Article pubs.acs.org/Langmuir
Optimizing Pin-Printed and Hydrosilylated Microarray Spot Density on Porous Silicon Platforms Dustin T. McCall,† Yi Zhang,† Daniel J. Hook,‡ and Frank V. Bright*,† †
Department of Chemistry, Natural Sciences Complex, SUNY-Buffalo, Buffalo, New York 14260-3000, United States Bausch + Lomb Incorporated, 1400 Goodman St. N., Rochester, New York 14609, United States
‡
ABSTRACT: Microarrays of spatially isolated chemistries on planar surfaces are powerful tools. An important factor in microarray technology is the density of chemically unique spots that can be formed per unit area. In this paper, we use contact pin-printing and evaluate how to decrease contact pin-printed spot diameters on porous silicon (pSi) platforms. Using hydrosilylation chemistry to covalently attach chemistries to the pSi surface, the variables studied included pSi porosity and surface polarity, active agent viscosity, and pin diameter. The spot characteristics were assessed by Fourier transform infrared spectroscopy (FT-IR) microscopy and X-ray photoelectron spectroscopy (XPS). Spot size decreased as pSi porosity increased in accordance with molecular kinetic theory and Darcy’s law of imbibition. Increasing active agent viscosity and pin diameter (volume of printed agent) led to larger spot diameters in accordance with molecular kinetic theory and Darcy’s law. Oxidizing the pSi with H2O2 increased the surface polarity but had no detectable impact on the spot size. This is consistent with formation of an oxide layer atop an unoxidized pSi sublayer.
■
⎛ 1000 ⎞2 ⎟ D = 100⎜ ⎝ CTC ⎠
INTRODUCTION Porous silicon (pSi), first discovered in the 1950s,1 exhibits bright red-orange photoluminescence2 caused by quantum confinement3 and can be easily incorporated into existing silicon integrated circuits.4,5 pSi has been used as an optical sensing platform for detecting gases, liquids, and vapors,6−8 and biomolecules9 by exploiting conductance,10 impedance,6,11 photoluminescence,7,8,12 reflectance,13,14 and resistance15 measurements. Functionalizing pSi to produce other surface chemistries can impart chemical stability,16 provide sites for secondary chemical reactions and modifications,17 and improve selectivity.18 Hydrosilylation16,18,19 is an attractive functionalization method because it yields monolayers at the Si surface and it yields Si−C bonds which are more stable in comparison to the native SiHx (x = 1−3) residues that exist on the pSi surface. Hydrosilylation of pSi with aliphatic chains, carboxylic acids, cyclic oligosaccharides, and esters in concert with photoluminescence and reflectivity measurements20−22 has been used to selectively detect organic analytes in a one-at-a-time fashion. Clearly, delivering multiple active agents on a pSi sample in a spatially controlled manner could lead to pSi-based microarrays23−25 and simultaneous multianalyte detection strategies. The ability to create discrete chemistries (spots) on planar solid surfaces has led to the development of microarrays in areas ranging from glucose26 and oxygen sensing27 to bacteria,28 biomolecule,29 gene,30 and protein31 identification and quantification. One important factor in modern microarray technology is the spot density (D, spots/cm2) which depends on the center-to-center (CTC, μm) spacing between adjacent spots:32 © 2015 American Chemical Society
(1)
In this expression, CTC represents the actual spot diameter multiplied by 1.2−1.4 (a 20−40% buffer between spot edges). As examples, microarrays with 100 μm diameter spots yield D values in the 5100−6900 spots/cm2 range whereas microarrays that have 10 μm diameter spots exhibit D values in the 510 000−690 000 spots/cm2 range. Clearly, producing spots with smaller diameters has substantial appeal and benefit. Although numerous strategies exist for creating chemically unique spots on planar surfaces (e.g., contact pin-printing,33 dip-pen nanolithography,34 inkjet printing,35 microstamping,36 and photolithography37), pins remain the most common tools for creating microarrays38 due to their relatively inexpensive costs compared to newer piezoelectric printing techniques.39 Commercially available microarray pins claim 37.5 μm spot diameters;40 however, contact pin-printed spot diameters on nonporous materials can be reduced by decreasing substrate wettability and print agent viscosity,41 and spot-to-spot size reproducibility depends on factors occurring during the time between reagent uptake by the pin from the well and pin− substrate contact.42 Previous work by our group reported on a solid fused silica pin for creating 9 μm diameter spots on glass.32 More recently, we described contact pin-printing of silicon alkoxides and silica sols on oxidized pSi (ox-pSi) for high-throughput screening applications;23 however, there was Received: July 21, 2015 Revised: September 23, 2015 Published: September 30, 2015 11370
