Porous Aluminium Oxide Coating for the Development of Spectroscopic Ellipsometry Based Biosensor: Evaluation of Human Serum Albumin Adsorption

An electrochemically synthesised porous anodic aluminium oxide (pAAO) layer has been analysed by means of spectroscopic ellipsometry. The determined thickness of the formed pAAO layer obtained from spectroscopic ellipsometry measurements and modelling was 322.75 ± 0.12 nm. The radius of the nanopores estimated from SEM images was 39 ± 5 nm and the distance between nanopores was 107 ± 6 nm. The investigation of human serum albumin (HSA) adsorption on the pAAO coating showed that: (i) the protein concentration inside nanopores, depending on exposure time, approximately was from 200 up to 600 times higher than that determined in buffer solution; (ii) the initial phase of the adsorption process is slow (3.23 mg·cm−3·min−1) in comparison with the protein desorption rate (21.2 mg·cm−3·min−1) by means of pAAO layer washing; (iii) conventional washing with PBS solution and deionised water does not completely remove HSA molecules from pAAO pores and, therefore, the HSA concentration inside nanopores after 16 h of washing still remains almost 100 times higher than that present in PBS solution. Thus, due to such binding ability, HSA can be successfully used for the blocking of the remaining free surface, which is applied for the reduction in non-specific binding after the immobilisation of biorecognition molecules on the pAAO surface. It was determined that some desorption of HSA molecules from the pAAO layer occurred during the sensor’s surface washing step; however, HSA concentration inside the nanopores still remained rather high. These results recommend the continued application of pAAO in the development of biosensors.


Introduction
The investigation of protein adsorption/desorption to/from various nanoporous materials is an important task due to the ability to use such materials for proteins separation, purification, and in the development of biosensors, which can be applicable for the determination of various analytes that are important in biology, biochemistry, medical diagnostics, pharmacy, or food industry [1]. The ability of large molecules such as proteins to adsorb within nanopores significantly depends on The porous anodic aluminium oxide (pAAO) layer was formed on high purity (99.999%, 100 mm × 100 mm × 0.32 mm, Good Fellow, Lille, France) aluminium substrate, using a standard two step anodization protocol [20][21][22]. The aluminium substrate was cut into 10 mm × 18 mm size pieces, degreased, and electrochemically polished in an ice cold perchloric acid-ethanol mixture (1 part 60% HClO 4 ; 4 parts 96% C 2 H 5 OH) at 15 V. Then, the electrochemically polished aluminium substrate was anodized for 2 h in 0.3 M oxalic acid electrolyte at a constant 40 V anodization potential. The obtained pAAO layer was removed using a chromic-orthophosphoric acid mixture (4% H 2 CrO 4 ; 11% H 3 PO 4 ) at 70 • C. The removal of the pAAO coating creates a pattern of concaves on the surface of the aluminium substrate. This patterned aluminium substrate was anodized again to create a more uniform pAAO coating. pAAO with~300 nm thickness was synthesised during the 80 s lasting second anodization step under the same conditions as was performed during the first anodization step. After the second anodization step, pores of oxide were cleaned and widened by immersion into 5% orthophosphoric acid solution for 1 min at room temperature, washed with deionised water, and then, dried in the air.
Ellipsometric measurements were performed using a J.A. Woollam M2000X spectroscopic ellipsometer (Lincoln, NE, USA) with a rotating compensator and an angle of incidence was 70 • . All spectra were recorded in a 200-1000 nm wavelength range. Appropriate optical models for characterisation of optical properties of all layers were performed using Complete EASE software (version 5.08gcs) from J.A. Woollam (Lincoln, NE, USA).
In this work, two samples were analysed: pAAO and Al 2 O 3 . The pAAO substrate was used as a sensing surface for non-specific HSA adsorption to nanopores. Ellipsometric data of Al 2 O 3 were simulated using "Complete EASE" software in order to reveal the differences in optical properties between pAAO and Al 2 O 3 . Before incubation of the pAAO substrate in the HSA-containing solution, the optical properties of pAAO in air were determined by spectroscopic ellipsometry. After this, the same substrate was immersed into PBS solution for 5 min, and then, incubated in the HSA-containing solution for 1, 2, and 3 h periods. Immediately after an incubation lasting 3 h, the pAAO coating was removed from the solution and "washed" with PBS for 15 min. After each pAAO immersion in HSA solution and washing step, ellipsometric parameters Ψ(λ) and ∆(λ) were registered. Optical models were constructed in order to calculate the thickness of pAAO and refractive index dispersion for the effective pAAO layers and refractive index inside nanopores.

