Fabrication of ordered metallic glass nanotube arrays for label-free biosensing with diffractive reflectance

Author’s Accepted Manuscript Fabrication of Ordered Metallic Glass Nanotube Arrays for Label-free Biosensing with Diffractive Reflectance Wei-Ting Chen, May-Show Chen, Jinn P. Chu, Chung-Kwei Lin, Jem-Kun Chen www.elsevier.com/locate/bios

PII: DOI: Reference:

S0956-5663(17)30682-6 https://doi.org/10.1016/j.bios.2017.10.023 BIOS10048

To appear in: Biosensors and Bioelectronic Received date: 21 July 2017 Revised date: 29 September 2017 Accepted date: 12 October 2017 Cite this article as: Wei-Ting Chen, May-Show Chen, Jinn P. Chu, Chung-Kwei Lin and Jem-Kun Chen, Fabrication of Ordered Metallic Glass Nanotube Arrays for Label-free Biosensing with Diffractive Reflectance, Biosensors and Bioelectronic, https://doi.org/10.1016/j.bios.2017.10.023 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Fabrication of Ordered Metallic Glass Nanotube Arrays for Label-free Biosensing with Diffractive Reflectance

Wei-Ting Chena, May-Show Chenc, d, Jinn P. Chua, *, Chung-Kwei Lin b, e, *, Jem-Kun Chen a,*

a

Department of Materials Science and Engineering, Chemical Engineering, National Taiwan University of Science and Technology, 43, Sec. 4, Keelung Road, Taipei, 106, Taiwan, ROC b Research Center of Digital Oral Science and Technology, College of Oral Medicine, Taipei Medical University, Taipei 11031, Taiwan c School of Oral Hygiene, College of Oral Medicine, Taipei Medical University, Taipei 11031, Taiwan d Department of Dentistry, Taipei Medical University Hospital, Taipei 11031, Taiwan e School of Dental Technology, College of Oral Medicine, Taipei Medical University Taipei 11031, Taiwan

[email protected] [email protected] [email protected]

*

Corresponding Author: Chung Kwei Lin. E-mail address:

*

Corresponding Author: Jinn P. Chu. E-mail address:

*

Corresponding Author: Jem-Kun Chen E-mail address:

1

Tel: +8862-2736-1661, ext.5115

Tel: +886-2-2730-3292, Fax: +886-2-27376544 Tel: +886-2-2737-6523, Fax: +886-2-27376544

Abstract In this study, a photoresist template with well-defined contact hole array was fabricated, to which radio frequency magnetron sputtering process was then applied to deposit an alloyed Zr55Cu30Al10Ni5 target, and finally resulted in ordered metallic glass nanotube (MGNT) arrays after removal of the photoresist template. The thickness of the MGNT walls increased from 98 to 126 nm upon increasing the deposition time from 225 to 675 s. The wall thickness of the MGNT arrays also increased while the dimensions of MGNT reduced under the same deposition condition. The MGNT could be filled with biomacromolecules to change the effective refractive index. The air fraction of the medium layer were evaluated through static water contact angle measurements and, thereby, the effective refractive indices the transverse magnetic (TM) and transverse electric (TE) polarized modes were calculated. A standard biotin–streptavidin affinity model was tested using the MGNT arrays and the fundamental response of the system was investigated. Results show that filling the MGNT with streptavidin altered the effective refractive index of the layer, the angle of reflectance and color changes identified by an L*a*b* color space and color circle on an a*b* chromaticity diagram. The limit of detection (LOD) of the MGNT arrays for detection of streptavidin was estimated as 25 nM, with a detection time of 10 min. Thus, the MGNT arrays may be used as a versatile platform for high-sensitive label-free optical biosensing.

