Improvement of Mechanical Characteristics on Ultra-High Molecular Weight Polyethylene Surface through Zinc Oxide Atomic Layer Deposition Film

,


Introduction
Polymer materials have been widely used in biomedical applications owing to their light weight, structural rigidity, conformability, electromagnetic inactivity, biochemical stability, and biocompatibility [1,2]. However, the polymer material applications are limited by minimal loading conditions because the mechanical strength of polymers is generally inferior to that of common metal-based materials [3][4][5]. A joint prosthesis is an application that requires mechanical properties to resist body weight and abrasive wears on articulating surfaces. Although ceramic materials are ideal for minimizing wear damage, the brittle and nonconformable characteristics of ceramics often lead to fractures in joint implants [6,7]. Tus, modern joint prostheses are designed to form an articulating surface between metal (cobalt-chromium alloy) and ultra-high molecular weight polyethylene (UHMWPE) compartments. Te high crack resistance and toughness of UHMWPE [8][9][10] are useful in preventing highly concentrated contact stresses on articulated surfaces. However, wear failures in relatively weak UHMWPE compartments are known to limit the life of a joint implant [11]. Gamma irradiation and vitamin E stabilization techniques are widely used to improve the wear characteristics of UHMWPE by increasing the cross-linking between molecular chains [12][13][14]. However, the life of joint prostheses is not sufcient for clinical implementation.
Atomic layer deposition (ALD) is a popular method to create mechanically strong ceramic flms on the surface of various materials at relatively low temperatures [15,16]. Because conventional surface coating methods such as physical vapor deposition (PVD) and chemical vapor deposition (CVD) require high-temperature conditions that are generally much higher than the melting temperature of polymers [17][18][19], the ALD process can be considered as a potential solution to form wear-resistant ceramic layers on the surface of UHMWPE structures without causing thermal damage to native polymer chains [20][21][22]. Previous studies have reported the mechanical efects of ALD ceramic flms on the surface of various materials. Aluminum, titanium, and zinc oxide (ZnO) layers have been shown to signifcantly increase the hardness and elastic modulus of wafer and glass surfaces [23][24][25][26]. Aluminum oxide flms on polyethylene naphthalate have been examined for the improvement of critical strain energy in crack propagation [27]. ZnO flms have been reported to decrease the friction coefcient and increase the wear resistance of stainless steels [28].
Adhesion failures between ALD flms and underlying materials are a common problem when there are large diferences in the thermal expansion coefcients of each material [29]. Diferent thermal contraction ratios in the two materials may create separations between the ALD flms and the underlying materials, which may result in multiple cracks on the surface of the ALD flms [30,31]. Recent studies introduced mixtures of ALD flms and organic material layers in molecular units through a molecular layer deposition (MLD) process to minimize adhesion problems owing to mismatches in the thermal expansion coefcients [31]. Composing organic MLD layers using trimethylaluminum with ethylene glycol has shown signifcant reductions in the surface damage of aluminum oxide ALD flms on Tefon-fuorinated ethylene propylene [31]. Because diferences in the thermal expansion coefcients of ceramics (ZnO: 1.57 × 10 −5 K −1 , aluminum oxide: 1.03 × 10 −5 K −1 ) and UHMWPE (18 × 10 −5 K −1 ) are distinct, MLD organic layers might be a good solution to reduce surface failures on ALD ceramic flms.
Although previous ALD studies have primarily focused on microelectronics applications, our goal is to improve the mechanical properties of polymer surfaces by creating ceramic flms on the surface of polymer structures. In this study, we aimed to examine the improvements in the mechanical properties of UHMWPE surfaces after applying ZnO and organic material hybrid flm. We decided to use ZnO as the material for ceramic flms because ZnO is known to efectively reduce the coefcient of friction of articular surfaces [32][33][34]. ZnO nanoparticles have been widely used for the delivery of biomolecules [35][36][37], treatments of various diseases [38,39], and tissue engineering applications [40,41]. Biocompatibility and toxicity of ZnO-related materials are already well proven [42,43]. We tested (1) the efect of MLD organic layers on minimizing surface failures and (2) the mechanical properties of ZnO hybrid flms with diferent ratios of ZnO and organic bufer layers. Finally, we aimed to estimate whether ALD ceramic flms could be a potential solution for improving the wear characteristics of polymer compartments in joint prostheses.