Results
In order to evaluate the differences of the sensing substrates as Al 2 O 3 and pAAO, scanning electron microscope images and ellipsometric study (measurements and simulations) were performed. SEM images of pAAO were used for the first approximation of the thickness and porosity. Then, ellipsometric measurements of pAAO/air, pAAO/HSA (1 h incub.) , pAAO/HSA (2 h incub.) , pAAO/HSA (3 h incub. + 15 min wash) , and data simulation for Al 2 O 3 and Al 2 O 3 /HSA were performed in order to compare the spectral shift for non-structured (Al 2 O 3 ) and nanostructured (pAAO) substrates, which were used as sensing layers for HSA protein adsorption.

SEM Measurements
The morphology of pAAO was investigated using a scanning electron microscope (SEM). As can be seen from Figure 1A,B, the obtained substrate thickness (including the intermediate Al/Al 2 O 3 layer) was of 339 ± 10 nm. Pore radius (r p ) and distance between pores (D int ) were estimated using SEM micrographs presented in Figure 1A,C, considering that pore diameter represents a larger (less intensive) spot in Figure 1C and the width of the trench visible in the cut-off of the layer (see bottom of Figure 1A). It was determined that r p ≈ 39 ± 5 nm and D int ≈ 107 ± 6 nm. The porosity (P) of the pAAO layer was evaluated using the formula for the hexagonal structure of pores displacement [23].
Coatings 2020, 10, 1018 4 of 10 Using this formula, it was calculated that in this case, P was about 51% ± 10%.

Simulation of Non-Structured Al2O3 Optical Properties
A comparison of the optical properties for the two different samples, pAAO and Al2O3, was performed. Values of ellipsometric parameters vs. wavelength were measured for the pAAO substrate before and after immersion into PBS with 1 mg/mL of HSA in order to determine the wavelength shift of the ellipsometric parameters when a higher amount of proteins adsorbs to the nanopores. For the same reason, simulation of the Al2O3 ellipsometric parameters spectra before and after the formation of the HSA layer was performed. As was noticed in the introduction part, the ellipsometric parameter Δ is more sensitive when it is used for the analysis of non-homogeneous samples. Due to this, in this research, only ellipsometric parameter Δ vs. λ is presented. These data are zoomed to a spectral range from 300 to 400 nm because in this range, the spectral shift was clearly noticeable. The simulated structure consisted of aluminium substrate, an intermix layer of 13.5 nm, and a 322.75 nm Al2O3 layer ( Figure 2, curve 1). Then, an additional HSA layer with the thickness of 8 nm (the average dimension of a HSA molecule [24,25]) was simulated on top of Al2O3 ( Figure 2, curve 2 A2O3/HSA). The simulation of ellipsometric parameter Δ(λ) of the Al2O3 layer and Al2O3/HSA is presented in Figure 2 in order to reveal the effect of the nanopores on proteins adsorption. Thus, such simulated substrate was considered as a closely packed and planar Al2O3 layer. As can be seen from Figure 2, the Al2O3 peak of the ellipsometric parameter Δ is located at λ = 353 nm. In the case of HSA adsorption on the Al2O3 layer, there was a shift of 3 nm towards longer wavelengths and a peak was observed at λ = 356 nm. During the simulations, aluminium and Al2O3 refractive index values were taken from the database of the software. In this case, the HSA layer was simulated as a Cauchy function, which is typically used in ellipsometry for dielectric materials as proteins: 2 4 ( ) ... B C n A = + + + λ λ λ (2) where Cauchy coefficients were applied as A = 1.575; B = 0.01; C = 9.8 × 10 −5 ; n(λ) refractive index dependence on wavelength