Keywords: metallic glass; nanotube; ordered structure; grating

2

1. Introduction Metallic glasses have attracted attention for decades because of their outstanding properties, including high strength and hardness, a high elastic strain limit, and good corrosion resistance (Lee et al., 2011; Chu et al., 2012; Hasan et al., 2016). Without any influence from crystalline facets or grain boundaries, metallic glasses are promising materials for applications in optical reflection. The absence of long-range atomic order and the lack of crystalline defects (e.g., dislocations and grain boundaries) make the behavior and properties of metallic glasses notably different from crystalline alloys. Previous efforts in the development of metallic glasses and amorphous alloys have focused mostly on bulk metallic glasses (BMGs); however, they were limited by brittleness. Recently, thin film metallic glasses (TFMGs) have become noticeable materials in fields related to micro-electro-mechanical systems (MEMS) (Chu et al., 2012) as well as in optical (Lee et al., 2010) and biomedical devices. Well-ordered periodic nanostructures are being developed for a growing number of diverse applications in modern optical communication and sensor technology including pH sensors (Chen et al., 2011; Chen et al., 2012), volatile organic compound sensors (Chen et al., 2013), DNA detectors (Chen et al., 2014), and immunosorbent assays (Chen et al., 2013). Surface patterning can be an important mean of controlling the response of materials to incident light (Muskens et al., 2008; Zhu et al., 2009); especially, the properties of nanostructured surfaces (e.g., reflectance and absorption of light) have been studied extensively (Ray et al., 1980; Hauser et al., 1979).Well-ordered periodic nanostructures over a large substrate area are often prepared through bottom-up microfabrication processes, including advanced lithography and surface modification (Chen et al., 2014; Saotome et al., 2007). The ability to controllably pattern metallic glasses through simple molding offers the opportunity

to

realize

textured

metal

surfaces

displaying

tunable

reflectance

and

absorption—potentially within desirable spectral ranges for specific applications. Deposition process can be an inexpensive, simple, and rapid fabrication route for preparing nanostructures over

3

large areas and with high aspect ratios (Grünzweig et al., 2008; Kim et al., 2013). Previous studies have demonstrated that bulk metallic glasses were fabricated into various three-dimensional microscale structures (Schroers et al., 2010; Lee et al., 2015; Yashiro et al., 2014). Optical sensing systems are required to be ideally operated under visible lights; thus, highly stable nanoarrays should be composed of relatively low refractive index materials, thereby enabling efficient overlap of the waveguide mode with its sensing zone aligned to the nanostructure (Skivesen et al., 2004). The refractive index of a well-defined nanotube array is significantly determined by the media filling within the nanotubes. Thus, a well-defined nanotube array filling of various molecules could result in various optical properties, appropriate for applications in optical sensing. Nevertheless, it has remained a challenging task to fabricate well-defined nanotube array possessing particular optical properties as an optical sensor, despite many efforts. In this study, very-large-scale integration (VLSI) process was used to synthesize well-defined patterns of contact hole arrays on Si wafers (Chen et al., 2012; Zeng et al., 2017; Huang et al., 2013). A layer of Zr-BMG (Zr55Cu30Al10Ni5) was then deposited onto the photoresist-patterned Si surface. Removing the photoresist with a solvent resulted in nanotube arrays of Zr-metallic glass. The development of manufacturing methods allows the creation of regular nanotube structures based on metallic glass. (Chen et al., in revision) Arrays of vertically aligned metallic glass nanotubes (MGNTs) can possess exceptional optical properties that are strongly dependent on the physical and geometrical parameters. The well-defined MGNT array could be fabricated as a specific biosensor due to its high strength and hardness, elastic strain limit, and good corrosion resistance. As our knowledge, the unique nanotube arrays could not be achieved with other materials. Filling with various media results the change of MGNT’s refractive index as well as other optical properties; therefore, the well-ordered MGNT array has the advantage to serve as a label-free optical biosensor. Here, the focus is on specular reflectance measurements, investigating the polarization states of both the incoming and reflected light. The integrated diffuse reflectance at various incident angles and

4

various wavelengths were combined to characterize the optical properties of textured MGNT surfaces. Additionally, this demonstrates that diffuse reflectance surfaces displaying low diffuse reflectance and wavelength-dependence can be realized through control of their topography, which is promising in terms of both the manufacturing and applications of metallic glasses. 2. Experimental Section 2.1 Materials 6” one-side-polished Si(100) wafers were used the substrates and supplied by Hitachi (Japan). To remove dust particles and organic contaminants, the Si surfaces were ultrasonically rinsed sequentially with MeOH, acetone, and CH2Cl2 for 10 min each and subsequently dried under vacuum (Chen et al., 2010). Biotin (C10H16N2O3S; Molecular Weight: 244.31 g/mol, ≥99%), lyophilized powder, and Streptavidin (lyophilized from 10 mM potassium phosphate, ≥13 U/mg protein) were purchased from Sigma. AffiniPure goat anti-rabbit IgG (AGRI) was purchased from Jackson ImmunoResearch Lab. A buffer solution (10 mM phosphate buffer, 100 mM NaCl,

