596

IEEE TRANSACTIONS ON NANOTECHNOLOGY, VOL. 9, NO. 5, SEPTEMBER 2010

Size-Dependent Inltration and Optical Detection of Nucleic Acids in Nanoscale Pores

Jenifer L. Lawrie, Yang Jiao, and Sharon M. Weiss, Member, IEEE
AbstractExperiments and complimentary simulations are presented to demonstrate the size-dependent inltration and detection of variable length nucleic acids in porous silicon with controllable pore diameters in the range of 1560 nm. The pore diameter must be tuned according to target molecule size in order to most effectively balance sensitivity and size-exclusion. A quantitative relationship between pore size (1560 nm), nucleic acid length (up to 5.3 nm), and sensor response is presented with smaller molecules detected more sensitively in smaller pores as long as the pore diameter is sufcient to enable molecular inltration and binding in the pores. The density of probe molecules on the pore walls and subsequent hybridization efciency for target molecule binding are also reported and are shown to depend strongly on the method of inltration as well as the target molecule size. Index TermsDNA, optical sensor, pores, porous silicon, size effects, waveguide.

I. INTRODUCTION

varied. Previous reports have shown that small molecules can enter nanoscale porous silicon and bind to the pore walls on a timescale of seconds to minutes [13], [14]. Moreover, in one report, the reaction kinetics of enzymes in porous silicon were found to be consistent with the kinetics of the enzyme in free solution, suggesting that for molecules much smaller than the pore diameter, the pores do not inhibit molecule activity [14]. Initial studies also suggest that there is a pore size dependence to the sensitivity of detecting molecules within a porous material [15], [16]. Here, we quantitatively demonstrate how both the pore size and the molecule size impact the inltration and subsequent detection of molecules in a porous medium where the pore size is signicantly less than the interrogating wavelength of light. We also report, based on experiments and complimentary simulations, the surface coverage density of small molecules in nanoscale silicon pores under various inltration conditions.

NDERSTANDING the inltration of target species exposed to porous media is of great importance for a plethora of applications ranging from drug delivery and sensors to photovoltaics, fuel cells, and optical interconnects [1][10]. In most cases, the primary advantages of using porous materials are the large internal surface area and capability for size-selective ltering. For example, in medical applications, the large internal surface area of porous polymers has allowed controlled loading and subsequent release of drugs at well-regulated timescales [11]; the size selectivity of ultrathin porous membranes made from silicon has led to sharp molecular weight cutoffs of complex protein mixtures and the promise of faster, more efcient hemodialysis devices [12]. Extensive research is ongoing to develop new fabrication methods for porous materials, better understand mass transport in complicated pore geometries, and explore new properties and applications of porous materials. This paper focuses on the study of size-dependent molecular inltration in porous silicon pores for biosensing applications; both the pore size and the molecule size are systematically

II. EXPERIMENTAL PROCEDURE A. Nanoscale Porous Silicon Fabrication Nanoscale porous silicon samples with pore diameters in the range of 1560 nm were prepared by electrochemical etching in electrolytes containing hydrouoric acid (HF). Double layer porous silicon with nominal pore diameters of 20 nm in the top layer were formed on highly doped p-type (0.01 cm) silicon wafers, using an electrolyte containing 15% HF in ethanol. The double layers of porous silicon with distinct porosities were fabricated by applying consecutive current densities of 5 mA/cm2 (90 s) and 48 mA/cm2 (53 s), respectively. The porous silicon samples were subsequently exposed to a 9 mM solution of KOH in ethanol for 30 min to widen the pore opening. The resulting top porous silicon layer had a porosity of 76% and thickness of 230 nm, and the bottom layer had a porosity of 84% and a thickness of 1333 nm, as estimated based on MaxwellGarnett effective medium theory and SEM imaging. Single-layer porous silicon samples with average pore diameters of approximately 15, 30, or 60 nm, porosities of 30%, 35%, 57%, and thicknesses of 1.55 m were formed on highly doped n-type (0.01 cm) silicon wafers, using an electrolyte containing 5.5% HF in DI water and applied current densities of 8 mA/cm2 (180 s), 10 mA/cm2 (82 s), and 30 mA/cm2 (72 s), respectively. Following formation, these porous silicon layers were removed from the silicon substrate by applying a series of high current pulses [17]. SEM images of the porous silicon samples are shown in Fig. 1. Note that the distribution of pore diameters within a single-layer lm of porous silicon can be as high as several nanometers [18].