Sample Preparation.
We prepared 15 × 15 × 5 mm 3 UHMWPE (Mitsubishi Chemical group, Tokyo, Japan) blocks (Figure 1(a)). Te surface roughness of all the samples was controlled by a fve-step sanding process with sandpapers of #200, #400, #800, #1600, and #2400 grits in a sequential order. Final polishing was performed using a polishing cloth with a 1 μm diamond suspension solution. Te polished specimens were cleaned with isopropyl alcohol and distilled water for 10 min in an ultrasonic washing machine (JAC Ultrasonic KODO, Yong In, Korea). We then applied O 2 plasma treatment (Femto Science, Hwaseong, Korea) to the surface of the UHMWPE blocks for 60 s with a discharging power of 100 W to improve chemical reactions of UHMWPE with coating materials.

ALD Window for ZnO Layers on UHMWPE.
Te deposition rate of the ZnO layer on the UHMWPE blocks was measured at diferent working temperatures between 60°C and 120°C, below the melting temperature of UHMWPE (130°C). A total of 450 cycles of ZnO deposition were applied to all the samples, and the thickness of the ZnO layers was measured.

ALD and MLD Processes.
A custom holder for UHMWPE specimens was developed for the efective and uniform heating of the polymer samples. Two metal bars were placed in contact with both sides of the polymer block to transmit heat energy through the side and bottom surfaces of the sample (Figure 1(b)). All samples were preheated for an hour at 100°C, as determined from the ALD window examination results. First, we created UHMWPE samples with pure ZnO flms. Te precursor for the ZnO layer was diethylzinc (DEZ, EG Chem, Gongju, Korea). Te ZnO ALD process consisted of alternating pulses of DEZ and distilled water (H 2 O). Nitrogen gas (N 2 , 99.99%) was used to purge the gases remaining inside the chamber between the DEZ and H 2 O pulses. Tus, a complete ZnO deposition cycle consisted of sequential pulsing of DEZ, N 2 , H 2 O, and N 2 for 0.5, 30, 0.5, and 30 seconds, respectively. Te ZnO ALD deposition cycle was repeated 1200 times. Te chemical reactions in the ZnO layers are expressed in the following formulas (the asterisks denote the surface groups). (1) We also fabricated UHMWPE samples with hybrid flms of ZnO and organic layers. Te deposition process of the organic layer was similar to that of pure ZnO flms, but H 2 O was replaced with hydroquinone (HQ, Sigma-Aldrich Chemistry, St. Louis, USA). A complete organic layer deposition cycle consisted of sequential pulsing of DEZ, N 2 , HQ, and N 2 for 0.5, 30, 0.5, and 60 seconds, respectively. Te chemical reactions in the organic layers are expressed in the following formula. 2 Advances in Materials Science and Engineering We controlled the ratio of the ZnO and organic layers by changing the number of cycles in each layer ( Figure 2). A single layer of ZnO or organic material was designed to be formed with a thickness of 0.3 nm. Five sample groups with diferent ZnO to organic layer ratios were prepared (ZnO to organic layer ratios of 1 : 1, 2 : 1, 3 : 1, 4 : 1, and 5 : 1), and the entire deposition cycle was maintained for 1200 cycles for all specimens. At the end of the deposition process, the deposition chamber was gradually cooled to room temperature for 180 min to minimize the thermal stresses in the samples.

Coating Characterization.
Surface images of the ALDcoated UHMWPE samples were acquired using a feldemission scanning electron microscope (FE-SEM; FEI Quanta 250 FEG, Philips, Amsterdam, Netherlands). Te thicknesses of the ALD flm layers were measured using focused ion beam scanning electron microscopy (FIB-SEM; LYRA3 XMH, TESCAN, Brno, Czech Republic). Te chemical characteristics of the ALD flm layers were analyzed using grazing incidence X-ray difraction (GIXRD; Smart Lab, Rigaku, Tokyo, Japan) with an X-ray power of 9 kW and a wavelength of 0.15 nm. Fourier transform infrared spectroscopy (FTIR; Cary 630, Agilent, Santa Clara, CA, USA) was used to measure the spectrum of the molecular structures in the ALD flms (frequency range: 650-4000 cm −1 , frequency interval: 2 cm −1 ). Te hardness and elastic modulus of an ALD-coated UHMWPE surface were measured by nano-indentation examinations (UNHT 3 , Anton Parr, Graz, Austria) at 15 randomly selected locations in a sample. Te indentation depth was determined to be 25% of the entire ALD flm thickness for each specimen.