DOI: 10.1021/acs.langmuir.5b02692 Langmuir 2015, 31, 11370−11377
Article
Langmuir
The pSi thickness was determined by using an Alpha Step IQ surface profiler (KLA Tencor). Here, individual pSi wafers were subjected to 12 mL of 1:5 (v:v) ethanol:1 M KOH solution with sonication for 10 min to dissolve away the pSi layer. These samples were then rinsed with ethanol and dried in a vacuum oven. The pSi thickness was determined by measuring the average depth from the region where pSi had been removed. Static contact angle measurements were performed by using a model 190-F1 contact angle goniometer (Ramé-Hart Incorporated). Measurements were made in triplicate with 40 μL of deionized water and the DROPimage CA analysis software. The high surface tension of water, the larger volume of water used, and the small pSi pore sizes retarded H2O imbibition into the pSi pore network allowing for accurate measurements. Sample masses were determined by using a UMX2 ultramicrobalance (Mettler Toledo). PSi Preparation.23 Silicon wafers were cut into 1.6 × 1.6 cm squares by using a glass slide cutter. The wafer matte side was then sanded with 1200 grit silicon carbide sandpaper to remove surface oxides. The wafer was then rinsed in ethanol for 150 s and gently dried with lint free tissue paper to remove silicon dust and organics. A thin Ga/In eutectic layer was applied to the wafer matte side with a clean cotton swab to create an ohmic contact between the silicon and copper electrode. The silicon anode and copper contact electrode were then placed between the standard etch cell and base so that the polished silicon side is exposed to the cell well. The cell well is filled with 4.5 mL of 1:2 (v:v) HF:ethanol solution. The HF solution is initially allowed to remain in contact with the silicon surface for 60 s prior to initiating anodization. This step helps to remove any surface oxides. The platinum cathode was shaped into a 1.7 cm diameter ring, and the ring plane is placed parallel to and 1.9 cm above the silicon wafer surface. A 21 mA/cm2 current density is applied for 30, 60, 150, 300, and 600 s to yield pSi with different porosities (vide inf ra). The pSi wafers were subsequently removed from the cell and carefully rinsed consecutively in 20 mL of 1:2 (v:v) water:ethanol, ethanol, and pentanes for 150 s each. All residual Ga/In eutectic was removed from the pSi sample with ethanol. The as-prepared pSi (ap-pSi) wafer was then placed in an evacuated desiccator (125 Torr vacuum) for 15 min to remove residual pentanes. The ap-pSi was kept in the evacuated desiccator for any temporary storage. The ap-pSi samples were completely functionalized within 6 h after preparation. Porosity and Depth Calculations. The Morales-Masiś et al. gravimetric analysis method was used to determine the average pSi porosity.49 PSi Oxidation.44 Separate ap-pSi samples (150 s anodization times) were fully submerged into H2O2 at room temperature for 15, 30, 60, and 120 min to create ox-pSi with different degrees of oxidation as a way to adjust the pSi surface polarity. The ox-pSi samples were subsequently rinsed and stored as described for the ap-pSi samples (vide supra). Pin-Printing and Hydrosilylation. Each pure hydrosilylating agent was individually contact pin-printed onto the pSi sample surface. In all cases, the pin contact acceleration was 500 mm/s2, the maximum pin contact speed was 7.5 mm/s, and the pin−surface contact time was 7 ms. To initiate hydrosilylation, a pSi sample that had been pinprinted upon was placed into a flow cell that had a quartz window and dry Ar gas was flowed over the pSi surface at 45 mL/s to minimize surface photo-oxidation.50 The flow cell was also water cooled with a copper jacket and fan cooled to prevent pSi surface cracking due to heat from the lamp. The pSi chip face was then exposed to UV electromagnetic radiation for 5 h to perform a combination of exciton mediated hydride abstraction18,19 and UV photoinduced radical51 hydrosilylation. After the hydrosilylation reaction was ended, the pSi microarray was removed from the flow cell and soaked in 20 mL of anhydrous npentane, a hydrosilylation agent soluble solvent, for 15 min to remove any unreacted hydrosilylating agent. We were unable to detect any free hydrosilylating agent within or on the pSi by monitoring the CC stretching band using FT-IR spectroscopy. The pSi microarrays were subsequently stored as described for the ap-pSi samples (vide supra).