Simulation of Non-Structured Al 2 O 3 Optical Properties
A comparison of the optical properties for the two different samples, pAAO and Al 2 O 3 , was performed. Values of ellipsometric parameters vs. wavelength were measured for the pAAO substrate before and after immersion into PBS with 1 mg/mL of HSA in order to determine the wavelength shift of the ellipsometric parameters when a higher amount of proteins adsorbs to the nanopores. For the same reason, simulation of the Al 2 O 3 ellipsometric parameters spectra before and after the formation of the HSA layer was performed. As was noticed in the introduction part, the ellipsometric parameter ∆ is more sensitive when it is used for the analysis of non-homogeneous samples. Due to this, in this research, only ellipsometric parameter ∆ vs. λ is presented. These data are zoomed to a spectral range from 300 to 400 nm because in this range, the spectral shift was clearly noticeable. The simulated structure consisted of aluminium substrate, an intermix layer of 13.5 nm, and a 322.75 nm Al 2 O 3 layer ( Figure 2, curve 1). Then, an additional HSA layer with the thickness of 8 nm (the average dimension of a HSA molecule [24,25]) was simulated on top of Al 2 O 3 ( Figure 2, curve 2 A 2 O 3 /HSA). The simulation of ellipsometric parameter ∆(λ) of the Al 2 O 3 layer and Al 2 O 3 /HSA is presented in Figure 2 in order to reveal the effect of the nanopores on proteins adsorption. Thus, such simulated substrate was considered as a closely packed and planar Al 2 O 3 layer. As can be seen from Figure 2, the Al 2 O 3 peak of the ellipsometric parameter ∆ is located at λ = 353 nm. In the case of HSA adsorption on the Al 2 O 3 layer, there was a shift of 3 nm towards longer wavelengths and a peak was observed at λ = 356 nm. During the simulations, aluminium and Al 2 O 3 refractive index values were taken from the database of the software. In this case, the HSA layer was simulated as a Cauchy function, which is typically used in ellipsometry for dielectric materials as proteins: Coatings 2020, 10, 1018 where Cauchy coefficients were applied as A = 1.575; B = 0.01; C = 9.8 × 10 −5 ; n(λ) refractive index dependence on wavelength.

Optical Properties of pAAO
A schematic representation of pAAO formed on aluminium substrate is presented in Figure 3. pAAO layer thickness calculated using the optical model applied for the analysis of ellipsometric measurements was estimated to be of 322.75 ± 0.12 nm and surface roughness was 39.4 ± 3 nm. Bruggeman effective media approximation (EMA) was used in order to calculate the porosity of pAAO, as was described by other authors [26]. The effective media layer (EMA) of pAAO was modelled as a layer consisting of two materials: Al2O3 and air (52% and 48%, respectively). EMA was used in order to include the porosity of pAAO. Starting values of thickness and surface roughness, which were applied for fitting, were taken from SEM images. During the regression analysis, thickness and surface roughness were treated as free fitting values and the refractive index of Al2O3 as a fixed value. The intermix layer of 13.5 ± 0.46 nm in the optical model was applied to simplify the approximation of the intermediate layer between aluminium and pAAO, and this includes the Al2O3 "barrier layer", which is formed on aluminium substrate. In the optical model, it was estimated that

Optical Properties of pAAO
A schematic representation of pAAO formed on aluminium substrate is presented in Figure 3. pAAO layer thickness calculated using the optical model applied for the analysis of ellipsometric measurements was estimated to be of 322.75 ± 0.12 nm and surface roughness was 39.4 ± 3 nm. Bruggeman effective media approximation (EMA) was used in order to calculate the porosity of pAAO, as was described by other authors [26]. The effective media layer (EMA) of pAAO was modelled as a layer consisting of two materials: Al 2 O 3 and air (52% and 48%, respectively). EMA was used in order to include the porosity of pAAO. Starting values of thickness and surface roughness, which were applied for fitting, were taken from SEM images. During the regression analysis, thickness and surface roughness were treated as free fitting values and the refractive index of Al 2 O 3 as a fixed value.