pH

7.4) was used to prepare the following solution: 0.5 mM of Sulfo-NHS-LC-LC-biotin solution, and a series of concentrations of streptavidin. Milli-Qwater (18.2MΩ cm, Millipore Corp., Bedford, MA) was used throughout the experiments. All other chemicals and solvents were of reagent grade and purchased from Aldrich Chemical. 2.2 Zr-Based MGNT Arrays Figure 1 outlines the strategy for the fabrication of MGNT arrays using a VLSI process. A: The photoresist was spun on the HMDS-treated Si wafer at a thickness of approximately 780 nm; contact hole arrays of various dimensions were successively patterned on the photoresist through a lithography process. B: The as-prepared photoresist templates on the Si substrates were subjected to deposition through sputtering of an alloyed Zr55Cu30Al10Ni5 target, using a radio frequency magnetron sputtering system, without intentional substrate heating for 225, 450, and 675 s; the operating conditions for the sputtering system were set at base and working pressures of

5

approximately 5  10–4 and 3 mTorr, respectively, with a working distance of 10 cm under argon. C: The remaining photoresist was further removed from the silicon surface by rinsing with toluene for 30 s, leaving behind the MGNT array brush surface. For comparisons, a Si substrate prepared without a photoresist template was also subjected to deposition by the same target for 450 s, providing the structure denoted herein as TFMG. D: In order to generate surface silane groups for the

Sulfo-NHS-LC-LC-biotin

immobilization,

an

activation

step

is

usually

necessary

(Kambhampati et al., 2001). The substrate with MGNT arrays was first immersed in a mixture of 0.1M KH2PO4/0.1M K2HPO4 solution for 30 min and kept in the oven at 60 °C for 2h, followed by modification using 3-aminopropyltriethoxysilane (APTES) by soaking in a 5 wt% APTES solution (in dry methanol) for 24 h. Then the substrate was washed with methanol and dried in a stream of N2. For covalent linking of the silane groups to the surface of silica, the substrate was heated at 120 °C for 2h. After this step, the substrates were ready for the step of the biotin anchoring. Commercially available recombinant biotin that minimizes nonspecific interactions with serum/cell proteins was covalently anchored on the nanotube surface by conjugation to the amine groups. The amine groups on the surface of the nanotube were activated by adding a solution of EDC and NHS. This mixture was allowed to shake for 90 min, diluted to 3 mL of PBS, and then 132 μL of biotin (1 mg/mL) in PBS was added and the solution shaken for 3 h at 10 °C. The biotin-modified MGNT was immediately cleaned ultrasonically for 90 min at 10 °C, the supernatant was decanted, PBS was added, and the cleaning process was repeated three more times to eliminate the excess of biotin. Finally, the resulting MGNT is kept frozen until further streptavidin detection. Biotin concentrations anchored on the MGNT surface were determined by the Bradford protein assay. The result indicates that the concentration of biotin on the flat silicon substrate and MGNT surface were ca. 12.81 and 161.23 g/cm2, respectively, indicating the high surface area of MGNT. Static water contact angles (SWCAs) were measured while increasing the water drop volume and recorded the angles using a GH-100 Contact Angle System (KRÜSS, Germany) equipped with a

6

rotatable stage. Each SWCA was determined by fitting a Young–Laplace curve around the water drop. The experiment was performed under normal laboratory ambient conditions (25 °C, 40% relative humidity). The morphologies of the MGNT arrays were analyzed using a high-resolution scanning electron microscope (HR-SEM, JEOL JSM-6500F, Japan).

Figure 1 Schematic representation of the process used to fabricate MGNT arrays. A: Photoresist spun onto HMDS-treated Si wafer at a thickness of ca. 780 nm; contact hole arrays with various resolutions on the photoresist were successively patterned by lithography process. B: As-prepared photoresist templates on Si substrates subjected to deposition through sputtering of an alloyed Zr55Cu30Al10Ni5 target. C: Remaining photoresist removed from the silicon surface, leaving