Manuscript received December 22, 2009; revised May 12, 2010; accepted June 18, 2010. Date of publication June 28, 2010; date of current version September 9, 2010. This work was supported in part by the Army Research Ofce under Grant W911NF-081-0200 and by the National Science Foundation under Grant ECCS0746296. The review of this paper was arranged by Associate Editor J. Rogers. The authors are with the Department of Electrical Engineering and Computer Science and the Interdisciplinary Graduate Program in Materials Science, Vanderbilt University, Nashville, TN 37235 USA (e-mail: jenifer.l.lawrie@ vanderbilt.edu; yang.jiao@vanderbilt.edu; sharon.weiss@vanderbilt.edu). Color versions of one or more of the gures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identier 10.1109/TNANO.2010.2055580

1536-125X/$26.00 2010 IEEE

LAWRIE et al.: SIZE-DEPENDENT INFILTRATION AND OPTICAL DETECTION OF NUCLEIC ACIDS IN NANOSCALE PORES

597

propriate surface chemistry for the immobilization of thiolmodied presynthesized DNA oligonucleotides. The DNA molecules were attached in both types of porous silicon samples. DNA oligonucleotides of length up to 24 bases were grown by a process of in situ synthesis inside the double-layer porous silicon samples, using sequential phosphoramidite reactions [19]. This base-by-base assembly of the oligonucleotides allowed for exibility in dening both the sequence and length of the DNA molecules. For the single-layer porous silicon samples, the cross-linker Sulfo-SMCC (Pierce) in water and ethanol was attached to the aminosilane monolayer in the pores to provide the necessary maleimide derivatization for subsequent immobilization of thiol-tagged DNA oligonucleotides (Eurons MWG Operon) [16]. The DNA oligos in HEPES buffer were mixed 1:1 by volume with TCEP (Pierce) in water and ethanol [16], and directly inltrated into the maleimide derivatized single-layer porous silicon samples. Different length of DNA oligos between 8 and 24 bases were obtained for the single-layer porous silicon experiments. In this paper, complementary, uncharged peptide nucleic acid (PNA) molecules (BioSynthesis, Inc.) were utilized for hybridization experiments in order to eliminate the porous silicon corrosion that has been associated with hybridizing negatively charged DNA molecules [20], [21]. Target PNA solutions were prepared in DI water or HEPES buffer, and hybridization experiments were performed by incubation of the functionalized waveguides in the target PNA solution at room temperature for 1 h. C. Optical Measurement Technique
Fig. 1. (Top) Cross-sectional SEM image showing the morphology of n-type porous silicon layers produced by different current densities (from top to bottom: 30 mA/cm2 , 10 mA/cm2 , and 8 mA/cm2 ). As the current density decreases from 30 to 8 mA/cm2 , the pore diameter decreases from approximately 60 to 15 nm. We note that the three layers are shown together for ease of comparison; experiments utilized only single-layer membranes. Insets with 200 nm dimensions illustrate the nominal 60, 30, and 15 nm pores. (Bottom) Top-view SEM image showing the surface morphology of p-type porous silicon. Pore openings are approximately 20 nm in diameter.

