ARTICLE pubs.acs.org/Langmuir
Preparation and Characterization of Pore-Wall Modification Gradients Generated on Porous Silicon Photonic Crystals Using Diazonium Salts Corrina M. Thompson,† Anne M. Ruminski,‡ Adrian Garcia Sega,‡ Michael J. Sailor,‡ and Gordon M. Miskelly*,† † ‡
School of Chemical Sciences, University of Auckland, Private Bag 92019, Auckland 1142, New Zealand Department of Chemistry and Biochemistry, The University of California, San Diego, 9500 Gillman Drive, La Jolla, California 92039-0358, United States
bS Supporting Information ABSTRACT: One-dimensional photonic crystals (rugate filters) constructed from porous silicon were modified by the chemical hydrosilylation of terminal alkenes (decyl, 10-carboxydecyl, and 10-hydroxydecyl) in the presence of a concentration gradient of diazonium salt initiators. The concentration gradient was generated by vertically orienting the Si wafer containing the porous Si layer in an alkene solution and then introducing the diazonium salt at the bottom edge of the wafer. Slow diffusion of the salt led to a varying density of grafted alkene across the surface of the porous layer. The modified surfaces were end-capped with methyl groups by electrochemical grafting to impart improved stability and greater hydrophobicity. The surface modified with 10-carboxydecyl species was ionized by deprotonation of the carboxy groups to increase the hydrophilicity of this porous silicon surface. The pore-wall modification gradients were characterized using attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) and scanning electron microscopy (SEM) with energy-dispersive spectroscopy (EDS). The more hydrophilic portion of the gradient changes color when water infiltrates the porous nanostructure because of a shift in the stop band of the photonic crystal. The more hydrophobic portion of the gradient excludes water, although mixtures of water and ethanol will infiltrate this region, depending on the concentration of ethanol in the mixture. A simple visual sensor for small quantities of ethanol in water, capable of detecting ethanol concentrations of between 0 and 8% with a resolution of 1% is demonstrated.
’ INTRODUCTION A cross-reactive sensor array typically consists of a 2D assembly of individual sensor elements whose responses to a given analyte differ from element to element.1 4Although such devices are usually constructed in parallel, each physically and chemically distinct element generally requires a different synthesis process, which can be tedious and complicated to manufacture. A simple method of incorporating compositional variations in a sensor array is to artificially generate a concentration gradient in one or more of the reactants during the synthesis or deposition of the array. This approach has been reported on flat surfaces,5 8 but it has been limited by the detection methods required to monitor the array. Porous Si is readily adapted to a gradient approach because of the optical methods used to probe the sensor. For example, Fabry Perot interference films, microcavity structures, or 1D photonic crystals can be made from porous Si to provide optically probed devices that can detect chemical or biological compounds very sensitively.9 17 The use of an asymmetrical electrode configuration, either during electrochemical etching of the nanostructure18 20 or during electrochemical grafting,21 provides a simple means to generate morphological or chemical gradients (in the x y plane) in these devices that yield an improved ability to discriminate between analytes.18,19,21,22 r 2011 American Chemical Society
In our prior efforts to generate chemical gradients in a porous Si film, we were unable to obtain a transverse gradient modification that was constant through a thicker porous silicon layer (∼10 μm) using electrochemical grafting methods. However, a dip-coating method has been shown to generate a gradient pore modification in oxidized porous silicon films using chlorosilane solutions.23 We hypothesized that such an approach would provide a chemical gradient that would be useful in producing an optical array-based sensor. In this work, we chose to use the reaction of porous Si with a solution containing diazonium salts and alkenes to generate the chemical functionality on the porous Si films. The reaction has been reported by Wang and Buriak to lead to the attachment of organic groups to the porous silicon via Si C bonds24 and is thought to proceed through the creation of silicon radicals on porous Si via the spontaneous one-electron reduction of the diazonium salt. The silicon radicals can then react with either an aryl group of the diazonium salt or with a terminal alkene, alkyne, or alky/arylselenoether group. The silicon radicals react Received: April 6, 2011 Revised: June 6, 2011 Published: June 23, 2011 8967
dx.doi.org/10.1021/la201272e | Langmuir 2011, 27, 8967–8973
Langmuir
ARTICLE
preferentially with alkenes and alkynes to form Si C bonds with only low levels of concomitant oxidation. This reaction occurs within 3 h at room temperature and accommodates various functional groups such as carboxylic acids and hydroxyl groups on the opposite end of the alkene, although molecules with terminal halogen atoms could not be attached.24 Here we report that the chemical hydrosilylation of alkenes by porous Si in the presence of a diffusion gradient of diazonium salts can lead to pore-wall modification gradients that are uniform through thicker porous silicon layers (>3 μm). The resulting structures display spatially dependent wettability, and a simple demonstration of the discrimination capability of the graded surface is provided: the detection of low concentrations (0 8%) of ethanol in water.