7

behind the MGNT array layer. D: The substrate with MGNT arrays was modified using 3-aminopropyltriethoxysilane (APTES), and then anchored biotin by EDC and NHS reaction. The biotin-modified MGNT was exploited to detect streptavidin. 2.3 Reflectance spectroscopy A subwavelength structured (SWS) grating does not generate real diffracted light waves other than zero-order diffraction waves. For a periodic relief grating having a period of Δ, the grating vector G has a magnitude of 2π/Δ. When the angle of the incident light beam is θl and k is a wave number of the light in a vacuum, the conditions of a zero-order grating are expressed as (Kikuta et al., 2003)



i = 1, 2, 3, … and m = ±1, ±2, ±3, … n k sin  mG  n k , 1 1 i

(1)

where nl and n2 are the refractive indices of the incident space and transmitted space, respectively, and n3 is the mean refractive index of the grating layer. Optical reflectance spectra of the MGNT arrays were measured as a function of the incident angle (Chu et al., 2007; Wang et al., 2013). S(transverse electric, TE) or p- (transverse magnetic, TM) polarized light could be adjusted using the polarizer. Optical experiments were performed to examine the potential sensing applications and to evaluate the sensitivity of the MGNT arrays by measuring their angular reflection spectra (reflectivity vs. incident angle). When the laser beam was directed on the MGNT arrays, various optical waveguide modes were excited in the waveguiding layer at specific incident angles with an electric field polarized either perpendicular or parallel to the film surface (s- and p-polarizations, respectively), recognized as the waveguide coupling dips in the angular reflection spectrum (Lau et al., 2004), and the guided waves efficiently coupled to the surface of the MGNT arrays as well as the filling media. 2.4 Performance of Biotin-modified MGNT Arrays for streptavidin Detection Biotin-modified MGNT was incubated with streptavidin, by controlling BSA and AGRI concentrations in PBS solution possessing 0.5% Tween 20 (pH 7.4, 10 mM) at 10 °C overnight. When the reaction accomplished, the MGNT arrays were irradiated using a home-made laser

8

system to obtain the reflectance, and the detection selectivity of the MGNT was also investigated under competitive conditions. Then, the MGNT arrays of 1  1 cm2 were placed in contact with a binary mixture system, in which the concentration of streptavidin was 25 nM, and coexisting BSA or AGRI were various times of the streptavidin. The mixture was shaken for 12 h at 25 C prior to use. The selectivity was determined with the detection capacities of biomacromolecules in binary mixtures. 3. Results and Discussion 3.1 Surface Topography

Figure 2 Top-view and tilted SEM images of (a) 500-nm-resolution contact hole arrays, generated through the lithography process, and (b–d) MGNT arrays, deposited for 225, 450, and 675 s, after removing the photoresist template. 9

Figure 2a presents top-view and 45°-tilted SEM topographic images (resolution: 500 nm) of a photoresist template possessing hole arrays, designed as a duty ratio of 1.5. The Zr-based TFMG was deposited on the photoresist template by sputtering a quaternary alloy target for 225, 450, and 675 s (Chiang et al., 2006). The deposition rate of the TFMG through sputtering was approximately 800 nm/h. The as-prepared samples were cleaned ultrasonically in acetone for 30 s to remove the photoresist template. Figures 2b-d display top-view and 45°-tilted SEM topographic images of the template that had been subjected to deposition of TFMG for 225, 450, and 675 s, respectively, after ultrasonic cleaned with acetone. As depicted in Figure 2b, the MGNT formed a dense distinctive overlay with tube arrays on Si surface, with an average height of 784 nm, an outer diameter of 500 nm, and a tube wall thickness of 98 nm. The tube wall thickness increased as the deposition time increased to 450 s, forming a regular nanotube structure over a large area (Figure 2c). With a deposition time of 675 s, the nanotube array of MGNT exhibited a slight irregularity at the tube wall (Figure 2d). In addition, the MGNT with various hole diameters including 600, 700, and 800 nm, were fabricated using a deposition time of 450 s. The wall thickness of MGNTs decreased upon increasing the MGNT dimensions (Table S1). It is noticeable that the wall thickness of MGNT determined the quality of the MGNT array. Wall thicknesses of MGNT ranging from 40 to 150 nm provided ordered arrays of better quality over large areas. For a geometrically rough surface, contact angle hysteresis originates primarily from the rough contact interface, and depends upon the contact area of water with the structured surface. Wenzel (Wenzel et al., 1936) and Cassie (Cassie et al., 1944) developed the earliest models of liquid drops on rough surfaces. The Cassie–Baxter models are described by the equation f MG 

