Room temperature TiO₂ atomic layer deposition on collagen membrane from a titanium alkylamide precursor

Atomic layer deposition (ALD) is well known for its unique capabilities within chemical vapor deposition processes. It offers excellent composition tunability, and precise thickness control and uniformity in deposited very thin metal oxides films.^1–3 It can also facilitate conformal deposition across three dimensional substrates.^4,5 ALD of an oxide is a gas phase cyclic process, and one cycle of ALD process typically consists of four main steps: precursor pulse, precursor purge, oxidizer pulse, and oxidizer purge. Pulsing time, purging time, and reaction temperature are important parameters for optimal growth during ALD processes.⁶ Different oxidizers and precursors play important roles in the thin film growth of a given metal oxide. For a typical ALD process, the oxide film thickness generally increases linearly with increasing number of ALD cycles. The substrate also plays a significant role in the growth rate and oxide film quality.⁶ 

ALD is typically carried out at a relatively high temperature to obtain optimal film growth and quality. High temperature ALD is used in the semiconductor industry.¹ However, low temperature or room temperature ALD is needed when the substrate is heat-sensitive. Precursors with high vapor pressure can facilitate this type of ALD process. This low temperature ALD enables deposition on heat-sensitive substrates such as polymers, organic and biological materials. Knez et al. reported different low temperature ALD processes in their review.⁷ In most of those ALD processes, water (H₂O) was used as the oxidizing agent. In 1994, Gasser et al. first published successful room temperature ALD of silicon dioxide (SiO₂) from tetraisocyanate silane [Si(NCO)₄] and H₂O.⁸ Room temperature ALD of cadmium sulfide (CdS) from dimethylcadmium [Cd(Me)₂] and hydrogen sulfide (H₂S) was also reported by Lou et al.⁹ ALD below 50 °C was reported previously for boron trioxide (B₂O₃) [at 20 °C from boron tribromide (BBr₃) and H₂O (Ref. 10)], SiO₂ [at 27 and 30 °C from silicon tetrachloride (SiCl₄) and H₂O (Refs. 11 and 12)], aluminum oxide (Al₂O₃) [at 33, 35, and 45 °C using trimethylaluminum (TMA) and H₂O (Refs. 13–15)], and TiO₂ {at 35 °C from titanium isopropoxide [Ti(OiPr)₄] and H₂O (Ref. 14)}. These processes were used to deposit thin films on heat-fragile polymer substrates.

After low temperature ALD was reported, it became an increasingly used tool in nanotechnology to functionalize surfaces of different biomaterials. Such a type of attempt was first documented on tobacco mosaic virus on which Al₂O₃ and TiO₂ were deposited at 35 °C using TMA and titanium(IV) isopropoxide [TIP, Ti(OiPr)₄] precursor, respectively, and water as oxidizer.¹⁴ Next, ALD was performed on some proteins such as ferritin¹⁶ (deposition of Al₂O₃ and TiO₂) and S-layers¹⁷ (deposition of HfO₂). ALD was also performed on peptides^18–20 and on DNA molecule^16,21 to synthesize functionalized nanofibers or nanowires. Recently, applications of ALD were reported on other types of biomaterials such as naturally occurring collagen (collected from inner shell membrane of hen's egg),^22,23 spider silk,²⁴ cellulose fiber (from paper^25,26 and cotton^27,28), bristles of sea mouse,²⁹ butterfly wings,^30,31 fly eyes,³² legumes,³³ legs of water strider,³⁴ etc. Some of these materials had high aspect ratio structures where ALD might be the only possible way to obtain uniform conformal coating of an oxide layer.

Many of the above mentioned ALD were used for TiO₂ and Al₂O₃ thin oxide layers. Due to its attractive physicochemical properties, like large band gap,³⁵ chemical stability,³⁶ high dielectric constant,^37,38 highly photoactive surface,³⁶ high refractive index,^39,40 and nontoxic environment-friendly nature,³⁶ TiO₂ has a wide range of applications in modern technologies such as photocatalyst in solar cell,⁴¹ waste water purification,⁴² oxygen gas sensor,^43,44 memory devices and capacitors,⁴⁵ food additives,⁴⁶ pharmaceuticals,⁴⁷ and in paint and cosmetics.⁴⁵ Therefore, this oxide appears to have become a preferred coating material to functionalize biomaterials.