B. Molecular Inltration and Attachment The double-layer and single-layer porous silicon samples were thermally oxidized in air at 800 C and 500 C, respectively, for several minutes in order to form a thin, hydrophilic oxide layer on all surfaces of the porous silicon [17], [19]. Silane chemistry was utilized to provide a link between the inorganic oxidized silicon surface and organic molecules. In the case of the porous silicon double layers, a 4% solution of N-(3-triethoxysilylpropyl)-4-hydroxy-butyramide TEOS-HBA, Gelest) in ethanol and DI water was utilized, as previously reported [19]. The TEOS-HBA silane provides the hydroxyl functional group necessary for subsequent attachment of DNA via the in situ phosphoramidite synthesis. For the single-layer samples, a solution of 4% 3-aminopropyltriethoxysilane (3-APTES, Aldrich) in methanol and DI water was utilized, as previously reported [17]. The APTES functionalization provides the ap-

In order to facilitate accurate detection of molecules inltrating the pores, the design of the two types of porous silicon samples was such that the characteristics of the porous silicon upon exposure to variable sized molecules could be monitored based on optical waveguides analysis. Details regarding the method for designing resonant porous silicon waveguides can be found in [17] and [22]. Resonant structures allow for stronger electric eld connement and sharper spectral features, which lead to more accurate detection of small quantities of small molecules. As shown schematically in Fig. 2 (inset), the porous silicon double layer acts as a waveguide by conning light in the upper porous silicon layer; the bottom porous silicon layer with lower refractive index and air gap above the upper porous silicon layer act as waveguide cladding layers [19], [21], [23]. In this conguration, light is concentrated in the porous silicon layer, where molecules will rst inltrate, increasing the sensitivity of detecting the presence of these molecules. The conguration for the single-layer porous silicon waveguide, as shown in Fig. 4 (inset), is similar; cladding layers are air and a low-refractive-index polymer layer on either side of the porous silicon membrane [17], [24]. Light from a 1550-nm diode laser is coupled into the waveguides at discrete angles using a cubic zirconia prism. In order for light to be coupled into the waveguides, the horizontal component of the wave vector in the prism must match the wave vector of a guided mode in the porous silicon waveguide [25]. The wave vector in the prism is changed by simply varying the

598

IEEE TRANSACTIONS ON NANOTECHNOLOGY, VOL. 9, NO. 5, SEPTEMBER 2010

Fig. 2. Attenuated total reectance spectra of the porous silicon two-layer waveguide after oxidation (left), inltration of TEOS-HBA silane (middle), and in situ synthesis of 16mer DNA (right). The resonance shifts to higher angle as layers of biomolecules are attached to the internal pore surfaces of the waveguide. Solid lines indicate experimental results, and dashed lines correspond to calculations that assume 70% silane coverage and 40% DNA coverage in the pores. The inset shows a schematic of the two-layer waveguide and measurement setup.

angle of incidence light. The resolution of the Metricon 2010 prism coupler used in this study is 0.002 . The guided mode wave vector is primarily determined by the refractive index of the waveguide. Hence, as the refractive index of the porous silicon waveguide changes when molecules are inltrated, the wave vector of the guided mode is modied, and the incident angle of light in the prism must be changed, accordingly, to allow light to couple into the waveguide. The presence of molecules in the pores is therefore detected by monitoring the coupling angle of light into the waveguide or, equivalently, monitoring the amplitude of light exiting the prism at a xed angle. Note that the refractive index of the porous silicon waveguide changes upon molecular inltration because porous silicon acts as an effective medium with pore diameters that are much smaller than the wavelength of light [26]; before molecular inltration, the refractive index of the waveguide is a weighted average of the refractive indexes of silicon, air, and oxide, while the refractive index of the waveguide after inltration is a weighted average of the indexes of silicon, air, oxide, and the molecules. III. SIZE-DEPENDENT MOLECULAR INFILTRATION In order to investigate size-dependent molecular inltration in 1560 nm pores, experiments and complimentary modeling were performed to ascertain the size-dependent surface density of molecules inltrated into double- and single-layer porous silicon waveguides. Fig. 2 shows the change in the resonant coupling angle of the double-layer waveguide upon inltration of TEOS-HBA silane and 16mer in situ synthesized DNA probe molecules. The magnitude of the resonance shifts directly corresponds to the quantity of each molecule immobilized in the pores. The TEOS-HBA silane molecule is approximately 0.9 nm in length [27]. The 16mer DNA molecule is approximately 3.52 nm long [28]. As expected, inltra-