1  cos  MGNT 1  cos  MG

(2)

where θMGNT and θMG are the equilibrium (Young’s) SWCAs of the rough surface of MGNT and the smooth surface of TFMG, respectively; and fMG represents the volume fraction of the MGNT of the waveguiding layer. The SWCAs on these tube structure were recorded to calculate the values of fMG 10

(Table S1). The trapped air prevents the intrusion of water into the nanostructures, resulting in large water contact angles on the surface. Thus, the SWCA increased as a result of the surface roughness and the air trapped within nanotubes. Eqn (2) assumes that the liquid does not completely wet the rough surface. Once air is trapped in the interstices of a rough surface, the liquid droplet interacts with a composite surface comprised of a solid substrate and air pockets. From eqn (2), a very hydrophobic surface would be realized if the air fraction were sufficiently large. For the MGNT array surfaces, the air fractions were evaluated to be in the range 0.36–0.74 (Table S1). For TFMG, the air fraction was 0, indicating that no air occupied its surface. For the MGNT array surfaces, the air fraction decreased upon increasing the dimension of the nanotubes; that is, the MGNT surfaces with larger dimensions facilitated the containment of air inside the nanotubes with lower degrees of leakage. The air fraction reached as high as 0.74, meaning that approximately 74% of the MGNT array surface was occupied by air. Therefore, the high air fraction of the MGNT arrays surfaces was expected to decrease the effective refractive indices for TM and TE polarizations. The ordered structure of the MGNT arrays enhanced the electromagnetic field of waveguide modes confined within the layer, due to its lower refractive index. 3.2 Optical Properties of MGNT Arrays Figure 3 presents the diffractive reflectance of the MGNT arrays possessing dimensions of 500, 600, 700, and 800 nm, for both TE and TM polarization. The diffractive reflectance intensity of the MGNT arrays increased with the dimensions in both cases. This observation suggests that MGNT arrays of smaller dimensions facilitated the trapping of light within the MGNT, resulting in lower diffractive reflection. Indeed, the MGNT array having a dimension of 500 nm exhibited the lowest diffractive reflectance intensity, due to its highest tube density, for both TE and TM polarization. Two reflective peaks appeared for the MGNT arrays at 576 and 759 nm for TM polarization. The reflective peak positions shifted gradually from 576 and 759 nm to 418 and 516 nm, respectively, upon increasing the dimension of the MGNT to 800 nm (Figure 3a). Only one diffractive reflective

11

peak, at 713 nm, appeared for TE polarization; which then shifted gradually to 577 nm as the dimensions decreased (Figure 3b). Notably, the diffractive reflective peak position varied with respect to the angles between the TE and TM directions, due to the different values of neff for TE and TM polarization. Therefore, the geometric structures of the MGNTs predominately determined the diffractive reflective peak position.

Figure 3 Diffractive reflectance spectra of the MGNT arrays possessing dimensions of 500, 600, 700, and 800 nm under (a) p- and (b) s-polarized light. In addition to the wavelength and polarization of the incident lights, the refractive index can also affect the angle of the diffractive peak for certain waveguide mode orders, m (Figure 4a). The optical waveguide mode for two kinds of polarized light for the 500-nm MGNT arrays was observed to be in the angular range 30–45°. In the angular reflection spectra measured with air, the TM1 waveguide mode was excited by p-polarized light within the MGNT arrays and the TE0 mode was excited by s-polarized light. The layer composed of the MGNT arrays and the filling medium acted as the waveguiding layer of the MGNT array, the effective refractive index of which can be expressed using the infinite, prolate ellipsoid approximation of the Maxwell–Garnett theory (Aspnes et al., 1982):

12

For p-polarized light: 2 2 1  f MG  nT2M  nMG f MG  nmed

(3)

For s-polarized light: 2 nT2E  nmed

 

2 2 2 2nmed  1  f MG  nMG  nmed 2 2 2 2nmed  1  f MG  nMG  nmed

 

(4)

where nTE and nTM are the effective refractive indices of the waveguiding layer for s- and p-polarized light, respectively; and nmed and nMG are the refractive indices of the surrounding medium and the TFMG, respectively. When combined with eqn (2), eqns (3) and (4) provide the following equations: For p-polarized light: 2 nT2M  nMG

