide (TIP) while this work utilized titanium chloride (TiCl 4 ) precursor as the source for titanium. These two distinct precursors would consequently produce TiO 2 films with completely different properties and growth mechanism as made evident by a lower growth rate and need for higher temperature for TIP compared to TiCl 4 [26]. Besides the work by Bai et al. have considerable insufficient methodological transparency as they failed to outline the equipment and precursors they utilized for their ALD process. In this regard, there is still limited understanding in the growth mechanism of ALD-coated TiO 2 films on AC substrates and developing a technique that produces films with excellent uniformity and conformity on the complex 3D porous structure still remains to be a challenge. Looking from the electrochemistry perspective of TiO 2 ALD-coated AC electrodes, the existing literature provides only a portion of their electrochemical behavior among a vast selection of electrolytes. The works conducted by Tan et al. and Bai et al. both employed tetraethylammonium tetrafluoroborate (TEABF 4 ) and methyl triethylammonium tetrafluoroborate (MeEt 3 NBF 4 ) acetonitrile based organic electrolytes, respectively. In general, EDLCs with organic based electrolytes achieve a wide operating cell voltage but has low specific capacitance values due to their large ion size which could have difficulties in pore accessibility [27]. Such valuable interface kinetics should be explored using a different type of electrolyte, especially in the case of uniform and conformal TiO 2 films on an intricate porous structure of ACs. It would be interesting to observe their electrochemical behavior using a neutral aqueous electrolyte solution of sodium chloride (NaCl) which has significantly smaller ions of Na + and Cl − compared to the large Et 4 N + , MeEt 3 N + , and BF 4 − ions of previous studies. In this study, we present a detailed and systematic investigation of an R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 2 optimized ALD technique for depositing uniform, conformal TiO 2 nanofilms on porous AC electrodes for SCs. Earlier works [24,25] missed crucial methodological details and overlooked key factors that affect film quality. We carefully studied precursor pulse times, growth rates, and film thickness control to achieve excellent surface coverage while preserving the complex 3D porous network that drives electric-double layer formation. The deposition temperature is only 120 ◦C, much lower than the 180–300 ◦C usually reported [28–30]. This low temperature broadens ALD’s compatibility with thermally sensitive binders and flexible substrates. As a result, our method is practical for a wide range of SC designs, including printed and flexible devices that cannot tolerate high heat. Another distinctive feature is the use of neutral aqueous NaCl electrolytes instead of bulky organic ones. This simple change provides new insights into ion transport and interface kinetics, which were largely missing from earlier studies. Together, these advances close critical gaps in understanding TiO 2 ALD growth on ACs. They also introduce a scalable and energy-efficient coating method. This approach (i) achieves uniform and conformal deposition on complex 3D porous structures, (ii) adds surface-redox pseudocapacitance for higher energy storage, (iii) allows precise control of film thickness to optimize electrochemical performance, and (iv) offers mechanistic insights for designing highenergy-density SCs. In short, this work pushes SC technology toward environmentally sustainable, high-performance energy storage. 2. Experimental section 2.1. Electrode fabrication The activated carbon (AC) ink was prepared using 90 wt% Kuraray YP-80F AC as the electrode material and 10 wt% of a styrene-butadiene rubber (SBR) and carboxymethyl cellulose (CMC) binder mixture. Kuraray YP-80F is a coconut-shell-derived AC characterized by a high specific surface area (approximately 2200–2300 m 2 /g), well-controlled micropore size distribution, and low ash and moisture content. While the specific activation process for YP-80F is proprietary, AC materials such as YP-80F are typically produced by carbonizing carbon-rich precursors (e.g., coal or wood) followed by physical activation at elevated temperatures (600–1200 ◦C) using oxidizing gases such as steam or carbon dioxide. This activation process generates a porous structure with a large surface area, imparting the high adsorption capacity typical of AC materials. Prior to ink formulation, binder solutions were prepared by diluting CMC to 2 wt% and SBR to 40 wt% in deionized (DI) water. To prepare the AC ink, 28 g of the CMC solution and 0.12 g of laboratory-grade Triton X-100 (Sigma-Aldrich) were gradually added to 14 g of YP-80F AC under stirring. Subsequently, ~30 g of DI water was