In this study, we carried out ALD of TiO₂ at room temperature in a custom made ALD reactor, on a commercially available resorbable collagen membrane (Fig. 1). Collagen is a naturally occurring fiber and one of the most abundant proteins found in the animal kingdom. ALD on fibrous materials has sparked a lot of interest as ALD can improve and functionalize these materials through successful infiltration into their complex structure. These ALD coated fibers have many applications such as in photovoltaic cells, chemical separations, barrier layers for medical, food packaging, organic electronics, surface engineering, textiles, biomedical industry, and other industries.^48–50 Previously, ALD was reported on different organic fibrous materials (Table I) like cellulosic fibers of cotton and paper,^{26–28,48,49,51,52} spider silk,²⁴ hen's inner egg shell membrane derived-natural collagen.^22,23 Cellulose consists of polysaccharide (d-glucose units) and it has better resistance to heat treatment unlike protein-based biomaterials. Kemell et al. first attempted TiO₂ ALD on natural cellulose fiber from paper at 150 and 250 °C.²⁶ In their study, TiO₂ replica of cellulose structure was obtained at 150 °C, and a photocatalytic crystalline TiO₂/cellulose composite was prepared at 250 °C. Cellulose fibers from cotton were also used as ALD substrates where the cellulose was highly crystallized. Hyde et al. performed Al₂O₃ ALD on this type of cotton fiber at 100 °C.²⁷ They reported a uniform growth on convoluted fiber surface and this growth behavior was reported to be significantly different from that on planar surface processed at the same conditions. In their follow-up work, they also showed the tunability of wetting behavior of ALD-treated cotton fibers by varying the oxide film thickness, although the role of surface roughness changes with processing may need to be studied further.²⁸ 

ALD of a biocompatible TiN film was also reported on cotton fiber by Hyde et al.⁵¹ In their study, increased cellular adhesion was observed for the thin (<10 nm-thick) and most hydrophobic ALD coating, whereas the cell adhesion density decreased on the thicker (>100 nm-thick) ALD coating. Jur et al. reported Al₂O₃ ALD at 60–90 °C (Ref. 48) and ZnO ALD at 115 °C on cotton fiber.⁵² Uniform Al₂O₃ growth was reported after 100 ALD cycles. Also, an effectively conductive cotton fiber was obtained with a ZnO ALD coating. Follow-up work by Lee et al. showed transition of the wetting behavior after performing ALD of Al₂O₃ and ZnO on cotton fibers.⁴⁹ After an initial ALD, the cotton surface became hydrophobic, from hydrophilic, but with further ALD cycles, the surface became hydrophilic again. Apart from cellulose, ALD was also reported on some other types of fibrous biomaterials such as spider silk and inner egg shell membrane. Lee et al. reported increased toughness in dragline spider silk after deposition of ZnO, TiO₂, and Al₂O₃ at 70 °C.²⁴ A photocatalytic polycrystalline film was also obtained on eggshell membrane with ALD of TiO₂ and ZnO.²² Recently, TiO₂ ALD on collagen extracted from eggshell matrix was reported at 70 °C using TIP precursor and water as oxidizer.²³ ALD on this naturally occurring collagen substrate was carried out at higher than 60 °C (i.e., 70–300 °C) using different oxidizer and precursor source from those in our study.

In our study, it is the first time room temperature TiO₂ ALD was performed on commercially available collagen membrane using a novel precursor for this process, tetrakis(dimethylamido)titanium (TDMAT), and ozone as oxidizing agent (Fig. 1). This can open up the potential of tuning the properties of commercial and other collagen.