’ RESULTS AND DISCUSSION The monitoring of changes in optical reflectance across a compositional gradient porous silicon sample is most readily accomplished if the sample has uniform optical reflectance prior to the pore-wall modification. Therefore, the uniformity of porous silicon samples is important for optical sensing for constant, comparable, and reliable results. A second important consideration is the internal structure of the porous silicon layer. Sensing changes in the effective refractive index of porous silicon using simple sensors, cameras, or the human eye is easier with porous silicon rugates (1D photonic crystals) rather than using the interference fringes of a Fabry Perot constant-porosity layer. These rugate porous silicon structures are created using a sinusoidally varying current during the etching process to form the porous silicon and usually have 30 100 periods and thicknesses on the order of 10 50 μm.25 27 The rugate reflector porous silicon samples used in this study showed more uniform optical reflectance across a given sample than did the Fabry Perot porous silicon samples used in our previous study of poremodification gradients.21 Rugate reflector porous silicon samples tend to be thicker than the Fabry Perot samples because a large number of porosity changes cycles are required to create a strong rugate reflectance color that can be used to monitor changes with the human eye or with simple instrumentation. These thicker layers were not amenable to pore-wall gradient modification using the electrochemical method we reported earlier, so we have explored chemical methods to prepare these gradient-modified porous silicon samples. Preparation of Pore-Wall Modification Gradients. The chemical hydrosilylation of alkenes in the presence of diazonium salts24 was explored as a method for gradient modification of the pore walls of porous silicon because this reaction occurs at room temperature and over a period of several hours, which allows variations of exposure time and concentration to create porous silicon samples with different extents of pore-wall modification. Lewis acid-catalyzed hydrosilylation28 31 was not explored for gradient formation because of the longer reaction times required (>12 h) for alkenes,31 which is longer than the diffusion time of a solution across a porous silicon sample of diameter ca. 10 mm. Vertical diffusion gradients of diazonium salts across porous silicon surfaces were prepared under Schlenk line conditions. A solution of the alkene was added to immerse the vertical silicon sample, and then a gradient of the diazonium salt was generated by carefully introducing the diazonium salt solution at the bottom of the alkene solution and allowing it to diffuse up past the porous silicon sample over a period of 3 to 4 h. One-directional pore-wall modification gradients on porous silicon were created using
Figure 1. Functional groups that have been grafted onto porous silicon by the chemical hydrosilylation of alkenes in the presence of a diffusion gradient of diazonium salts: (a) a decyl surface, (b) a 10-carboxydecyl surface, (c) a 10-carboxylatodecyl surface, and (d) a 10-hydroxydecyl surface.