introduced, and the suspension was homogenized using an Ultra-Turrax T 10 basic disperser. Finally, 3.2 g of the SBR solution was added while stirring, and small amounts of DI water were incrementally incorporated until the desired viscosity was achieved. The resulting ink was transferred into a syringe and stored in a desiccator until use. The supercapacitors (SCs) were fabricated using YP-80F as the active electrode material and EDAG graphite ink as the current collector, deposited on 100 μ m Kapton polyimide (PI) film substrates. A schematic overview of the electrode fabrication process is presented in Fig. 1. For patterning, a 50 μ m Kapton PI film was employed as a stencil, with electrode shapes cut using a Silhouette Cameo 3 plotter. The PI film substrate was positioned beneath the stencil, and graphite ink was applied across the openings. A 220 mm mtv-messtechnik doctor blade coater was then used to deposit the graphite current collectors, which were subsequently dried in an oven at 95 ◦C for 1 h. The AC electrode layer was fabricated in a similar manner. The prepared YP-80F-based ink was printed atop the dried graphite layer and allowed to dry overnight at room temperature. After drying, the layer thicknesses were measured using a Mitutoyo 543-250B digital indicator. The graphite current collector and AC electrode exhibited thicknesses of approximately 25–35 μ m and 30–40 μ m, respectively. 2.2. Atomic layer deposition Titanium dioxide (TiO 2 ) nanofilms were deposited using a Beneq TFS 200 atomic layer deposition (ALD) system, directly onto the fabricated SC electrodes as well as 50 mm diameter silicon wafers (Siegert Wafer), Fig. 1. Schematic illustration showing the fabrication process of the graphite current collector and AC electrodes on PI substrates using doctor blade coater. The fabricated electrodes with/without ALD coatings are then utilized in a two-electrode system as symmetrical SC devices, and in a three-electrode system with Ag/AgCl reference electrode and platinum counter electrode. R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 3 which served as dummy substrates for thickness calibration. A schematic representation of a single ALD cycle is provided in Fig. 2. Titanium tetrachloride (TiCl 4 , Volatec Ltd.) was employed as the titanium precursor, while deionized (DI) water was used as the oxygen source. The overall surface reactions involved in the TiO 2 growth mechanism are summarized as follows [31]: (A)TiOH (s) + TiCl4(g)→TiO −TiCl3(s) + HCl (g)(1) (B)TiCl (s) + H2O(g)→Ti −OH (s) + HCl (g)(2) The sequential completion of reactions A and B constitutes one ALD cycle of TiO 2 deposition. The process begins at the hydroxyl (–OH) terminated surface of either the AC electrode or the previously deposited oxygen layer. Each ALD cycle consists of four steps: (i) the surface hydroxyl groups react with the TiCl 4 precursor; (ii) excess TiCl 4 and the HCl by-products are removed by nitrogen purging; (iii) the surfacebound TiCl species react with the H 2 O precursor; and (iv) excess H 2 O and the resulting HCl by-products are purged with nitrogen. These chemical reactions define the practical requirements for precursor pulse duration. In an ideal ALD growth mechanism, full surface saturation of TiO–TiCl 3 and Ti–OH species should occur during steps (ii) and (iv). Insufficient precursor pulsing restricts the formation of these critical intermediates, leading to non-uniform TiO 2 film growth. To ensure complete saturation, the pulse durations of both precursors were incrementally increased until a stable growth rate was obtained. The ALD process was carried out at a deposition temperature of 120 ◦C. Both nitrogen purge steps were fixed at 5 s, while the TiCl 4 pulse duration was varied at 50 ms, 100 ms, 150 ms, 200 ms, 250 ms, 300 ms, and 450 ms, in combination with H 2 O pulse durations of 150 ms, 300 ms, and 450 ms. After determining the optimal pulse times for both precursors, the number of ALD cycles was adjusted to investigate the effect of film thickness. The TiO 2 ALD-coated electrodes used in this work, along with the corresponding film thicknesses measured on coloaded silicon substrates, are summarized in Table 1. 