no consideration of spot density or bona fide microarray development on pSi in our earlier work. In this paper, we explore the effects of pSi porosity and surface polarity, pin diameter, and hydrosilylation agent viscosity on the contact pin-printed spot size. pSi porosity was controlled by adjusting the pSi anodization time,43 and pSi surface polarity was controlled by using H2O2 oxidation.44
■
EXPERIMENTAL SECTION
Materials, Supplies, and Reagents. The following were used: ptype B-doped ⟨100⟩ CZ processed 8−12 Ω·cm Si wafers (Alsil Supply Division, Y Mart Int.); acrylic acid (99.5%) and HF (48−51%) (Acros Organics); 200 proof ethanol (Decon Laboratories, Inc.); pentanes (99.7%) (Fisher Scientific); allylamine (99%), Ga/In eutectic (99.99%), and anhydrous n-pentane (≥99%) (Sigma-Aldrich); 1heptadecene and 1-octadecene (90% GC grade) (TCI America); H2O2 (30%) and solid KOH (87.5%) (J.T. Baker); argon gas (Jackson Welding Supply Co., Inc.); 19 GA pure Pt round wire (Hoover and Strong); 30 GA ⟨110⟩ electrical grade Cu sheet (MSC Industrial Supply Co.); and deionized H2O produced by a 50k light Silex deionizer (AmeriWater, Inc.). Equipment. The following were used: a polytetrafluoroethylene (PTFE) etch cell modeled after Sailor;45 a PTFE flow cell with a quartz window; a Keithly model 2400-LV dc power supply; a UVG-54 hand-held UV lamp (254 nm, 7 mW); a model 1410 vacuum oven (Sheldon Manufacturing); a model 75HT ultrasonic bath (Aquasonic); a ProSys 5510 pin-printer (Cartesian Technologies); 50 μm diameter MicroSpot C ceramic tipped pins (Apogent Discoveries); and 200 and 400 μm solid tungsten pins (Point Technologies). Instrumentation. XPS measurements were performed by using a model Quantera SXM scanning X-ray microprobe with Al Kα source (1486.6 eV, 38.7 W, 200 μm beam diameter), Ar+ sputter gun (4 keV, 25 nA), and an electron filament charge compensation gun (1.0 V, 20 μA) (ULVAC-PHI, Inc.). The sample was held at a 45° angle with respect to the hemispherical analyzer, and 112 eV pass energy was used for all measurements. Zalar rotation46 was used to avoid differential sputtering and increase spatial depth resolution. Under these conditions, the sputter area diameter was ∼4 mm and the sputter rate was 25 nm/min (vide inf ra). The O 1s, C 1s, and Si 2p binding energy spectra were recorded with a resolution of 0.1 eV while alternating between sputter and analysis cycles (Table 1).