decene, 9-decen-1-ol, or 10-undecenoic acid. This modification could be followed by electrochemical methylation with iodomethane with symmetric electrode placement to end-cap many of the reactive Si H groups remaining within the porous silicon layer and so improve the stability of the porous silicon toward oxidation.32 Using a magnesium anode instead of platinum for this step allowed the use of longer electrochemical-modification times with less oxidation of the porous silicon,21 which allowed for higher levels of end-capping through the porous silicon layer. The porous silicon modified with 10-carboxydecyl groups can be treated with ethanolic potassium hydroxide solution to give a carboxylate-terminated surface to increase the hydrophilicity of the porous silicon. The moieties grafted to porous silicon during this study are shown in Figure 1. Characterization of Pore-Wall Modification Gradients. The porous silicon samples with pore-wall modification gradients have been characterized using ATR FTIR spectroscopy and SEM EDS, together with examining their interactions with water. The ATR FTIR spectra of a methyl-end-capped 10-hydroxydecyl gradient-modified rugate reflector porous silicon sample obtained at different positions across the surface are shown in Figure 2. The C H stretching bands (ca. 2900 cm 1) show that the highest level of modification of the porous silicon is around 2 mm from the lowest edge during the diffusion/hydrosilylation process, and there is a decrease in the intensity of these IR bands with distance from this edge. Because the intensity of the ATRFTIR spectra varies slightly depending on the contact made between the ATR crystal and the sample, these spectral changes along the pore-wall modification gradient are best evaluated by comparing the intensity of the C H stretching bands corresponding to the 10-hydroxydecyl groups with that of the Si-CH3 rocking mode of the end-capping methyl groups at 770 cm 1 and the bands at 600 700 cm 1 that contain contributions from Si H, Si C, and Si Si vibrations. A comparison of these band intensities along a transect parallel to the direction of the pore-wall modification gradient shows a large change in the surface density of 10-hydroxydecyl groups, ranging from a level 8968
dx.doi.org/10.1021/la201272e |Langmuir 2011, 27, 8967–8973
Langmuir
Figure 2. ATR FTIR spectra of a 10-hydroxydecyl-gradient-modified rugate porous silicon sample prepared using 3,4,5-FBD, measured at different positions along the gradient direction. The sample has been methylated using a magnesium anode for 120 s at 9 mA cm 2 after the gradient-modification step.
of modification similar to that seen for samples prepared with a uniform modification via the diazonium salt/alkene reaction (Supporting Information) to no 10-hydroxydecyl addition over approximately 10 mm. Pore-wall modification gradients of decyl or 10-carboxydecyl groups on these porous silicon samples have also been characterized and show similar trends in surface composition across the porous silicon surface (Supporting Information). The intensities of the FTIR stretching bands noted above are constant along a transect perpendicular to the gradient direction for these samples. A small amount of the aromatic moiety from the diazonium salt is also incorporated during the hydrosilylation reaction on porous silicon, as indicated by the aromatic bands in the FTIR spectra around 1500 cm 1 (Figure 2) and an increased fluorine signal observed by SEM EDS (Figure 3). A decyl-modified porous silicon sample shows a 2:1 ratio of decyl groups to 3,4,5-trifluorobenzene groups as determined by the carbon and fluorine signals in SEM-EDS and by comparison of the relative sizes of the ATR FTIR bands observed for modified porous silicon samples to those observed for a dichloromethane acetonitrile solution containing a 1:1 mixture of decene and FBD (Supporting Information). On the basis of the ATRFTIR results, the decyl-modified porous silicon had a greater incorporation of 3,4,5-trifluorobenzene groups than the undecanoic acid-modified porous silicon samples. Thus, porous silicon layers modified using diazonium salt-initiated hydrosilylation with alkenes have a mixed surface of alkyl chains and aromatic rings, and this ratio can vary with the alkene used for the modification.
ARTICLE
Figure 3. Atomic compositions determined by SEM-EDS for rugate porous silicon samples modified uniformly with (a) decyl groups using diazonium salts, (b) decyl groups using diazonium salts and then endcapped with methyl groups, and (c) decyl groups using thermal hydrosilylation. The porous silicon layer thickness was approximately 10 μm, and measurements were made at five equally spaced depths within the porous silicon.
Figure 4. ATR FTIR spectra of (a) 10-carboxydecyl-modified porous silicon and (b) 10-carboxylatodecyl acid-modified porous silicon, prepared by deprotonation with ethanolic potassium hydroxide. Both samples are end-capped with methyl groups.