2.3. Material characterization The growth behavior of TiO 2 nanofilms was examined using a Rudolph Research AutoEL ellipsometer to measure film thickness on coloaded silicon substrates. X-ray diffraction (XRD) analysis was performed with a Malvern Panalytical Empyrean system to study the crystal structure of the fabricated AC SC electrodes and the deposited TiO 2 coatings. Raman spectroscopy was used to analyze the orientation and surface defects of bare and TiO 2 ALD-coated AC electrodes. Measurements were carried out with a Renishaw inVia Qontor spectrometer equipped with a 50 ×objective confocal microscope. A 532 nm diodepumped solid-state laser operating at 30 mW was used as the excitation source. Each spectrum was recorded as the average of 10 acquisitions of 40 s, covering the range of 100–3200 cm −1 . X-ray photoelectron spectroscopy (XPS) was conducted to identify surface functional groups, using an Omicron Nanotechnology GmbH ESCA Model 3000 instrument under ultra-high vacuum (UHV) conditions with a base pressure below 1 ×10 −10 mbar. Monochromatized Al K α radiation (h ν =1486.5 eV) was employed as the excitation source. Data acquisition and analysis were performed with CasaXPS software (version 2.3.22 PR1.0). Spectral fitting included Shirley background subtraction and peak deconvolution using a symmetrical Gaussian–Lorentzian function, with equal Gaussian and Lorentzian contributions (50:50) for all component peaks. The surface morphology of AC electrodes before and after TiO 2 ALD deposition was characterized using a high-resolution TESCAN CLARA field-emission scanning electron microscope (FESEM). Elemental composition was analyzed with a Zeiss Ultraplus SEM equipped with an Oxford Instruments X-MaxN energy-dispersive X-ray spectroscopy (EDX) detector. The uniformity and conformality of TiO 2 nanofilms Fig. 2. Schematic illustration of the ALD of TiO 2 on AC electrodes involving a TiCl 4 precursor and H 2 O precursor pulses followed by a nitrogen purge. This four-step process constitutes one ALD cycle which is repeated over several times until the desired nanometer thickness is achieved [31]. Table 1 Summary of the ALD-coated AC electrodes used in this study. TiAC20 TiAC40 TiAC60 TiAC80 TiAC100 No. of ALD cycles 20 40 60 80 100 TiO 2 film thickness ≈1.6 nm ≈1.8 nm ≈2.3 nm ≈3.0 nm ≈4.0 nm R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 4 were investigated on cross-sectional regions of AC electrodes. Lamellae were prepared with a Helios 5 UX DualBeam focused ion beam scanning electron microscope (FIB-SEM). To protect the surface, a carbon layer was first deposited using an electron beam (e-dep. C), followed by a thicker platinum layer deposited with an ion beam (i-dep. Pt). The lamella surface was not exposed to electrons or ions prior to protective layer deposition. The final lamella thickness was below 100 nm. Crosssections of the AC electrodes were further examined with a FEI Tecnai G2 F20 transmission electron microscope (TEM) equipped with an EDX detector. Nitrogen sorption isotherms at 77 K were measured on the fabricated electrodes using a Quantachrome NOVA 4200K instrument. The surface area, pore volume, and pore size distribution were calculated using analytical models including MultiPoint Brunauer–Emmett–Teller (BET), Dubinin–Radushkevich (DR), and Density Functional Theory (DFT). 2.4. Electrochemical measurements The fabricated AC electrodes, with and without TiO 2 nanofilms, were assembled into symmetrical SC devices as illustrated schematically in Fig. 1. A pre-cut 3M 468 MP adhesive transfer tape was applied onto the graphite and PI substrate surface, leaving the AC layer exposed. A 40 μ m Dynacap GT 0.45/40 cellulose separator, soaked in a 5:1 mass ratio of water and NaCl (99.5 %, Sigma-Aldrich), was then placed on the exposed electrode area with adhesive. A second AC electrode was subsequently aligned in an opposing orientation and pressed firmly onto the prepared electrode to form a sealed, symmetrical cell. The assembled SCs were initially evaluated in a two-electrode configuration to determine their specific capacitance (Cs) and equivalent series resistance (ESR), using a Maccor 4300 workstation. These performance parameters were derived from galvanostatic charge–discharge (GCD) measurements in accordance with the IEC 62,391-1 industrial standard [32]. The protocol consisted of four charge–discharge cycles at applied currents of 1, 3, and 10 mA within the selected operating voltage window, with the fourth cycle held at the upper voltage limit for 30 min. To assess the impact of TiO 2 nanofilms under different conditions, the upper voltage limit was varied at 1.0, 1.2, and 1.4 V. The Cs and ESR values were then calculated using the following equations: Cs=4×IΔt mΔV(3) ESR =Vdrop ΔI(4) Here, I is the applied current; Δt is the discharge time between 80 % and 40 % of the fourth cycle at 1 mA; m is the combined mass of the two AC electrodes; V (drop) is the voltage at the IR drop of the fourth cycle at 10 mA; and ΔI is the change in current during V(drop). Following the initial evaluation, bare AC and TiO 2 -coated AC (TiAC60) symmetrical supercapacitors were subjected to additional GCD measurements within a potential window of 0–1.0 V at varying current densities, and their Cs values were calculated using the same procedure. Once the optimal TiO 2 nanofilm thickness was determined based on Cs and ESR, the electrochemical behavior of the corresponding ALD-coated electrodes was further investigated in