Table 1. Experimental Parameters for XPS Depth Profiling of pSi pSi porosity (%) 26
40 or 47
64
binding energy peak oxygen 1s carbon 1s silicon 2p oxygen 1s carbon 1s silicon 2p oxygen 1s carbon 1s silicon 2p
integration time per spectral step (s)
steps per spectrum
spectra averaged per cycle
acquisition/ 60 s sputter cycles
0.050
200
8
27
0.050
200
8
51
0.050
200
8
101
FT-IR spectra were measured by using an infrared microscope (Bruker, Hyperion 3000 microscope interfaced to a Vertex 70 FT-IR bench) equipped with a single-element HgCdTe (MCT) detector. Spectra were recorded over 50 × 50 μm areas in the transmission mode with 4 cm−1 resolution using an air background and by averaging 32−128 scans. Vector scans across a contact pin-printed spot typically used 50 μm step sizes. All FT-IR spectra were normalized to the Si−Si stretching mode at 620 cm−1 to correct for any differences between pSi samples and to compensate for multiple reflections.47,48 11371
DOI: 10.1021/acs.langmuir.5b02692 Langmuir 2015, 31, 11370−11377
Article
Langmuir
τ(x) describes the time at which the pores at radius x begin to imbibe with the liquid. Numerical evaluation of eqs 2−4 has been used to successfully describe and model liquid spreading on porous membrane materials.55,56 Inspection of eqs 2−4 reveals that drop spreading rates across the face of a porous material are directly proportional to the reagent volume on the surface at any time. Higher volumetric flow rates into the porous material will lower the instantaneous liquid volume remaining on the surface at any given time. Thus, careful selection of liquid viscosity, molecular volume, and surface tension in concert with control over substrate material porosity, permeability, and pore radius can control spreading (i.e., spot size) on porous materials like pSi.
IR and XPS Analysis. IR absorbance line profiles were constructed by measuring the FT-IR spectrum between 600 and 4000 cm−1 at each spatial position across each contact pin-printed spot. The acrylic acid-, 1-octadecene- and 1-heptadecene-, and allylamine-derived spot diameters were determined by following the position-dependent 1731 cm−1 (CO stretch),52,53 2916 cm−1 (CH2 asymmetric stretch), 52,53 and 2921 cm −1 (CH 2 asymmetric stretch) 52,53 absorbance bands, respectively. XPS depth profiles for C, Si, and O were acquired by measuring the C 1s, Si 2p, and O 1s peak areas, respectively, for hydrosilylated 1octadecene-derived spots formed on ap-pSi. The XPS depth was calibrated by using the known pSi thickness determined by profilometry (vide supra) and the Si 2p and O 1s signals. In all cases, the point where the O 1s signal equaled the noise and the Si 2p atomic concentration neared 100% was coincident with the pSi depth/ thickness determined by profilometry. Using this information, the sputter rate was determined (25 nm/min); hence, the depth anywhere within the pSi sample as a function of sputter time is known. Data Analysis and Statistics. For all data, statistical significance was assessed by using one-way ANOVA at the 95% confidence level with pairwise comparison (Holm-Sidak test) (p < 0.05 being significant). In all cases, the power of performance test exceeded 0.96.
■
RESULTS AND DISCUSSION pSi porosity depends on anodization time.43 Figure 1 presents the anodization time-dependent pSi porosity under our
■
THEORY SECTION There have been several reports on factors governing spot size from contact pin-printing on planar, solid surfaces.41,42 When a liquid drop contacts such a surface, the instantaneous dynamic contact angle, θD, will relax in a time-dependent manner from 180°, at initial contact time, toward the static contact angle, θS, at equilibrium.54 Simultaneously, the liquid moves across the solid surface with a radial velocity, U (i.e., ∂r/∂t) that is given by the molecular kinetic theory:55 ⎛ γ(cos(θS) − cos(θD)) ⎞ 2κ 0hλ ∂r = U = S sinh⎜ ⎟ ∂t ην 2nkBT ⎝ ⎠
Figure 1. Effects of anodization time on ap-pSi porosity. Current density = 21 mA/cm2.