The deposition of TiO₂ was performed in a custom-made tubular, hot wall ALD reactor.⁵³ The reactor can be heated up to 600 °C and its base pressure is a few mTorr. This reactor has four precursor delivery lines and is capable of delivering four different types of oxidizers: ozone, oxygen, water vapor, and small molecular weight alcohols. During the deposition, the substrate and reactor were at room temperature while the precursor bubbler was kept at 50 °C and the delivery line in between bubbler and reactor was kept 20–30 °C higher than the bubbler temperature. The ALD chamber pressure was kept at 500 mTorr during deposition. TDMAT (Sigma Aldrich, 99.999%) and ozone (generated just upstream the ALD chamber with custom made UV lamp system) were used as precursor and oxidizer, respectively. The precursor and the oxidizer were introduced sequentially into the reactor using computer controlled pneumatic valves. Argon (99.999%) was used as precursor carrier gas and purging gas. Geistlich Bio-Gide^® (Geistlich Biomaterials), commercially available collagen membrane was used as substrate and p-type Si(100) silicon wafer (University wafer, Inc.) was used as reference substrate to measure the deposited film thickness. The thickness of the deposited TiO₂ film on the reference silicon substrate was measured using spectral ellipsometry (Model M44, J.A. Woollam Co., Inc.). A custom made sample holder was used to hold the collagen sample and the reference silicon wafer inside the chamber during deposition (Fig. 2). The surface chemical composition of the control collagen surface and ALD TiO₂-coated collagen substrates was studied using a high resolution x-ray photoelectron spectrometer (Kratos AXIS-165, Kratos Analytical, Ltd., UK) equipped with a monochromatic Al Ka (1486.6 eV) x-ray source operating at 15 kV and 10 mA. Surface morphology of native collagen and TiO₂-coated collagen was analyzed using a high resolution field emission scanning electron microscopy (SEM) (JEOL JSM-6320F, JEOL, Inc.). Prior to SEM, samples were gold-coated using a sputter coater to make them conductive. imagej software was used to measure the fiber diameters from the SEM images of samples. Glancing incidence x-ray diffraction (GIXRD) spectra of deposited TiO₂ film were obtained using a high resolution x-ray diffractometer (X'Pert, PANalytical, BV Co., Netherlands) configured with 0.1542 nm x-ray emission line of Cu. Diffraction spectra were collected at an incidence angle of 1° to enhance sensitivity for thin films and to reduce substrate interference.

Room temperature growth behavior of TiO₂ deposited on our reference silicon substrate was investigated for 50–600 ALD cycles. Figure 3 shows the deposited TiO₂ film thickness on silicon substrate as a function of the corresponding number of TiO₂ ALD cycles. The standard deviation of the measurements at different positions across the substrate is indicated by the error bars in the graph. The film thickness was found to increase linearly with increasing number of ALD cycles without any apparent growth lag or rush. The growth rate is found to be 0.06 nm/cycle from the linear fit of the data. The apparent linear growth behavior indicates that this ALD process offers a surface saturated growth and excellent tunability of film thickness. Additionally, the small error bars in Fig. 3 are indicative of the uniformity of the room temperature ALD of TiO₂ films across the substrate surface. Deposition of TiO₂ from TDMAT and ozone was reported before for different substrates at higher temperatures, 60–300 °C.^{36,45,54–59} In 2008, Katamreddy et al. first reported TiO₂ ALD using TDMAT and ozone with a growth rate of 0.065 nm/cycle at 225 °C; they also reported an increase in the deposition rate at temperature higher than 225 °C due to the decomposition of the TDMAT precursor.⁵⁶ The deposition rate varies with process temperature, and with gradually increasing temperature it first decreases, then reaches saturation and finally increases again. Specifically, with increasing substrate temperature from 75 to 150 °C, the TiO₂ film growth rate first decreases from 0.052 to 0.045 nm/cycle; at a temperature 150–250 °C, a saturation phase of the growth rate at ∼0.046 nm/cycle is found; a further increase in the temperature results in strongly increasing growth rate again.^45,57 Rose et al. also reported a growth rate of 0.04 nm/cycle at 180 °C.⁵⁹ At lower temperatures, 60–65 °C, the growth rate remained ∼0.06 nm/cycle,^54,55 which is similar to the one obtained at room temperature (20–25 °C), in our study.

Figure 4(a) shows the XPS spectra of native collagen substrate (i.e., control sample), collagen-150 cycles, collagen-300 cycles, and collagen-600 cycles. Three distinct peaks were observed with the control collagen samples. Those peaks are attributed to oxygen (O 1s: 532 eV), carbon (C 1s: 285 eV), and nitrogen (N 1s: 398 eV), which are the basic chemical components of collagen protein. In addition to O, C, and N peaks, two other titanium peaks (Ti 2s: 560 eV and Ti 2p: 455 eV) were observed on the collagen samples with as-deposited TiO₂ after 150, 300, and 600 ALD cycles.