Scratch Test of ZnO Films. Scratch tests of ZnO flms
were performed by using a nano-scratch device (UNHT3, Anton Parr, Graz, Austria) to examine adhesion characteristics of pure ZnO and hybrid ZnO (ZnO:organic layer � 1 : 1) flms on the surface of UHMWPE. A normal compressive load was gradually increased from 0 to 600 mN through a scratch length of 3 mm with a scratch speed of 6 mm/min. A Rockwell diamond tip with a tip radius of 50 μm was used to provide a maximum contact pressure of approximately 64 MPa (Hertzian contact theory) which was comparable to the contact pressure of 20-30 MPa in total knee arthroplasty [44]. Energy dispersive spectrometer (EDX) technique was used to measure atomic ratios of ZnO flms in and outside of the scratch tracks. Locations of EDX measurements were indicated by star ( * ) symbols in Figure 3.

ALD Window.
Te growth rate of the ZnO layers on the UHMWPE surface was observed at diferent temperatures ( Figure 4). Te growth rate gradually increased and stabilized at a chamber temperature of 100°C. Tus, we decided to use 100°C as the ALD processing temperature for subsequent experiments.

Chemical
Analysis. GIXRD measurements of the samples with pure ZnO flms showed three distinct peaks at 31.8°, 34.5°, and 36.3°, which are indicators of the wurtzite crystal structures representing the degree of crystallinity of the ZnO layers [45][46][47]. Tese three peaks gradually decreased as the ratio of ZnO to the organic layer decreased ( Figure 5(a)). A (020) peak around 30°, which is one of typical difraction signals of UHMWPE, is also shown [48,49]. C-H, C-O, and C � C aromatic stretching modes in organic materials, respectively [50][51][52]. Tese peaks increased as the ratio of ZnO to the organic layer decreased (Figure 5(b)). Additional distinct peaks around 750 cm −1 and 1450 cm −1 indicate CH 2 rocking of methyl and CH bending of methylene groups in UHMWPE [53][54][55].

Surface Images of UHMWPE with Various ZnO Films.
Te FE-SEM images of the UHMWPE samples with diferent ALD flms are shown in Figure 6. Roughness of UHMWPE surfaces seemed to change with hybrid flms of ZnO and organic layers. It might be that ZnO grains and organic HQ-DEZ bonding structures increased irregularity of surface profles. However, the size and distribution of the ZnO grains were uniform throughout all the samples. However, severe cracks on the surface of the ZnO flms were distinct in the samples with pure ZnO flms ( Figure 6(g), ZnO 100%). Although surface cracks were found in the samples with a ZnO to organic layer ratio of 5 : 1 (Figure 6(f ), ZnO:organic layer � 5 : 1), no cracks were found in hybrid flm samples with other ratios.

Mechanical Properties.
Te thickness of the ALD flms was approximately 200 nm (Table 1). Te hardness and elastic modulus of the UHMWPE samples with pure ZnO flms signifcantly increased compared to those of the native UHMWPE blocks (Tukey's HSD post hoc signifcant difference test, p < 0.01, Figures 7(a) and 7(b)). Te hardness and elastic modulus of the UHMWPE specimens with hybrid flms were signifcantly greater than those of the native UHMWPE (Tukey's HSD test, p < 0.01), but signifcantly smaller than the values in the samples with pure ZnO flms (Tukey's HSD test, p < 0.01). Diferences in hardness and elastic modulus values between hybrid flms with diferent ZnO to organic layer ratios were not signifcant (Tukey's HSD test, p > 0.05).    Advances in Materials Science and Engineering Te hardness to elastic modulus ratios (H/E ratio), indicating the wear resistance of a solid material [56,57], were calculated for diferent flms (Figure 7(c)). Te H/E ratios increased for all types of ALD flms compared to the values of native UHMWPE (Tukey's HSD test, p < 0.01). Although pure ZnO flms showed higher H/E ratios than the hybrid flms, the diferences were not signifcant except for 4 : 1 hybrid flm sample (Tukey's HSD test, p > 0.05).