Fig. 3. Resonance shift upon inltration of probe DNA into two-layer porous silicon waveguides. Black square data points give the shift due to inltration and immobilization of presynthesized DNA probes, ranging from 1.76 to 5.28 nm length [16]. Immobilization chemistry of the presynthesized DNA is described in the single-layer system, utilizing APTES and Sulfo-SMCC. Red circular data points give the shift observed when probe DNA of the given length is synthesized base by base in the waveguide using the in situ method. The x-axis corresponds to the number of nucleotide bases in the oligo sequence. Each base has a nominal length of 0.22 nm. Data for the in situ synthesis contains a correction factor that reduces the measured shift by 0.012 to account for the presence of a DMT protecting group, which does not appear on the inltrated probe.

tion of the silane molecules gives a smaller resonance shift (0.680 ) compared to the subsequent synthesis of the larger 16mer probe molecules (1.195 ). Transfer matrix theory [17] and Lugos three component effective medium model [29] were used to determine the percent surface coverage of each type of molecule by tting the model to the experimentally derived magnitude of the respective resonance shifts. Details of this tting procedure are described elsewhere [17]. Complete monolayer coverage of TEOS-HBA silane would yield refractive index changes of ncore = 0.042 (upper porous silicon layer) and ncladding = 0.033 (lower porous silicon layer), while the measured resonance shift after silane inltration corresponds to ncore = 0.028 and ncladding = 0.022, suggesting nearly 70% surface coverage of the silane molecules. For the 16mer probe DNA, 100% surface coverage is calculated to give refractive index changes of ncore = 0.151 and ncladding = 0.125, while the measured resonance shift corresponds to ncore = 0.056 and ncladding = 0.046, suggesting approximately 40% surface coverage of the 16mer DNA molecules. The simulated resonances for 70% silane coverage and 40% DNA coverage in the pores are shown in Fig. 2. The discrepancy in resonance depth is due, in part, to neglecting scattering losses in the simulations. Porous silicon nonuniformity within the measurement spot likely also contributes to the broadening of reectance spectra [26]. Scattering losses, and thus signal resolution, may be improved through reduction of roughness on the porous silicon surface. Low-temperature etching of porous silicon has been shown to reduce porous silicon surface roughness [30], [31]. Fig. 3 shows the evolution of the two-layer waveguide resonance shift during in situ synthesis of DNA molecules up to

LAWRIE et al.: SIZE-DEPENDENT INFILTRATION AND OPTICAL DETECTION OF NUCLEIC ACIDS IN NANOSCALE PORES

599

24 bases in length. The data points in the gure are a concatenation of measurements from several different samples. Hence, the somewhat abrupt slope changes may be due, in part, to sampleto-sample variation. Nevertheless, the overall trend is clear that the resonant coupling angle progressively increases as additional bases are introduced into the porous silicon waveguides. Based on the magnitude of the resonance shifts, we estimate approximately 50% surface coverage for 8mer in situ synthesized DNA molecules, 40% surface coverage for 16mer molecules, and 35% coverage for 24mer molecules. Hence, because the size of the protected, single-DNA nucleotides is small enough to be unhindered by size of the pore opening, the surface coverage of the in situ assembled DNA molecules with up to 24 bases is only slightly dependent on the nal DNA molecule size. We note that synthesis of larger DNA molecules has not yet been investigated, but it is expected that such molecules could be assembled in the pores. The size dependence of the surface coverage of in situ synthesized DNA molecules is in direct contrast to the trend reported previously for inltration of presynthesized DNA molecules into porous silicon waveguides [16]. As reproduced in Fig. 3, the percent surface coverage falls off sharply for DNA molecules composed of more than 16 bases. Moreover, the shift due to 8mer DNA molecules when inltrated as 8-base molecules instead of single base molecules that assemble in the pores to become 8mer molecules is only 10%. The impact of the larger surface coverage for in situ synthesized probe DNA molecules for sensing applications is reported in Section IV. The tunable pore diameter of the single-layer porous silicon waveguides offers a convenient platform for further investigation of the size-dependent inltration of molecules into porous media. For sensing applications, it is important to understand how different sized molecules inltrate and attach within various sized pores. For example, even when in situ synthesized probe molecules are utilized, the inltration efciency of fullsized target molecules must be considered. The data shown in Fig. 3 suggest that although 24mer DNA probes can be assembled efciently in 20 nm pores, the detection of complementary 24mer DNA target molecules would be challenging. Fig. 4 shows the change in the resonant coupling angle of a single-layer porous silicon waveguide with 60-nm-diameter pores upon inltration of 3-APTES (1.15 ), Sulfo-SMCC (1.73 ), and 10-M 16mer thiolated DNA probe (0.55 ) molecules. The 3-APTES molecules and Sulfo-SMCC molecules have been shown to form 0.8 and 1.16 nm monolayers on silica surfaces [17], [32]. We note that the magnitude of the resonance shifts for the singlelayer waveguides cannot be directly compared to those of the two-layer waveguides since the sensitivity of porous silicon waveguides has been shown to depend on initial resonance angle; higher resonance angles yield improved detection sensitivity [17]. By tting the experimental resonance shifts with calculations based on transfer matrix theory and Lugos effective medium model as done previously with the two-layer waveguides [17], [29], the refractive index changes due to 3APTES, Sulfo-SMCC, and 16mer probe DNA were found to be 0.016, 0.025, and 0.011, respectively. These refractive index changes correspond to surface coverages near 100% for both the 3-APTES and Sulfo-SMCC molecules, and a surface coverage