a three-electrode configuration. This setup used a silver/silver-chloride (Ag/AgCl, redoxme AB) reference electrode, a platinum coil counter electrode (redoxme AB), and the same NaCl electrolyte solution. Cyclic voltammetry (CV) measurements were performed using a Zahner Zennium potentiostat. CV analyses were carried out within a potential window of 0–0.8 V vs Ag/AgCl, with the upper potential limit extended to 1.0, 1.2, and 1.4 V at a scan rate of 10 mV/s. Electrode–electrolyte stability was further examined in the potential window of 0–1.2 V vs Ag/AgCl by varying the scan rate from 10 mV/s up to 100 mV/s. 3. Results and discussion 3.1. Initial challenges and strategies of ALD Our early attempts to deposit nano thin films of TiO 2 were done with plasma-enhanced ALD (PE-ALD) processes, a coating technique that offers several advantages such as faster nucleation of precursors and higher growth rate. The thermal energy required to activate the surface for deposition is greatly reduced due to the highly reactive plasma species [33]. The PE-ALD technique was performed on AC electrodes based on our previous works [34] made from a mixture of commercial Kuraray YP-80F powder and chitosan binder (Sigma-Aldrich, 50494) with graphite (Acheson EDAG PF-407C) current collectors and aluminum/polyethylene terephthalate (Al/PET) substrates as shown in Fig. S1. The growth rate was calculated to be 1 Å/cycle and the number of ALD cycles varied between 20 and 100 resulting in film thicknesses of 2 nm and 30 nm, measured on co-loaded silicon substrates via ellipsometry. In contrast, the surface of the coated electrodes appears to have a strong discoloration of a dark metallic blue which obviously indicates that the TiO 2 coatings are thicker than 2 nm. This signifies an uncontrolled and non-saturating growth mechanism similar to chemical vapor deposition. More importantly, the coated electrodes disintegrate when it comes into contact with a liquid solution as shown in Fig. S2. This occurs when the binder of the AC electrode suffers extreme decomposition from the bombardment of the energy from the highly reactive plasma, causing the YP-80F powder to become loose and separate easily from the graphite current collector. Thus, the PE-ALD technique was deemed unfit for the application of this study, and thermal ALD was subsequently investigated. Besides the ALD processes itself, the various material components of the AC electrodes must be taken into serious consideration for compatibility. Because the traditional thermal ALD technique used in this study is performed with a deposition temperature of 120 ◦C and the PET component of the substrates have a glass transition temperature within the range of 50 ◦C–90 ◦C [35], a more thermally stable substrate should be utilized instead. Thus, Kapton polyimide films (PI, DuPont de Nemours, Inc.) were used as substrates in this study. Their favorable properties include high thermal stability which fits very well for ALD depositions, good electrical insulation ideal for electrode substrates, and excellent mechanical flexibility that is advantageous in wearable and portable device applications of SCs [36]. Following these, TiO 2 nanofilms were successfully deposited onto the AC-chitosan electrode with PI substrates at various number of cycles and subsequently assembled into symmetrical SCs for electrochemical characterizations as shown in Fig. S3. However, the Cs and ESR values reveal negligible differences with the pristine sample and declines beyond 100 ALD cycles. Upon closer inspection of the coated electrodes, the chitosan binder undergoes degradation similar to the PE-ALD, however, it is thermally induced and only becomes apparent after prolonged exposure to high temperature, such as during a high number of ALD cycles. Fig. S4 shows the exfoliation of the AC electrodes into bulk powders upon exposure to deionized (DI) water. For this reason, the AC-chitosan electrodes are incompatible for ALD applications and a mixture of styrene-butadiene rubber latex (SBR, MTI Corporation) and sodium carboxymethyl cellulose (CMC, Sigma-Aldrich) binder was substituted in subsequent experiments for its appealing properties. The CMC matrix acts as the main binder while the SBR improves mechanical flexibility and electrode adhesion [37]. This binder combination has an impressive thermal stability at a relatively high temperature of 288 ◦C with a negligible small weight percentage loss of 0.8 % [38]. 