Scratch
Characteristics. SEM images of the scratches on UHMWPE surfaces with pure ZnO and hybrid ZnO flms are shown in Figure 3. Changes in indentation loads and displacements during scratch tests are also shown for both pure and hybrid ZnO flms ( Figure 3). Multiple cracks were prominent from the early stage of the scratch, and the crack patterns became severer as the compressive load increased on the pure ZnO flm (Figure 3(a)). However, no surface cracks were found throughout the entire scratch track on the hybrid ZnO flm (Figure 3(b)). We were unable to measure distinct critical loads for failures in both types of flms. EDX measurements revealed that atomic ratio of Zn was reduced by approximately 20% in the scratch track for pure ZnO flms. However, the diference in atomic ratios of Zn in and outside of the scratch track was minimal for hybrid ZnO flms ( Table 2).

Discussion
We successfully produced ceramic flms with ZnO on the surface of UHMWPE using ALD, which is suitable for polymers because of its relatively low processing temperature. ZnO ceramic layers were shown to improve the mechanical and wear characteristics of the polymer surfaces. Organic layers were applied between each ZnO layer to minimize surface failures of ceramic flms, which might be caused by mismatches in thermal expansion ratios of different materials. Te sandwich-type hybrid ceramic flms of ZnO and organic layers (Figure 2) signifcantly increased the hardness and elastic modulus of the native UHMWPE samples. Although the mechanical properties of the hybrid flms were inferior to those of the pure ZnO flms, the hybrid flms resulted in markedly reduced crack failures on the surface of the ceramic flms. Moreover, the H/E ratio as an indicator of wear resistance in a solid material [58,59] was signifcantly improved in the samples with hybrid flms of ZnO and organic layers. Te H/E ratios of the hybrid flms were comparable to those of pure ZnO flms (Figure 7(c)).
Te formation of ZnO flm layers on UHMWPE surfaces was tested using GIXRD to measure the degree of crystallinity of ZnO. Distinct wurtzite crystal peaks at 31.8°, 34.4°, and 36.3°in the GIXRD measurements of the pure ZnO flms indicate the formation of ZnO layers [60,61]. Tese peaks were undetectable (31.8°and 34.4°) or decreased (36.3°) in the flms with organic layers. Because only a few cycles of ZnO (one to fve ZnO cycles in the flms with diferent ZnO to organic layer ratios) existed between each organic layer in the hybrid flms, the difraction signals of the ZnO crystal peaks may be weaker than those for the pure ZnO thick flms ( Figure 5(a)). Te formation of organic layers was also examined by FTIR by measuring the characteristic frequencies (788 cm −1 , 827 cm −1 , 1204 cm −1 , and 1492 cm −1 ) of HQ rings in the hybrid flms [50,62]. Strong peaks at these frequencies in all the hybrid flms suggest the stable formation of phenylene HQ rings [50,62]. Increases in the signals for HQ frequencies for the flms with greater bufer layer proportions also indicated the successful control of the bufer layers in our ALD design.
ZnO and organic layer hybrid flms exhibited signifcant reductions in surface cracks, although severe surface cracks were found in all tested samples with pure ZnO layers. Surface failures on pure ZnO flms might be the result of Advances in Materials Science and Engineering  Advances in Materials Science and Engineering diferences in the thermal expansion coefcient between ZnO (1.57 × 10 −5 K −1 ) and UHMWPE (18 × 10 −5 K −1 ) [63]. Because the thermal expansion coefcient of UHMWPE is approximately 11 times greater than that of ZnO, thermal contractions in the UHMWPE surface must be greater than those in ZnO flms during the cooling stage of the ALD process. Mismatches in the thermal contraction between the two materials may create compressive stress on the ZnO flms and result in surface cracks [30,31]. Continuous uplift patterns surrounding the surface cracks provided strong evidence for the formation of compressive stresses in the pure ZnO flms (Figure 6(g)). A mixture of organic layers in every few cycles of ZnO flms may act as cushions to release compressive stresses and reduce surface failures. However, these cushion efects of the organic layers seemed to decrease as the proportion of ZnO layers increased because surface cracks started to appear in the samples with a ZnO to organic layer ratio of 5 : 1 (Figure 6(f )). Further studies are needed to determine the optimum ratio between ZnO and the organic layers for protection against surface failure. Te mechanical properties of both the pure and hybrid ZnO flms were signifcantly better than those of the native UHMWPE samples. ZnO ceramic flms on the surface of    (Figures 7(a) and 7(b)). Reductions in the mechanical properties of the hybrid flms might be inevitable because the proportion of ZnO layers is reduced [64]. Although the hardness and elastic modulus of the