(2)
In this expression, κ0S represents the molecular jump frequency at n sites of liquid−solid interaction per unit area, h is Planck’s constant, λ is the molecular jump length (= 1/n1/2), η is the printed liquid viscosity, ν is the liquid molecular volume, γ is the liquid surface tension, kB is Boltzmann’s constant, and T is the absolute temperature. To fully describe the drop relaxation process requires an additional expression that relates the instantaneous drop radius at the surface, r, to θD. Fortunately, because our contact pin-printed drop sizes are on the micron scale, the Bond number will be small enough so we can neglect gravitational forces, the spherical cap approximation can be used, and the expression for r is given by55 ⎤1/3 ⎡ 3V sin 3(θD) ⎥ r=⎢ ⎣ π 2 − 3 cos(θD) + cos3(θD) ⎦
experimental conditions (21 mA/cm2). These results show that anodization times between 30 and 600 s yield ap-pSi with porosities between 26 and 75%, respectively. In our hands, increasing the anodization time beyond 600 s leads to ap-pSi samples that become too fragile to print upon. We next questioned how anodization time affected the pSi surface chemistry. Figure 2 presents typical anodization timedependent IR absorbance spectra across the Si−F3,57 Oy−Si− F57 (y = 1−3), and Si−O−Si58 (950−1200 cm−1) (panel A) and SiHx58 (x = 1−3) and Oy−SiH58 (y = 1−3) (2000−2300 cm−1) (panel B) spectral regions. The Si−F3/Oy−Si−F/Si−O− Si IR spectral region (Figure 2A) shows that the Si−O−Si absorbance remains relatively constant, and there is little detectable Si−F3/Oy−Si−F species until the anodization time reaches and exceeds 300 s. The Oy−Si−F57 (y = 1−3) and Si− F357 species are very prominent when pSi has been etched for 600 s. The presence of Si−F3 and Oy−Si−F at longer anodization times arises from HF stoichiometric limitations.59 Figure 2B shows that the silane absorbance increases with increasing anodization time, as one would expect for a layer of pSi that is increasing in thickness; however, oxidized silane species are also clearly observed when pSi has been etched for 600 s. At 600 s anodization times, HF is no longer present and reaction of Si−F3 and SiHx (x = 1−3) species with water to form Oy−Si−F (y = 1−3) and Oy−SiH (y = 1−3), respectively, becomes the dominant pathway.45 There is little contrast between the photoluminescence (PL) from ap-pSi/ox-pSi and the same materials when they have been hydrosilylated with the active agents used in this research
(3)
where V is the drop volume and all other terms have been defined earlier. On a planar, porous surface, the situation becomes more complex because V is no longer constant, and some fraction of the liquid drop volume imbibes into the porous material, again, in a time-dependent manner. To describe this process, one can invoke Darcy’s law in one dimension:55 ⎡ γk cos(θS) ⎤1/2 δV = φ⎢ ⎥ ⎣ ⎦ δt ηa
∫0
r
t
2πx
∫τ(x) (t0 + (t − τ(x)))−1/2 dt dx (4)
In this expression, ϕ, k, and a represent the material porosity, permeability, and effective pore radius, respectively. The term 11372
DOI: 10.1021/acs.langmuir.5b02692 Langmuir 2015, 31, 11370−11377
Article
Langmuir
increases as the pSi porosity increases. This occurs because the pin force remains constant, but the pSi is becoming progressively more fragile as its porosity increases. Finally, the average pSi porosity-dependent 1-octadecene-derived spot diameter ranges from 670 to 1940 μm (∼3 fold) as the pSi porosity increases (Figure 4), decreasing linearly (p < 0.001, r2
Figure 4. Effects of ap-pSi porosity on 1-octadecene contact pinprinted spot diameter.