3.2. Material characterizations To critically assess the amount of TiO 2 deposited onto the AC surface per ALD cycle, the measured thickness is normalized by the number of cycles used and defined as the growth rate or growth per cycle (GPC). R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 5 Fig. 3a shows the computed GPC values of the TiO 2 layer versus the TiCl 4 pulsing duration at three different values of H 2 O pulse time using a fixed number of 100 ALD cycles. The growth rate is observed to increase at a faster rate when the TiCl 4 pulse time is increased within the range of 50 ms–150 ms. However, the growth rate began to saturate when the precursor pulse time reached 200 ms and eventually plateaued beyond 300 ms at 0.6 Å/cycle. This growth behavior signifies the saturation point in the ALD process which occurs at the “self-limiting region” where no further chemical reaction from the precursors occurs. Interestingly, higher GPC values and increased uniformity of refractive index are observed as the H 2 O pulse times are increased. For this reason, the pulsing duration for H 2 O was determined to be optimal at 450 ms and was kept constant in the succeeding ALD processes. Concerning the growth mechanism on the silicon substrates, there is not any significant difference from increasing the TiCl 4 pulse time further than 200 ms as they all saturate at 0.6 Å/cycle. Nevertheless, it becomes a critical aspect, and the difference is magnified during the actual depositions when the AC electrodes are present in the ALD chamber. The TiO 2 film thickness had a notable reduction to 0.4 Å/cycle which is probably caused by a phenomenon in ALD technology known as the “edge effect”. In general, this effect dictates a difference in the film growth mechanism for substrates with many features or edges compared to a smooth surface. Hence, the interconnected 3D pores of AC complicate the initial nucleation of the precursors making it more efficient for the process to grow on the smooth surface of silicon substrates in the early stages of ALD (0–20 cycles). As a result, higher growth rates are measured for 20 cycles compared to the succeeding values in Fig. 3b. However, the precursors will eventually pass the threshold of initial nucleation onto the complex surface of AC substrates once a sufficient thickness of TiO 2 has been deposited (>20 cycles). Afterwards, a transition period in the growth mechanism is observed where the precursor nucleation is more favorable on the AC substrate. Because of its significantly higher surface area, most of the precursors nucleate onto the surface of AC to form the layer-by-layer growth of TiO 2 and reduce the film thickness in the silicon substrates. This is even more accentuated by the setup of the ALD where each process contains six AC samples to only one silicon wafer in the middle. In addition, the growth rate of the actual depositions at varying number of ALD cycles were done at 200 ms and 450 ms of TiCl 4 pulse time as shown in Fig. 3b. Increasing the TiCl 4 pulse time results in higher GPC values which indicates a non-saturating growth behavior at 200 ms, although it is within the self-limiting region in Fig. 3a. Although that is to be expected since the 200 ms pulse time is at the early stages of saturating behavior. Obviously, it can also be argued that increasing the pulse time higher than 450 ms could compensate for the “edge effect” and the increased surface area of ACs to increase the growth rate back from 0.4 to 0.6 Å/cycle. However, an overly excessive precursor pulsing duration poses certain risks in the ALD equipment as it dramatically increases the possibility of precursor condensation in the narrow lines and cause blocking issues. In retrospect, the pulse time of 450 ms could already be considered to be a prolonged exposure of precursors compared to the short pulse times of 60–200 ms employed by relevant Fig. 3. Material characterizations of the TiO 2 nanofilms deposited via ALD. Growth per cycle behavior (a) as a function of TiCl 4 pulse time at different H 2 O pulse times and (b) as a function of ALD cycle amount at 200 ms and 450 ms pulse time of titanium tetrachloride. (c) XRD graphs of bare and coated AC electrodes, insets are the reference peaks for graphite and anatase structure. (d) Raman spectroscopy of bare and coated AC electrodes. (e) XPS survey spectra of bare and coated AC electrodes. Deconvoluted and fitted peaks for the regions: (f) C 1s, (g) Ti 2p, and (h) O 1s. R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 6 studies [24,30,31]. Furthermore, Ritala et al. also observed a similar growth rate of 0.4 Å/cycle on soda lime glass substrates using identical precursors of TiCl 4 and H 2 O [31]. This comparative growth rates suggests that 450 ms is already sufficient pulse time for depositing TiO 2 film in AC substrate. Following the analysis of TiO 2 growth mechanisms, the deposited films with varying thickness underwent comprehensive materials characterizations based on their crystallinity, structural defects, surface functional groups, surface morphology, and elemental composition. Fig. 3c shows the XRD spectra of bare AC and coated electrodes TiAC60 and TiAC100. To further analyze the corresponding