The highest levels of carbon (63.5%) and nitrogen (15.7%) were detected in the control collagen, and these levels were found to decrease after the ALD of TiO₂ thin films on collagen (Table II). With increasing number of the TiO₂ ALD cycles, the Ti 2p peaks became more intense. These results on collagen are in qualitative agreement with our finding of increasing TiO₂ film thickness with increasing number of ALD cycles on the silicon reference substrate. The Ti peaks in the 454–468 eV region are the Ti 2p_3/2 and Ti 2p_1/2 peaks and are mostly attributed to Ti⁴⁺.^45,56 Table II shows the O:Ti ratio/stoichiometry estimated from quantitative analysis of the XPS data. This data revealed the ratio of O:Ti of the first 6–10 nm of the TiO₂ film deposited on the collagen substrate to be 2.7–3.4, indicating excess oxygen within the film, while the film mainly consists of TiO₂. Similar ratio of O:Ti was reported before for TiO₂ film grown from alkylamide-ozone system on inorganic substrates.^56,57 Yet, those depositions were carried out at significantly higher substrate temperatures (150–250 °C); thus, the O:Ti ratio apparently seems to be minimally affected by the deposition temperature used.⁵⁷ 

Additionally, the C 1s peak at 285 eV is attributed to C–C bonding and this indicates the existence of mostly organic-bonded carbon on the film.⁵⁷ The collagen protein structure itself contains several types of carbon bonds like C–C (285 eV), C–O (286 eV), and C=O bonds (288 eV).^23,27 As a result, XPS detected larger amount of carbon on the control sample and also on the collagen-150 cycles one. With a further increase in the number of ALD cycles, the signal from collagen C–O and C=O could barely be detected. The collagen-300 cycles and collagen-600 cycles samples exhibited peaks attributed only to C–C bonds, indicating mostly the presence of adventitious carbon [Fig. 4(b)]. Considering the sensitivity of the biological substrate, the collagen samples were not sputtered with an Argon (Ar⁺) beam before XPS scans, and this might lead to the presence of adventitious carbon on the ALD-coated collagen in spite of having a thicker TiO₂ coating. To investigate this issue further, XPS scans were performed on the corresponding Si reference samples (after 150, 300, and 600 TiO₂ ALD cycles) before and after sputter cleaning. The carbon content (at. %) of Si-150 cycles, −300 cycles, and −600 cycles before sputtering was found to be 28.3%, 34.0%, and 31.5%, respectively, while after sputter cleaning the carbon content was 7.4%, 5.7%, and 5.6%, respectively. This suggests that most of the carbon on the ALD-coated collagen is indeed adventitious carbon present on top of the ALD TiO₂ thin film. Additionally, ∼18% Ti and ∼48% O was detected before sputtering for these three reference Si samples with ALD TiO₂, while after removing the adventitious carbon through sputtering the Ti and O content were determined to be ∼30% and ∼62%, respectively. This result also indicated the deposition of mostly pure (uniform) TiO₂ film.

Surface morphology of the collagen samples was studied using SEM. Figure 5 shows the SEM images of all four sample groups. There is an apparent difference of fiber morphology among the samples. In the control sample, the collagen fibers are seen to be thin. The fibers became increasingly thicker in the collagen samples with increasing number of TiO₂ ALD cycles.

The average outer diameter of single fibers was measured from SEM images of each sample using the imagej software. For each sample group, three fibers were used for measurement and those fibers have been labeled with arrows in Fig. 5. Both conjugated fibers and single fibers are present in the collagen membrane; only single fibers were chosen. The results of these measurements are presented in Fig. 6. The single fiber diameter average of the control sample was 44.3 ± 2.9 nm; after 150, 300, and 600 TiO₂ ALD cycles, it was 60.3 ± 2.2, 106.7 ± 6.3, and 220.3 ± 17.3 nm, respectively. A stronger than linear increase in fiber diameter is therefore observed for as-deposited TiO₂ coated collagen samples with increasing number of ALD cycles. This finding also suggests the ability of TiO₂ ALD to uniformly coat three-dimensional (3D) substrates at room temperature.