= 0.9953). This behavior arises from increased pSi porosity leading to increased pSi permeability leading to a higher volumetric flow into the pSi pore network and less time for spreading on the surface before the hydrosilylation agent fully imbibes into the pSi (eqs 2−4). Another potential issue with contact pin-printing onto a porous specimen is the limited volume of reagent delivered during the printing process (nL-pL).33 If this were to become an issue, as the pSi thickness and porosity increase, one would encounter a situation where the pin-delivered reagents could become limiting. Figure 5 explores this issue using the data
Figure 2. Effects of anodization time on the pSi IR absorbance spectra. (A) Si−F3, Oy−Si−F (y = 1−3) and Si−O−Si region. Asymmetric (νa) and symmetric (νs) stretching bands are indicated. (B) SiHx (x = 1−3) and Oy−SiH (y = 1−3) region.
so spot diameters were determined by using IR microscopy. Figure 3 presents a typical series of porosity-dependent IR
Figure 3. Effects of ap-pSi porosity (%) on the IR absorbance line profiles across three 1-octadecene contact pin-printed spots formed on ap-pSi at 2916 cm−1 (asymmetric CH2 stretch). The individual profiles are offset along the y-axis for clarity.
Figure 5. Effects of ap-pSi porosity on IR absorbance at 2916 cm−1 (related to the 1-octadecene grafting density).
absorbance line profiles across three 1-octadecene-derived spots that were printed by using a 50 μm diameter ceramic pin. In these experiments, the 2916 cm−1 band was monitored, and the profiles are offset along the y-axis to aid visualization. The spot diameters were determined by measuring the full width at baseline (FWB) from these profiles. Inspection of these results reveals three interesting points. First, there is often a decrease in IR absorbance at the spot center, indicated by the dashed lines. This arises from pin-induced damage and cannot be eliminated but has been reduced through optimization of the pin-printing parameters.23 Second, the extent of pin damage
from Figure 3. The plateau observed at higher pSi porosities suggests that the reagents delivered by the pin become limiting as the pSi porosity increases. Given the results from Figure 5 and eqs 2−4, we questioned the pSi porosity-dependent depth to which 1-octadecence, delivered during contact pin-printing, imbibes into and is hydrosilylated to the pSi network. This is a complex process governed at the minimum by the 1-octadecence volume delivered during the printing process, the imbibing and spreading kinetics (eqs 2−4), the pSi thickness, and the ability of the UV electromagnetic radiation to access the pSi and 11373
DOI: 10.1021/acs.langmuir.5b02692 Langmuir 2015, 31, 11370−11377
Article
Langmuir
demonstrates that 1-octadecene penetrates into the pSi in a porosity-dependent manner to depths ranging from 0.5 to 1.5 μm for porosities between 26% and 64%, respectively. This behavior is in line with eqs 2−4.56 An illustration describing the porosity-dependent 1-octadecence penetration/hydrosilylation into ap-pSi is given in Figure 6D. (Note: there is a line break in the 64% porosity material.) For ap-pSi with 26% porosity, the entire pSi thickness is hydrosilylated whereas only 20% of the available pSi thickness is hydrosilylated for pSi having 64% porosity. (Note: the C 1s exhibits a baseline atomic concentration of ∼9% that arises from adventitious carbon species entrapped within the pSi pore network. This characteristic exists in ap-pSi, too.61 The C 1s signal decreases below the noise once bulk crystalline silicon is reached during sputtering.) The pSi polarity, through θS, is another factor that could control the spot size, and Figure 2 suggests that the pSi porosity is not being changed in isolation. To explore this issue in more detail, we measured the static water contact angle (θS,H2O) for our ap-pSi samples. The results (Table 2) demonstrate that
activate the hydrosilylation process. To address this point, we performed depth-dependent XPS experiments on 1-octadecene-derived spots formed on ap-pSi with porosities ranging from 26%−64%. (Note: the 75% porosity sample was not studied because the sample pore network could not withstand the XPS ultrahigh vacuum.) Figure 6 presents typical pSi porosity-dependent Si 2p (panel A), O 1s (panel B), and C 1s (panel C) depth profiles for 1-
Table 2. Effects of Porosity (Anodization Time) and H2O2 Surface Oxidation Time on pSi Contact Angle porosity (%)
H2O2 oxidation time (min)
26 40 47
0 0 0 15 30 60 120 0 0
64 75
θS,H2O (deg) 121.4 125.3 127.0