1  cos  MGNT cos  MG  cos  MGNT 2  nmed 1  cos  MG 1  cos  MG

(5)

For s-polarized light: n n 2 TE

2 med



2 2 2 1  cos  MG   2  cos  MGNT  cos  MG  nMG 2nmed  nmed 2 2 2 1  cos  MG   cos  MG  cos MGNT  nMG 2nmed  nmed







(6)

Both waveguide modes provided very sharp minima in their angular reflectivity curves, compared with those measured using the TFMG in a water environment, of which the TM1 and TE0 modes were excited. These modes were labeled with the number of nodes in the optical field distribution (Knoll et al., 1998). The angle of the reflectance (θd) was 32.65° for the TM1 mode in air. When water filled the bare MGNTs within a flow cell, a large shift in the value of θd was observed from 32.65 to 34.3°, for the TM1 mode (Δθ = 1.65°; Figure 4b). Upon modifying with biotin and loading streptavidin at 25 nM, shifts of the values of θd also occurred, to 37.25 and 39.5°, respectively. Subsequently, different concentrations of streptavidin (25, 50, 75, 100, 125 nM) were injected into the flow cellcirculating with a pump system at the rate of 10 μL/min. As the concentration of the streptavidin increases, the dip undergoes a shift in the angular reflection spectrum.

13

Figure 4 (a) Schematic representation of MGNT waveguide measurement. Reflectances of samples were observed as very sharp spots for TM and TE optical waveguide modes when using light from a He–Ne laser. (b) Angular reflectance spectra of the MGNTs filling with air (■), water (▲), biotin (◆), and streptavidin at 25 (●), 50 (●), 75 (●), 100 (●), and 125 (●) nM for p-polarized light. An important comparative parameter for sensing devices, the figure of merit (FOM), can be defined as follows: FOM 

 n  R

(7)

where Δθ is the shift in the value of θd, Δn is the effective refractive index change of the sensing medium, and ΔR is the full-width of the reflection peak at half-maximum. The FOM takes into account the sharpness of the reflection peak and, thus, can be an efficient standard for examining the sensitivity of the sensors based on measuring tiny angular changes. Three independent substrates were measured in this study. Figure 5a displays the FOM plotted as a function of concentration of streptavidin, BSA and AGRI for TM1 and TE0 modes. For the TM1 mode of the MGNT array, the FOM ranged from 30 to 60—values higher than that reported for an optical waveguide sensor based on polycyanurate thermoset nanorods (Gitsas et al., 2010), and much higher than those for surface plasmon resonance sensors based on regular angular modulation (Homola et al., 1999). The TM1 mode had a relative amplitude higher than that of the TE0 mode, indicative of a more efficient waveguide coupling for the TM1 mode (Lazzara et al., 2010), presumably because of the much narrower value of ΔR of the TE0 mode as well as uniformities in 14

the thicknesses and optical properties of MGNT arrays, resulting in the shallower depth of the waveguide mode. The large FOM of the MGNT can be ascribed to the very sharp minima of the waveguide modes and the wide field distribution at a deep region (Chiang et al., 2006).

Figure 5 FOM values for the TM1 and TE0 modes of the biotin-modified MGNT plotted as a function of (a) concentration of streptavidin, BSA and AGRI, and (b) concentration ratio of BSA or AGRI to streptavidin (25 nM) in binary mixtures. Relative standard deviations (RSD) is ca. 5%. In addition, the FOM values increased linearly with the streptavidin concentration for both of TM1 and TE0 modes, while the FOM values did not change significantly with the BSA or AGRI concentrations. The results indicate the biotin-modified MGNT could detect the streptavidin precisely with high selectivity. Even in the binary mixture samples including 25 nM streptavidin and ten times concentrations of BSA or AGRI, the FOM values were not significantly affected, indicating the high capacity of resisting disturbance.(Figure 5b) Thus, the MGNT arrays allowed measurements of changes in effective refractive index with high accuracy. Indeed, these novel and conveniently prepared MGNT arrays displaying selective, sensitive, and repeatable responses to media appear and suitable for applications as optical sensors. In addition, these MGNT arrays exhibited visibly composed color images; their colors can be quite sensitive to the incident angles 15