Optical Properties of pAAO
A schematic representation of pAAO formed on aluminium substrate is presented in Figure 3. pAAO layer thickness calculated using the optical model applied for the analysis of ellipsometric measurements was estimated to be of 322.75 ± 0.12 nm and surface roughness was 39.4 ± 3 nm. Bruggeman effective media approximation (EMA) was used in order to calculate the porosity of pAAO, as was described by other authors [26]. The effective media layer (EMA) of pAAO was modelled as a layer consisting of two materials: Al2O3 and air (52% and 48%, respectively). EMA was used in order to include the porosity of pAAO. Starting values of thickness and surface roughness, which were applied for fitting, were taken from SEM images. During the regression analysis, thickness and surface roughness were treated as free fitting values and the refractive index of Al2O3 as a fixed value.   The intermix layer of 13.5 ± 0.46 nm in the optical model was applied to simplify the approximation of the intermediate layer between aluminium and pAAO, and this includes the Al 2 O 3 "barrier layer", which is formed on aluminium substrate. In the optical model, it was estimated that this "barrier layer" consisted of 50% Al 2 O 3 and 50% aluminium. The presence of this "barrier layer" in the design of the optical model significantly improved the calculation quality and reduced the mean square error (MSE) of the calculated values. MSE in this case was in the range of 31-34. Optical models were designed to calculate the refractive index dispersion of the pAAO sample before and after the incubation of the pAAO substrate in the HSA-containing solution. The same pAAO substrate was investigated in the following way: firstly, the optical properties of the pAAO coating in air were registered. Then, the sample was immersed into 1 mg/mL of HSA-containing solution in PBS, pH 7.4, for 1 h. After that, the spectra of both ellipsometric parameters (∆ and Ψ) were recorded; then, the same sample again was immersed into the HSA solution for 2 h and again, the spectra of ellipsometric parameters were recorded. The same procedure was performed after 3 h immersion followed by 15 min washing in PBS solution. It was determined that when pAAO layer nanopores were filled with air, the determined ratio of pAAO layer composition according to calculations using the EMA-based optical model was 52% pAAO and 48% air.

HSA Adsorption to pAAO
HSA adsorption to pAAO was evaluated by assessment of "∆ spectral shift" and regression analysis, which are presented in Figures 4 and 5. Here, we present only the spectra of ellipsometric parameter ∆ zoomed from λ = 300 to λ = 370 nm, because in this range, a clear shift of the ∆ peak is observed. The determined shift of peaks' wavelength is compared to shifts of the peaks in simulated spectra for the Al 2 O 3 layer, presented in Figure 2. Starting with pAAO/air, the peak of ellipsometric parameter ∆ is observed at λ = 330 nm (Figure 4, curve 1). After 1 h of incubation in solution containing 1 mg/mL of HSA, ellipsometric parameter ∆ shifted towards longer wavelengths, resulting in the appearance of the peak at λ = 335 nm (Figure 4, curve 2); after 2 h of incubation in solution containing 1 mg/mL of HSA, the peak shifted even more significantly and appeared at λ = 341 nm (Figure 4, curve 3). After 3 h incubation in HSA solution and 15 min washing with PBS solution, a shift of ellipsometric parameter ∆ to a shorter wavelength was observed at λ = 336 nm and it indicated that part of the HSA molecules had desorbed from the pAAO surface ( Figure 4, curve 4). Finally, after washing with deionised water lasting 30 min, the peak position shifted even more to shorter wavelengths and appeared at λ = 333 nm ( Figure 4, curve 5).
Coatings 2020, 10, x FOR PEER REVIEW 6 of 10 this "barrier layer" consisted of 50% Al2O3 and 50% aluminium. The presence of this "barrier layer" in the design of the optical model significantly improved the calculation quality and reduced the mean square error (MSE) of the calculated values. MSE in this case was in the range of 31-34. Optical models were designed to calculate the refractive index dispersion of the pAAO sample before and after the incubation of the pAAO substrate in the HSA-containing solution. The same pAAO substrate was investigated in the following way: firstly, the optical properties of the pAAO coating in air were registered. Then, the sample was immersed into 1 mg/mL of HSA-containing solution in PBS, pH 7.4, for 1 h. After that, the spectra of both ellipsometric parameters (Δ and Ψ) were recorded; then, the same sample again was immersed into the HSA solution for 2 h and again, the spectra of ellipsometric parameters were recorded. The same procedure was performed after 3 h immersion followed by 15 min washing in PBS solution. It was determined that when pAAO layer nanopores were filled with air, the determined ratio of pAAO layer composition according to calculations using the EMA-based optical model was 52% pAAO and 48% air.