The significant difference of the growth rate per ALD cycle observed on the planar Si reference substrates from that on nonplanar collagen substrates, even though they were processed at the same conditions, is discussed next. A linear growth at a constant rate of 0.06 nm/cycle was observed on the planar Si reference substrate while much higher growth rate per cycle was observed on nonplanar collagen substrates, which was found to increase with further increasing number of TiO₂ ALD cycles. After 150 ALD cycles, the growth rate on collagen was about 0.05 nm/cycle, while after 300 and 600 cycles, it reached 0.10 and 0.14 nm/cycle, respectively. These deposition rates on nonplanar collagen samples were calculated based on the average outer diameter of single fibers measured from the corresponding surface SEM images. Cross-sectional interface study might be more effective to accurately measure the TiO₂ thickness and the growth rate for each of the coated collagen samples. Similar increased growth on nonplanar three dimensional substrates was reported previously for Al₂O₃ and TiO₂ films. Hyde et al. reported on planar Si substrate that the Al₂O₃ growth rate was 0.2 nm/cycle which increased to 0.3–0.5 nm/cycle on a nonplanar cotton fiber substrate.²⁷ Although in their report, the growth rate on nonplanar substrate was very high initially (0.5 nm/cycle for 50–100 ALD cycles) most likely due to the absorption of water oxidizer, the growth rate decreased (to 0.3 nm/cycle) as ALD proceed further. However, their growth rate was still higher on the nonplanar substrate than that on planar substrate with further ALD cycles. Deposited TiO₂ amount on planar Si substrate was also reported to increase significantly on irregular nonplanar multiwalled carbon nanotubes (MWCNTs) substrate.³⁶ 

There could be several reasons behind this increased growth on nonplanar complex 3D substrates. One reason could be the higher surface area of 3D substrates compared to two-dimensional ones. This exposure of higher effective surface area could lead to larger amount of deposited oxide, although the deposition rate could be the same. Deng et al. reported 183 times higher amount of TiO₂ deposited on irregular MWCNTs due to its 183 times higher surface area compared to the surface area of the planar Si substrate.³⁶ 

A possible reason for higher growth rate on nonplanar substrates could be the complex 3D structure of such porous fiber networks. Due to this 3D pathway, it may take a longer time for reactant precursor molecules and vapor byproducts to diffuse out of these tortuous fiber networks. As a result, excess precursor may remain inside the fiber structure, even after the end of the precursor purging steps and this could cause higher film growth on 3D substrates.²⁷ For optimal growth behavior, longer pulsing and purging of precursor might be an effective approach to avoid this kind of highly nonlinear growth rate in the case of porous 3D fiber substrates allowing longer diffusion time for the reactant molecules and reaction byproducts to get in and out.

GIXRD was performed to investigate the crystalline property of room temperature deposited TiO₂ film on collagen substrates. Samples with the thickest ALD TiO₂ film (i.e., after 600 cycles) were used. Figure 7 shows a GIXRD sample pattern. No distinct crystalline peak was observed, indicating the amorphous nature of the deposited oxide. This was not surprising, since the ALD of TiO₂ was carried out at room temperature, which is well below its reported crystallization temperature (>250 °C).⁴⁵ The broad peak shown in Fig. 7 is mostly from collagen. Lee et al. also reported similar diffraction pattern from collagen substrate, and their TiO₂ film deposited at 70 °C was reported to be amorphous too.²³ 

TiO₂ thin film was deposited on commercially available collagen using TDMAT precursor and ozone oxidizer in a custom ALD system. The deposition was carried out at room temperature, and this ALD process had a linear growth behavior of 0.06 nm/cycle on the planar Si reference sample. XPS data showed the basic elemental composition of collagen protein and it confirmed the presence of TiO₂ on all collagen samples subjected to TiO₂ ALD. GIXRD data indicated that the deposited film was amorphous. SEM images showed a distinct morphological difference among the sample groups. The average fiber diameter of the control sample was found to increase ∼36% after 150 cycles of TiO₂ ALD, and this corresponded to ∼0.05 nm/cycle. The fiber diameter was found to increase more strongly than linearly with further increase of the number of TiO₂ ALD cycles to 300 and then to 600, with growth rates of 0.10 and 0.14 nm/cycle, respectively. The deposition rate on collagen coated with ALD TiO₂ was therefore higher by a factor of 2–3 than that on the corresponding silicon reference substrate, most likely due to its 3D fiber network and diffusion processes in it. This TiO₂ ALD-coated dense collagen could become a novel biomaterial, the mechanical and chemical properties of which could be tuned through ALD so that it could be used in a variety of biomedical applications like bone-grafting, wound-healing, etc. Further studies are under way to investigate the extent of ALD-driven versatile applicability of collagen in material science and biomedical engineering.

This work made use of instruments in the Electron Microscopy Service (Research Resources Center, UIC). The authors are thankful to M. Sardela from Frederick Seitz Materials Research Laboratory Central Facilities, University of Illinois at Urbana-Champaign for his help with the GIXRD experiments. The project was supported by the National Science Foundation CBET No. 1067424 and DMR No. 1307052.