(Ma et al., 2012; Zeng et al., 2017). In order to examine this phenomenon, the angle of the incident light was strictly fixed to 8° during our color measurements, and the multi-color variation was defined using the uniform chromaticity scale diagrams L*a*b* color space and a*b* coordinates proposed by the Commission Internationale de l’Éclairage (CIE) in 1976 (Figure 6). Notably, the flat surface of TFMG cannot display any surface color or reflective peaks. While the regular structure can display a regular color in a stable environment, the media filling inside the MGNT and all over the surface, changed the effective refractive indices as well as the colors. Upon filling the components, these MGNT arrays exhibited glowing colored spectra, transforming from white to purple (Figure 6a). For the 500-nm MGNT arrays filled with streptavidin at 25 nM, the hues perceived by the naked eye were purple. With this MGNT array, the high sensitivity of human visual perception could lead to these hues appearing or disappearing across the colorless gray point. On the other hand, when focusing on “hues” in visual colorimetry, it is assumed that more hues appearing during transitions simplify visual perception. With increasing concentrations of streptavidin, the L* and a* of the color scale decreased and increased slightly; however, b* of color scale decreased obviously. (Figure 6b) It is supposed that accurate quantification can be realized through large hue variations. Thus, these novel and convenient MGNT arrays, with selective, sensitive, and repeatable responses to their environments, appear suitable for applications as optical sensors.

16

Figure 6 (a) L*a*b* color space proposed by CIE in 1976 and color circle shown on a*b* chromaticity diagram of TFMG and MGNTs filling with air, water, biotin, and streptavidin at 25 nM. (b) Dependence of the L*a*b* color scale of MGNT arrays with the concentration of streptavidin from 25 to 125 nM. RSD is 9 %. 4. Conclusions The well-defined MGNT layer having a dimension of 500 nm has been fabricated with VLSI and sputtering processes. After modification of biotin, the MGNT array acted as a waveguiding layer for an optical sensor for both the TM and TE waveguide modes. The optical waveguide mode was excited and efficiently coupled to the surface of the MGNTs, as well as the filling medium. The 500 nm MGNT sensor waveguide could readily detect the streptavidin by monitoring the shift. However, the components of biological samples are always complex. The 500 nm MGNT might not satisfy to detect all kinds of biological models. Fabrication of MGNT having various dimensions is important issue to the design of an optimal MGNT waveguide sensor that might be used as a high performance and versatile sensing platform for bioanalytical applications in real sample.

Acknowledgment We thank the Ministry of Science and Technology of the Republic of China for supporting this research financially, and the National Nano Device Laboratory for assistance with the I-line lithography.

17

References Aspnes, D.E., 1982. Thin Solid Films 89, 249–262. Cassie, A. B. D., Baxter, S., 1944. Trans. Faraday Soc. 40, 546–551. Chen, J. K., Bai, B. J., 2011. J. Phys. Chem. C 115, 21341–21350. Chen, J. Chen, W., Cheng,C., Yu, C., Chu J. P., Mater. Today, in revision. Chen, J.-K., Li, J.-Y., 2010. Sens. Actuators B 150, 314–320. Chen, J.-K., Pai, P.-C., Chang, J.-Y., Fan, S.-K. 2012. ACS Appl. Mater. Interfaces 4, 1935–1947. Chen, J., Wang, J., Cheng, C., Chang, J., Chang, F., 2013. Appl. Phys. Lett. 102, 151906. Chen, J.-K., Wang, J.-H., Fan, S.-K., Chang, J.-Y., 2012. J. Phys. Chem. C 116, 6980–6992. Chen, J., Zhou, G., Chang, C., Cheng, C., 2014. Sens. Actuator B 194, 10–18. Chen, J., Zhou, G., Chang, C., Lee, A., Chang, F.,2014. Biosens. Bioelectron. 54, 35–41. Chen, J., Zhou, G., Huang, C., Chang, J., 2013. ACS Appl. Mater. Interfaces 5, 3348–3355. Chiang, C. L., Chu, J. P., Liu, F. X., Liaw, P. K., Buchanan, R. A., 2006. Appl. Phys. Lett. 88, 131902 Chu, J. P., Jang, J. S. C., Huang, J. C., Chou, H. S., Yang, Y., Ye, J. C., Wang, Y. C., Lee, J. W., Liu, F. X., Liaw, P. K., Chen, Y. C., Lee, C. M., Li, C. L., Rullyani, C., 2012. Thin Solid Films 520, 5097–5122. Chu, J. P., Wijaya, H., Wu, C. W., Tsai, T. R., Wei, C. S., Nieh, T. G., Wadsworth, J., 2007. Appl. Phys. Lett. 90, 034101. Gitsas, A., Yameen, B., Lazzara, T.D., Steinhart, M., Duran, H., Knoll, W., 2010. Nano Lett. 10, 2173–2177. Grünzweig, C., Pfeiffer, F., Bunk, O., Donath, T., Kühne, G., Frei, G., Dierolf, M., David, C., 2008. Rev. Sci. Instrum. 79, 053703. Hasan, M., Kumar, G., 2016. Scr. Mater. 123, 140–143. Hauser, E., Zirke, R. J., Tauc, J., Hauser, J. J., Nagel, S. R., 1979. Phys. Rev. B 19, 6331.