HSA Adsorption to pAAO
HSA adsorption to pAAO was evaluated by assessment of "Δ spectral shift" and regression analysis, which are presented in Figures 4 and 5. Here, we present only the spectra of ellipsometric parameter Δ zoomed from λ = 300 to λ = 370 nm, because in this range, a clear shift of the Δ peak is observed. The determined shift of peaks' wavelength is compared to shifts of the peaks in simulated spectra for the Al2O3 layer, presented in Figure 2. Starting with pAAO/air, the peak of ellipsometric parameter Δ is observed at λ = 330 nm (Figure 4, curve 1). After 1 h of incubation in solution containing 1 mg/mL of HSA, ellipsometric parameter Δ shifted towards longer wavelengths, resulting in the appearance of the peak at λ = 335 nm (Figure 4, curve 2); after 2 h of incubation in solution containing 1 mg/mL of HSA, the peak shifted even more significantly and appeared at λ = 341 nm ( Figure 4, curve 3). After 3 h incubation in HSA solution and 15 min washing with PBS solution, a shift of ellipsometric parameter Δ to a shorter wavelength was observed at λ = 336 nm and it indicated that part of the HSA molecules had desorbed from the pAAO surface ( Figure 4, curve 4). Finally, after washing with deionised water lasting 30 min, the peak position shifted even more to shorter wavelengths and appeared at λ = 333 nm (Figure 4, curve 5).   for 1 h ( Figure 5B, curve 2), (iii) after the incubation of pAAO in PBS containing HSA for 2 h, (iv) after the incubation of pAAO in PBS containing HSA for 3 h and followed by a 15 min wash with PBS. In order to visualise differences between refractive indexes, the values at λ = 600 nm wavelength, which were taken for calculations, are presented in Table 1. Refractive indexes values at λ = 600 nm wavelength were taken in order to simplify the evaluation; these values are at the end of the abrupt part of dependence of the refractive index vs. wavelength.  Results obtained after the washing of HSA-modified pAAO substrate with PBS and deionised water show that the interaction of HSA protein with pAAO substrate is reversible. Additionally, this process depends on the composition of the selected washing solution. It was determined that pAAO/HSA structure more sensitive to deionized water because more HSA molecules were removed from the substrate by washing using deionized water, than by 0.01 M PBS, pH 7.4. Despite that, the desorption of HSA proteins from the surface of pAAO takes a long time due to the inclusion of proteins within nanopores in the Al 2 O 3 layer.
In order to assess changes in refractive index for the measured sample when using the EMA-based optical model for the pAAO layer, regression analysis was performed after all modification steps, namely after the formation of pAAO/air, pAAO/HSA (1 h incub.) , pAAO/HSA (2 h incub.) , and pAAO/HSA (3 h incub. + 15 min wash) structures. During the regression analysis, such parameters as thickness, roughness, and intermix layer thickness, which were applied in the EMA-based optical layer, were fixed values, which were calculated by modelling of the pAAO/air structure. Cauchy function parameters A, B, and the ratio of HSA and Al 2 O 3 in the EMA-based optical model were applied as free fitting parameters. Regression analysis resulted in a dispersion of refractive index, which is presented in Figure 5A,B for all evaluated structures: pAAO/air, pAAO/HSA (1 h incub.) , pAAO/HSA (2 h incub.) , pAAO/HSA (3 h incub. + 15 min wash) , and pAAO/PBS. Figure 5A shows the dispersion of effective refractive index for the pAAO layer, when this layer was modelled using the EMA-based optical model for the structures mentioned here: (i) pAAO in PBS while estimating that pAAO pores are filled with PBS ( Figure 5A Figure 5B shows refractive index dispersion inside nanopores of pAAO when pores were filled with: (i) PBS ( Figure 5B, curve 1), (ii) HSA after the incubation of pAAO in PBS containing HSA for 1 h ( Figure 5B, curve 2), (iii) after the incubation of pAAO in PBS containing HSA for 2 h, (iv) after the incubation of pAAO in PBS containing HSA for 3 h and followed by a 15 min wash with PBS. In order to visualise differences between refractive indexes, the values at λ = 600 nm wavelength, which were taken for calculations, are presented in Table 1. Refractive indexes values at λ = 600 nm wavelength were taken in order to simplify the evaluation; these values are at the end of the abrupt part of dependence of the refractive index vs. wavelength. In order to evaluate the HSA concentration inside the pores (C HSA ), we used the well-known fact that the refractive index of protein in water solution is the linear function of protein concentration. For this reason, the C HSA can be calculated by means of the formula obtained after a simple transformation of mathematical expression used for refractive index calculation at a known concentration [27]: Here, n PBS is a refractive index of PBS, which is equal to 1.333; I n is refractive index increment (the derivative dn/dc) according to the protein's concentration [28]. For the calculation of C HSA , we used I n = 0.18 mL/g that was determined at λ= 632.2 nm.
The results presented in Table 1 show the following important features of HSA adsorption on/within pAAO coating: (a) the protein concentration inside nanopores, depending on exposure time, is approximately 200-600 times higher than that in buffer solution; (b) the beginning of the absorption process is slow (3.23 mg·cm −3 ·min −1 ) in comparison with the desorption of proteins (21.2 mg·cm −3 ·min −1 ) during the washing of the pAAO/HSA structure; (c) the conventional washing procedure does not enable the desorption of all adsorbed HSA molecules.