Fig. 4. Attenuated total reectance spectra of a porous silicon single-layer waveguide with 60 nm pores after (from left to right) oxidation, inltration of 3-APTES silane, Sulfo-SMCC, and 10-M 16mer DNA. Solid lines indicate experimental results, and dashed lines correspond to calculations that assume 100% silane coverage, 100% Sulfo-SMCC coverage, and 25% DNA coverage in the pores. The inset shows a schematic of the single-layer waveguide and measurement setup.

of approximately 25% for the larger 16mer DNA molecules. The simulated resonances for 100% silane and Sulfo-SMCC coverage and 25% DNA coverage in the pores are shown in Fig. 4. The larger width of the experimental resonance dips can be attributed in large part to scattering losses due to the roughened porous silicon surface created by the series of high current pulses used to remove the porous silicon lm from the substrate. Scattering losses were not included in the calculation. Sample nonuniformity within the measurement spot also likely contributed to the resonance broadening. Although the pores size of the single-layer waveguide for this experiment was three times that of the two-layer waveguide investigated in Fig. 2, the probe DNA coverage was signicantly lower due to the method of inltration. Hence, there is a clear size-dependent inltration and surface coverage of molecules in porous silicon. Fig. 5 illustrates the trends of size-dependent inltration in porous silicon single-layer waveguides with pore diameters of approximately 15, 30, and 60 nm. Each sample was inltrated with 100 M of 8mer, 16mer, or 24mer DNA. Note that for the various pore sizes presented in Fig. 5, we can compare the magnitude of the resonance shifts since all single-layer waveguides were designed with similar initial resonance angle. The largest magnitude resonance shift occurs for the largest DNA molecule (24mer, 5.28 nm) inltrated into the largest pores (60 nm). In this case, there is no inhibition of molecule inltration and the larger DNA molecules induce a relatively larger refractive index change and resonance shift. In contrast, there is essentially no resonance shift induced from the 24mer DNA exposure to 15 nm pores, and it can be concluded that the pore size is too small compared to the molecule size to allow for effective inltration. For the smaller DNA molecules, the most appropriate size of pore diameter depends on the application. While the 60 nm pores give rise to a slightly larger resonance shift compared to the 30 nm pores upon inltration of 16mer DNA (3.52 nm),

600

IEEE TRANSACTIONS ON NANOTECHNOLOGY, VOL. 9, NO. 5, SEPTEMBER 2010

Fig. 6. Resonance shifts upon hybridization with 100% complementary PNA sequences for 8mer (left) and 16mer (right) DNA probes. Hybridization to the 8mer probe produces a shift of 0.194 , while hybridization to the 16mer probe results in a shift of 0.122 . The magnitude of the resonance dip is independent of the effective refractive index of the waveguide, relating only to the coupling efciency into the waveguide.