18

Homola, J., Koudela, I., Yee, S. S., 1999. Sens. Actuators B 54, 16–24. Huang, H., Chen, J., Houng, M., 2013. Sens. Actuators B 177, 833–840. Kambhampati, D.K., Jakob, T.A.M., Robertson, J.W., Cai, M., Pemberton, J.E., Knoll, W., 2001. Langmuir 17, 1169–1175. Kikuta, H., Toyotai, H., Yui, W., 2003. Opt. Rev. 10, 63–73. Kim, J., Lee, K.H., Lim, C.H., Kim, T., Ahn, C.W., Cho, G., Lee, S.W., 2013. Rev. Sci. Instrum. 84, 063705. Knoll, W., 1998. Annu. Rev. Phys. Chem. 49, 569–638. Lau, K. H. A., Tan, L.-S., Tamada, K. Sander,, M. S., Knoll,W., 2004. J. Phys. Chem. B 108, 10812–10818. Lazzara, T. D., Lau, K. H. A., Knoll, W., 2010. J. Nanosci. Nanotechnol. 10, 4293–4399. Lee,C. J., Lin, H. K., Huang, J. C., Kuan, S. Y., 2011. Scr. Mater. 65, 695–698. Lee, C. J., Lin, H. K., Sun, S. Y., Huang, J. C., 2010. Appl. Surf. Sci. 257, 239–243. Lee, S-W, Jafary-Zadeh, M., Chen, D. Z., Zhang, Y.-W., Greer, J. R., 2015. Nano Lett. 15, 5673– 5681. Ma, J., Yi, J., Zhao, D. Q., Pan, M. X., Wang, W. H., 2012. J. Appl. Phys. 112, 064505. Muskens, O. L., Rivas, J. G., Algra, R. E., Bakkers, E. P. A. M., Lagendijk, A., 2008. Nano Lett. 8, 2638–2642. Ray, S., Tauc, J., 1980. Solid State Commun. 34, 769–772. Saotome, Y., Fukuda, Y., Yamaguchi, I., Inoue, A., 2007. J. Alloys Compd. 434–435, 97–101. Schroers, J., 2010. Adv. Mater. 22, 1566–1597. Skivesen, N., Horvath, R., Pedersen, H. C., 2004. Sens. Actuators B 106, 668–676. Wang, W. H., Hsu, J. H., Huang, J. C., 2013. Appl. Phys. Lett. 106, 161906. Yashiro, W., Noda, D., Hattori, T., Hayashi, K., Momose, A., Kato, H., 2014. Appl. Phys. Exp. 7, 032501.

19

Zeng, J., Cheng, C., Chang, C., Huang, C., Chen, J., 2017. Dyes Pigments 139, 300–309. Zeng, J., Cheng, C., Huang, B., Huang, C., Chen, J., 2017. Sens. Actuators B 243, 234–243. Zhu, J., Yu, Z. F., Burkhard, G. F., Hsu, C. M., Connor, S. T., Xu, Y. Q., Wang, McGehee, Q. M., Fan, S. H., Cui, Y., 2009. Nano Lett. 9, 279–282.

Highlights  Ordered metallic glass nanotube (MGNT) array has been fabrucated on silicon surface.  Filling the MGNT with media results the significant change in diffractive reflectance. 

A standard biotin–streptavidin affinity model was tested using the MGNT arrays.

 Filling the MGNT with streptavidin results the color change identified by an L*a*b* color space.

 Sensitivity of detect the streptavidin reaches to 25 nM according to the change in reflectance.

20