Conclusions
A standard anodization protocol was successfully applied for the formation of pAAO coating of 322.75 ± 0.12 nm thickness. SEM images illustrate that the pores of pAAO are displaced in a regular hexagonal structure. The radius of the nanopores estimated using SEM micrographs was 39 ± 5 nm and the distance between nanopores was 107 ± 6 nm. These dimensions fit well the size of the HSA proteins molecules, as the average diameter of HSA is in the range of 8 nm. In this research, the evaluation of non-specific and reversible adsorption of HSA onto the pAAO layer was performed. Ellipsometric measurements showed that a part of HSA remained inside of the pAAO pores after washing with PBS solution and deionised water. Therefore, during the application of pAAO-based optical sensors, it should be taken into account that some of the HSA, which is rather fast adsorbed on/within pAAO, might desorb during washing of the sensor. However, it was determined that conventional washing with PBS solution does not fully desorb all the HSA molecules from pAAO pores and even after washing with deionised water lasting 16 h, HSA concentration inside nanopores still remains approximately 100 times higher than that in PBS solution. Hence, pores of pAAO operate as "traps" for HSA proteins, cumulating molecules of this material. However, HSA should be carefully used for the blocking of the remaining free surface of signal transducers based on pAAO when significant reduction in non-specific binding is expected.