peaks, reference peaks from the graphite (RRUFFID No. R050503) and anatase (RRUFFID No. R070582) crystal structures are shown as an inset at the bottom of the figure. All the samples exhibited a strong peak at 26.5◦and a slight peak at 54.5◦, consistent with the peaks present in the reference crystalline graphite. These two peaks correspond specifically to the (002) and (004) planes of the graphite crystal structure. Moreover, it can be observed that the crystalline graphite peaks reduce in intensity as the ALD cycles increase. This can be attributed to the increasing film thickness which decreases the penetration depth of the x-rays and reduces the signal coming from the AC layer. A few nm thin and most likely amorphous TiO 2 films are present on the surface, thus peaks corresponding to TiO 2 cannot be observed in the XRD data. Fig. 3d shows the Raman spectra of the bare and TiO 2 -coated AC electrodes where the presence of prominent D and G bands, and a weak 2D band signature are confirmed in the bare AC sample. However, in the case of the TiAC60 and TiAC100, the intensity of the prominent peaks seemingly diminishes and is unable to present any impression of the 2D band. This occurs due to the similar interactions which were observed in the XRD measurements where the signal generated from the AC layer decreases as the number of ALD cycles increases. Generally, the D band represents the inelastic scattering of the phonon from the defect sites and the G band infers the presence of ordered graphitic structure due to vibrational stretching of sp 2 hybridized carbon. Whereas the 2D band arises from a second-order overtone, as a result of the interactions between the incident laser and charge carriers causing two sequential inelastic scattering events from two distinct phonons instead of scattering from defects [39]. Aside from these characteristic peaks, a high intensity ratio between the D and G bands (I D /I G ) also provides information about the number of defects present in the surface. The I D /I G ratio was calculated to be 0.87 for AC, 1.01 for TiAC60 and 1.02 for TiAC100. The persistent structural disorder observed in AC is attributed to a high density of defect sites located at the edges of its flakes [40]. Although the AC samples coated with TiO 2 nanofilms have higher I D /I G ratio than the pristine sample, which infers higher surface defect sites, taking into consideration that these are muted peaks from the AC layer rather than the TiO 2 layer. Note that amorphous TiO 2 does not produce Raman peaks [41]. As such, the XRD and Raman studies provide inconclusive characterizations regarding the uniformity and overall quality of the TiO 2 nanofilms. Moreover, the surface morphology studies discussed later in this section provide a sounder explanation which has probably caused higher surface defects in the ALD-coated samples. The XPS survey scan of the bare and TiO 2 -coated AC electrodes presents all the constituent elements present on the surface as shown in Fig. 3e. From there, the peaks corresponding to titanium and oxygen such as O KLL, Ti 2s, O 1s, Ti 2p, and Ti 3p are already apparent and increase in intensity as the number of ALD cycles gets higher. While the C 1s peak decreases as the TiO 2 layer thickness increases, which is consistent with the XRD and Raman analysis. Each of the several significant peaks are then deconvoluted to provide a more in-depth analysis and reveal their specific functional groups. Fig. 3f shows the C1s spectrum which are mainly separated into four following components: sp 2 hybridized (C-C, 284.8 eV), hydroxyl (C-O, 285.9 eV), carbonyl (C=O, 286.8 eV), and carboxyl (O=C-O, 289.5 eV) [42–44]. The doublet peaks Ti 2p 1/2 (464.9 eV) and Ti 2p 3/2 (459.2 eV) associated to the Ti 2p are then shown in Fig. 3g. The peak separation of 5.7 eV between the two Ti 2p peaks is in excellent agreement with the reported values in previous relevant studies [28,29]. As expected, both of the Ti 2p peaks are exclusive to the TiO 2 -coated AC samples where the TiAC100 sample displayed the highest intensity. Lastly, the deconvoluted peaks associated with O 1s shown in Fig. 3h are primarily composed of only the carbonyl (C=O, 536 eV) and the hydroxyl (C-O, 533.1 eV) groups for the pristine AC samples [45]. Based on the chemical reactions involved in the ALD process of TiO 2 [31], the initial nucleation of TiCl 4 precursor occurs at the C-O groups present at the surface of AC. Afterwards, the number of C-O groups decrease as it gets replaced by a C-O-TiCl 3 bonding to initiate the succeeding layer-by-layer growth of TiO 2 as shown in Fig. 3f and h. In addition, Fig. 3h shows two additional peaks upon deposition of TiO 2 corresponding to titanium hydroxyl (Ti-OH) groups located at 532.4 eV and 531.7, and titanium oxide (Ti-O) groups located at 530.8 eV and 530.7 for TiAC60 