The deprotonation of 10-carboxydecyl acid groups by 17 mM ethanolic potassium hydroxide was performed to increase the hydrophilicity of the porous silicon surface, and the ATR-FTIR spectra observed before and after this step are shown in Figure 4. The FTIR band at 1716 cm 1 due to the CdO stretching mode almost completely disappears upon deprotonation, and 8969
dx.doi.org/10.1021/la201272e |Langmuir 2011, 27, 8967–8973
Langmuir new bands at 1558 and 1408 cm 1 associated with the asymmetric and symmetric stretching modes of the carboxylate anion are observed.33 A third band appears at around 1649 cm 1, and this has previously been assigned34 as the scissor mode of water molecules,35 suggesting that the deprotonated layer retains adsorbed water. The other FTIR bands associated with the methyl and 10-carboxydecyl acid groups are still present. The Si O stretching band around 1000 cm 1 has increased after deprotonation, indicating that a small amount of oxidation of the porous silicon layer has occurred.32 This may be due to the incomplete methyl end-capping throughout the 10 μm porous silicon layer. The uniformity of the pore-surface modification with depth into the porous silicon layer was investigated by measuring SEMEDS spectra within a 2 μm 11 μm region at different depths within the layer, with the smaller dimension being parallel to the normal through the layer. The 2 μm dimension is much larger than the repeat distance of the rugate porosity oscillation. We first investigated porous silicon samples that had been uniformly modified via either thermal or diazonium salt-initiated hydrosilylation for comparison with the gradient-modified samples discussed later. All uniformly modified porous silicon samples studied showed a decrease in the atom % of carbon from top to bottom within the porous silicon layer (Figure 3 and Supporting Information). This decrease within the layer was observed even when electrochemical methylation was performed for over 3 min with a magnesium anode following the initial porous silicon surface modification. A uniform decyl-modified porous silicon sample prepared by thermal hydrosilylation in the presence of Me3SiCl as a water scavenger also showed this trend in carbon level with depth into the porous silicon layer (Figure 3). This suggests either that there is a decrease in the mean porosity of the silicon with depth in the layer or that the reagents are not fully penetrating the bottom of the porous silicon layer or a combination of both of these factors. The integrated silicon areas in the EDS spectra show an increase with depth into the porous silicon layer, which indicates that a decrease in the porosity of the sample with depth is the dominant contribution to the changes in observed atomic composition. A uniformly thermally hydrosilylated porous silicon sample shows a higher coverage of decyl groups than a sample prepared by uniform hydrosilylation in the presence of diazonium salts, as indicated by the C content in the SEM-EDS analysis (Figure 3). This can be explained by the bulky diazonium salts present during the reaction of porous silicon with decene, reducing the surface density of surface-attached groups. However, methyl endcapping a decyl-modified porous silicon sample prepared using diazonium salts creates a surface with similar stability toward atmospheric oxidation as a decyl-modified porous silicon sample prepared by thermal hydrosilylation. This led to a similar limited extent of oxygen incorporation into the latter two porous silicon samples upon extended exposure to the laboratory atmosphere (Supporting Information). Thus, the level of oxidation of these two samples remained stable after a small initial increase whereas the decyl-modified porous silicon sample prepared using diazonium salts that had not been subsequently end-capped with methyl groups showed increasing oxidation over a 32 day period. Unmodified porous silicon also shows significant oxidation upon exposure to laboratory air over this time. The pore-wall compositional gradient porous silicon samples were then probed using SEM EDS. Measurements were made through the depth of the porous silicon layer along a transect across a sample that had been gradient-modified with
ARTICLE
Figure 5. SEM EDS results for a rugate porous silicon sample gradientmodified with 10-hydroxydecyl groups, end-capped with methyl groups, and then esterified with tridecafluoroheptanoyl chloride. The porous silicon layer thickness is approximately 10 μm, and the gradient was approximately 15 mm long. Measurements were made at three positions along the gradient (distances given in the legend, starting with the low amount of 10-hydroxydecyl modification) and at five positions through the depth of the porous silicon.