hybrid flms decreased, the H/E ratios remained similar in both the pure and hybrid ZnO flms (Figure 7(c)). Because the wear characteristics on a solid surface are known to be described by H/E ratios, these results indicate similar wear resistances in both flms. ZnO and organic material hybrid flms exhibited markedly improved scratch characteristics. EDX measurements in the scratch tracks showed reductions in atomic ratio of Zn for pure ZnO flms, suggesting delamination of ZnO flm under a physiologic level of contact pressure (64 MPa) while the diferences in Zn atomic ratios in and outside of the scratch were minimal for hybrid ZnO flms. Numerous cracks inside the scratch on pure ZnO flms might be the result of brittle failures from a large magnitude of vertical compression and deformation. However, organic MLD layers in the hybrid ZnO flms may increase the ductility of thin flms and resulted in reduced crack failures inside the scratch [65,66]. In this study, we were unable to measure distinct critical loads for flm failures. It might be the result of thick specimens (approximately 7 mm thick) of soft UHMWPE material to simulate the contact situation in a real joint prosthesis. Because vertical deformation in a thick material would be larger than the deformation in a thin flm, our ZnO flms may also deform through the profle of vertical deformation in UHMWPE blocks prior to a complete delamination of the flms [67,68]. Our scratch test confguration was diferent from previous studies on thin flm-type substrates which resulted in small vertical deformations and distinct critical loads for failures [69]. However, our examination on thick UHMWPE blocks might be a better representation of realistic contact mechanics in physiologic environments for joint prostheses.
Tis study has some limitations that need to be addressed. First, we applied only one cycle of the organic layer between ZnO flms, although the deposition cycle of ZnO varied from one to fve cycles in diferent types of hybrid flms. Our results suggest that organic layers reduce the thermal stresses on ZnO layers and minimize surface cracks. Tus, multiple cycles of organic layers may result in better resistance to surface cracks in ceramic flms. However, we did not examine multiple cycles of organic layers because the goal of this study was to establish the relationship between the proportion of organic layers and the reduction in surface cracks within a ceramic flm. Further studies on hybrid flms with multicycle organic layers are important to understand the optimal deposition conditions for various ceramic flms. Second, the wear resistance of the ZnO ceramic flms was estimated from the H/E ratios from the nano-indentation examination results. Te wear characteristics of a solid material are generally examined by measuring the volume or weight loss in the sample after ball-on-disk-or pin-on-disk-type wear tests [70][71][72]. Unfortunately, measurement of volume or weight losses in nanoscale flms was not possible. Tus, we described the changes in wear characteristics of ALD flms using H/E ratios, which have been widely used as an indicator of wear resistance [29]. Although future ball-ondisk or pin-on-disk wear studies on thicker flms would reveal the improvement in wear characteristics by measuring volume or weight losses in the flms, we believe that the current results of H/E ratios may still be useful for estimating changes in wear characteristics after ALD ceramic flm coating on polymer materials.

Conclusion
In this study, we created ALD ZnO flms on the surface of UHMWPE to improve the mechanical and wear characteristics of UHMWPE samples, and organic layers were mixed with ZnO flms to reduce surface damage. Te organic layers exhibited excellent reductions in the surface cracks in the ALD ZnO flms and signifcantly increased the hardness and elastic modulus of the UHMWPE surfaces. Te wear resistance of the hybrid flms with ZnO and organic layers was also signifcantly increased and was comparable to the value measured for pure ZnO flms. Because biomedical applications of polymers have been limited in nonweightbearing situations due to their poor wear characteristics, our ALD ceramic coating approach should be a method to improve the mechanical characteristics on the surface of polymer materials without causing chemical and thermal damage to native polymer structures. Although further research on macroscale mechanical examinations needs to be conducted, several promising results from this study suggest that the ALD ceramic layer with organic layer on polymer surfaces may become a potential solution for realizing a wear-resistant protection coating for the polymer compartment of joint prostheses.

Data Availability
Te data that support the fndings of this study are available from the corresponding author upon reasonable request.

Conflicts of Interest
Te authors declare that they have no conficts of interest.