and TiAC100, respectively. The Ti-OH peak is primarily due to the non-lattice adventitious oxygen atoms at the surface brought by the ALD deposition while the Ti-O peak corresponds directly to the crystal lattice of the deposited TiO 2 layer, specifically from the coordination of O 2- -Ti 4+ atoms to form the TiO 6 octahedron [46]. Thus, the XPS measurements undoubtedly confirm the presence of ALD-coated TiO 2 layers at the surface of the AC electrodes but further material characterizations such as SEM, EDX, and TEM are required to confirm their uniformity and conformity. To further assess the effects of the deposited TiO 2 nanofilms on the surface properties of AC substrates, a BET analysis were performed to provide comprehensive calculations on their surface area, pore volume, and pore size distribution. Fig. 4 shows the nitrogen sorption isotherms at 77 K and their corresponding arrangement of pore widths. The isotherm profiles of the bare and ALD-coated AC samples displays a type IV isotherms classification by the International Union of Pure and Applied Chemistry (IUPAC) [47]. Distinguished by their adsorption and desorption curves separated by a thin hysteresis loop which occurs in the relative pressure range of 0.1–1.0 p/p 0 . The rapid increase in adsorption rate below 0.1 p/p 0 directly corresponds to the one-layer nitrogen adsorption on the micropores while the area with decreased adsorption rate above 0.1 p/p 0 indicates the multilayer nitrogen adsorption on the macro and mesopores [48]. The isotherm profiles of both samples showed no significant difference in their adsorption and desorption curves confirming a similar morphology of their pore structure. It can be observed in the inset of Fig. 4a and b that the arrangement of the pores of AC and TiAC60 samples remains similar. Although, the TiAC100 sample showed a minimal decrease in intensity of its pores as shown in the inset of Fig. 4c which might be an indicative characteristic of pore blocking. Furthermore, the hysteresis loop occurs when a pore width reaches a certain critical width which triggers capillary condensation in the macro and mesopores [47,48]. Thus, the thinner separation in the hysteresis loop of the TiAC100 isotherm could be a result of fewer number of sites for capillary condensation which also suggests pore blocking. Table 2 shows the summary of calculated surface properties of the bare and ALD-coated AC based on the MultiPoint BET, DFT, DR analysis along with their mass normalized values. The sorptometry measurements showed negligible differences in the surface area and pore volume of the bare AC and TiAC60, especially even after these values were multiplied by the mass of the fabricated electrodes. However, the mass normalized values of TiAC100 have significantly lower surface area, micropore area, and pore volume compared to the previous two samples. As hinted by its nitrogen sorption isotherms and pore size distribution, having a high number of ALD cycles such as the case of TiAC100, would initiate several blocking of the porous networks which would lead to these lower values of the surface properties. In addition, the discrepancy in the average pore diameter and micropore width between all three samples could be considered insignificant. Although the increasing trend of the average pore diameter as the number of ALD cycles increases could also be explained by pore blocking. During the deposition process when the TiO 2 film thickness increases, the smaller R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 7 Fig. 4. Nitrogen sorption isotherms and pore size distribution (inset) for (a) bare AC, (b) TiAC60 electrodes, and (c) TiAC100. Table 2 Summary of material properties calculated from the sorptemetry analysis. MultiPoint BET surface area MultiPoint BET surface area (mass norm.) DR micropore area DR micropore area (mass norm.) DFT cumulative surface area DFT cumulative surface area (mass norm.) m 2 /g m 2 m 2 /g m 2 m 2 /g m 2 AC 85.24 0.26 76.48 0.24 57.42 0.18 TiAC60 78.51 0.28 71.30 0.26 51.54 0.19 TiAC100 78.47 0.22 70.23 0.20 53.67 0.15 DR micropore volume DR micropore volume (mass norm.) DFT cumulative pore volume DFT cumulative pore volume (mass norm.) Average pore diameter DR micropore width cm 3 /g cm 3 cm 3 /g cm 3 nm nm AC 0.02718 8.43 e −5 0.04643 1.44 e −4 2.398 1.504 TiAC60 0.02534 9.12 e −5 0.04433 1.60 e −4 2.466 1.594 TiAC100 0.02496 6.99 e −5 0.04523 1.27 e −4 2.511 1.450 Fig. 5. FESEM images at low and high magnifications of AC electrodes: (a,d) pristine, (b,e) TiAC60, and (c,f) TiAC100. (g) FESEM image of TiAC100 site selected for EDX mapping analysis of (h) C, (i) Ti, and (j) O. R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 8 the pore diameters are then the faster it would be filled and essentially clogging the pore. As a result, the smaller pores would be blocked