10-hydroxydecyl groups, end-capped with methyl groups, and then esterified with tridecafluoroheptanoyl chloride (Figure 5). The interpretation of the trends in these data was aided by comparison to the SEM-EDS analyses performed on uniformly modified porous silicon samples (Figure 3). Similar to the uniformly modified porous silicon samples, the gradient-modified sample showed a variation in the observed atom percentages of C, O, and F with depth into the layer. As noted earlier, this is most likely due to changes in porosity and pore accessibility with depth. For the gradient-modified sample at a given depth into the porous silicon, the C and F contents clearly show an increase along a transect parallel to the direction of the gradient (Figure 5), although the change observed (1.5- to 2-fold for F) is less than that observed in the ATR-FTIR spectra. There is a smaller change in atomic composition along the gradient direction at the outer surface of the porous silicon layer than through the remainder of the layer. Water droplets were placed upon the top surface of either 10-carboxylatodecyl or 10-hydroxydecyl to methyl pore-modification gradient porous silicon samples to demonstrate the difference in “wettability” across the samples (Figure 6). The samples showed a change in contact angle from the end with the highest concentration of methyl groups (75) to the end with the highest concentration of 10-carboxydecyl or 10-hydroxydecyl groups (60). This range of contact angles is smaller than is observed on flat surfaces and reflects the chemistry and morphology of the outer surface, rather than the internal pores. It is also consistent with the smaller changes in atomic content observed in the SEM-EDS analyses of the upper part of the gradient-modified porous silicon samples. An aqueous titration was performed to determine the effect of the pore-wall surface modification on wetting of the internal pores of the porous silicon layer. When a porous silicon rugate sample prepared from p-type silicon with a methyl-end-capped 10-hydroxydecyl pore-wall modification gradient was immersed 8970
dx.doi.org/10.1021/la201272e |Langmuir 2011, 27, 8967–8973
Langmuir
Figure 6. Images of water droplets placed along the direction of (a) a 10-carboxylatodecyl compositional gradient (droplets approximately 2.5, 6.5, and 10.5 mm along the gradient) and (b) a 10-hydroxydecyl compositional gradient (droplets approximately 4.0 and 12.5 mm along the gradient; arrows show the direction of each gradient from highest to lowest concentration of 10-carboxylatodecyl or 10-hydroxydecyl groups).
Figure 7. Optical photographs of a porous Si pore-wall modified with a gradient of 10-hydroxydecyl to methyl, immersed in water containing the indicated concentrations of ethanol. The gradient is horizontal in the images, with the maximum 10-hydroxydecyl surface concentration on the right-hand side. Stated values are the ethanol concentration (percent by volume).
in water, the water penetrated the pores only in a small region near the end with the highest concentration of 10-hydroxydecyl groups (Figure 7). As ethanol was added to the water, the region in which the pores were wetted by the solution increased, with complete wetting occurring only above 8 vol % ethanol. Thus, the porous silicon sample responds with a spatial color signal to a change in the ethanol concentration, a sensing modality that has not heretofore been reported for porous silicon photonic crystals. Such a spatial response means that the change can be detected by the human eye rather than requiring instrumentation. A similar methyl-end-capped 10-hydroxydecyl pore-wall modification gradient rugate porous silicon sample prepared from p++ silicon showed a response to ethanol over the range of 0 4% ethanol (Supporting Information). This difference in response range is due to the porous silicon prepared from p++ silicon having pores of larger average diameter. The visual response shown in Figure 7 requires that the liquid infiltrates a significant fraction of the entire porous layer. The extent of liquid infiltration is controlled by the surface tension of the liquid, the diameter of the pores, and the hydrophobicity of the pore walls. For a given surface chemistry, the lower the surface tension of the liquid, the smaller the pores that can be infiltrated.36 Porous Si
ARTICLE
samples prepared from p++ silicon possess both micropores (pore diameter