leaving only the bigger pores for the overall quantification of the average pore diameter for the samples with thicker TiO 2 films. The surface morphology of the AC electrodes was characterized through high-magnification SEM images to analyze pore blocking behavior and quality of surface coverage of the TiO 2 nanofilms as shown in Fig. 5a–f. The typical layered graphitic structure of bulk AC particles is observed across all samples similar to the YP-80F morphology reported by Fic et al. [49] at comparable magnifications. Although, interesting nanostructures were spotted on selected areas of the surface of TiO 2 -coated AC electrodes. Higher magnification SEM images (Fig. 5e and f) showed cubic-shaped crystal nanostructures with varying particle diameters of around 100 nm–400 nm. The TiAC100 sample was further studied using EDX and elemental mapping analysis was performed on a site containing that nanostructure to study its composition as shown in Fig. 5g and S5a. The distribution of the elements on the surface of the TiAC100 sample reveals a density of sodium (Na) and chlorine (Cl) elements localized on the area of the cubic nano structures shown in Fig. S5b and c, respectively. The elemental composition and cubic morphology of the nanostructures found on the TiO 2 -coated AC electrodes are consistent with NaCl nanocrystals synthesized by Wang et al. [50]. The formation of these compounds most likely occurred during the ALD process as a side reaction between the sodium atoms from the CMC binder and the highly reactive TiCl 4 precursors or HCl by-products. Also, these NaCl nanocrystals could be considered as defect sites on the surface of the electrodes from a surface point of view and may have been the cause for higher I D /I G ratio for the TiO 2 -coated ACs. Nevertheless, Fig. 5h–j shows the elements of interest such as C, Ti, and O to have a uniform distribution across the surface of TiAC100 sample. Most notably, the distribution of Ti and O elements were found to be reduced in the area of the nanostructures. As no other visible structures on the surface of TiO 2 -coated AC electrodes were observed at high magnification SEM images and the surrounding graphitic structures remain consistent across all samples, the elemental mapping suggests the deposited TiO 2 layer are nanofilms with excellent uniformity and coverage. Thus, TEM images of the AC, TiAC60, and TiAC100 were utilized to determine the uniformity and conformity of the TiO 2 layers on the AC electrode surface. As anticipated, the cross-sectional area of the pristine AC electrode only contains the AC layer, and the electron-beam deposited protective layer made of C as shown in Fig. 6a. While on the ALDcoated samples, a very thin and uniform layer of TiO 2 which appears like a hairline crack in between the AC and C layer is shown in Fig. 6b. Going to the TiAC100 sample, Fig. 6c revealed a much thicker line of the TiO 2 layer. Fig. 6d and e shows a high-resolution TEM images for TiAC100 sample where the higher contrast between the individual layers confirm a uniform and conformal nanofilm quality for the deposited TiO 2 . The thickness of these observed TiO 2 layers can be easily estimated with the 5 nm scale bar to be somewhere in the range of 2–4 nm, which are in excellent consistency with the measured TiO 2 film thickness of 2.3 nm for TiAC60 and 4.0 nm for TiAC100 based on the computed GPC values. Fig. 6f shows the TEM image of TiAC100 for the selected site of the line scan and the inset shows intensity variations of the Ti K α and Ti L α x-ray emission lines across the sample, measured along a 30 nm line from left to right (indicated by the orange line with a red-cross). Both profiles display an increase in Ti signal intensity peaking when TiO 2 nanolayer is reached, around the 15–18 nm mark. Thereafter, decreasing in intensity which indicates that the TiO 2 is diffused into nearby areas of the AC material and minimally for the C layer. Furthermore, a fast survey scan using EDX analysis was performed to analyze the elemental composition of each layer. The TEM images of the cross-sectional area of TiAC60 and the different sites marked O 1 to O 4 used for the study are shown in Fig. S6a and b. In the EDX spectra presented in Fig. S6c, a primary carbon peak associated with the AC layer remained constant in intensity across all four sites. Aside from this, a minor peak for copper is observed at all the sites as a result of background radiation from the copper omniprobe grid and copper-based Fig. 6. TEM images of AC electrodes: (a) pristine, (b) TiAC60, and (c) TiAC100. (d,e) High-resolution TEM images of the TiO 2 layer on the TiAC100 sample. (f) TEM image of TiAC100 site selected for 30 nm line scan and its corresponding profiles (inset) at Ti L α and Ti K α . R.I.M. Vitto et al. Journal of Power Sources 662 (2026) 238691 9