Fig. 5. Resonance angle shifts of single-layer porous silicon waveguide with different pore sizes for detection of 100-M DNA molecules of different lengths (2.2 A/base). For optimal performance as a sensor, the pore size must be appropriately tuned based on the size of the molecule of interest.

the 30 nm pores would enable enhanced size exclusion capabilities for sensing in complex media. When examining inltration of the smallest DNA molecules in this study (8mer, 1.76 nm), it can be seen that the smallest pores (15 nm) yield a slightly larger resonance shift compared to the larger pore diameters, likely due to their increased surface area. Importantly, it should be noted that this sensitivity enhancement of the smallest pores would increase for the detection of targets smaller than 8mer DNA [33]. For example, in aptamer-based detection schemes, DNA can be used as a probe molecule for the detection of small toxin or drug molecules that are considerably smaller than the probe DNA [34]. The limitations on target and probe inltration could vary signicantly in such a system and should be examined separately. IV. SIZE-DEPENDENT MOLECULAR DETECTION Nucleic acid probes in biosensors have applications ranging from simple DNA complement sensing for genetic analysis to use as aptamers for the detection of a wide variety of biochemical species. These applications, however, depend upon nding a balance between the inherent size-exclusion properties of the pores and the ability of target molecules to lter into the pores and interact with probes on the pore walls. In order to carefully examine the effect of pore diameter and probe density on target detection, single- and double-layer waveguide sensors were exposed to 10 M solutions of a 100% complementary PNA target molecule. The waveguides were incubated in the PNA solution for 1 h, followed by a 20 min. soak in DI water. After washing the sample in DI water, the resonance angle was measured. Fig. 6 shows the resonance shifts due to hybridization of 8mer and 16mer PNA in double-layer porous silicon waveguides (20-nm pore diameter) with in situ synthesized 8mer and 16mer DNA probes that occupy 40%50% of the pore surface. If all of the probe molecules capture a target PNA molecule, then a resonance shift of nearly equal magnitude to that shown by the waveguide after probe DNA attachment would be observed.

Fig. 7. Hybridization efciencies of various sensing systems toward 10-M PNA solution: (a) 8mer target exposed to relatively high probe density in 20 nm pores of two-layer waveguide, (b) 16mer target exposed to relatively high probe density in 20 nm pores of two-layer waveguide, (c) 16mer target exposed to relatively low probe density in 15 nm pores of single-layer waveguide, and (d) 16mer target exposed to relatively low probe density in 60 nm pores of single-layer waveguide.

However, given the size-dependent inltration data presented in Section III, we do not expect nor do we observe a comparable magnitude shift. Following a trend similar to that exhibited by the 15 nm pores in Fig. 5, the shift due to hybridization of 16mer PNA is less than that due to hybridization of 8mer PNA, owing to molecule size-dependent inltration challenges. While the in situ synthesis method allowed the 16mer probe to be synthesized in the pore with very little hindrance, the full-size 16mer target molecule inltrated into the pores with a much lower efciency. Fig. 7 presents hybridization efciency data for both the twolayer porous silicon waveguide and single-layer waveguide to shed light on the role of size dependence in target molecule detection. Here, hybridization efciency is dened as the ratio of the waveguide resonance shift after PNA attachment to the resonance shift after probe immobilization. For example, based on the resonance shifts, as shown in Fig. 6, due to target attachment

LAWRIE et al.: SIZE-DEPENDENT INFILTRATION AND OPTICAL DETECTION OF NUCLEIC ACIDS IN NANOSCALE PORES

601

and the in situ synthesized probe DNA shifts, as shown in Fig. 3, we nd that the hybridization efciency for the 8mer molecules is approximately 30% while that for 16mer molecules is approximately 15%. Due to sample-to-sample variations, we estimate that these hybridization efciencies are accurate to within 10%. In a similar manner, hybridization experiments show that the hybridization efciency of 16mer PNA molecules in 15 and 60 nm pores is approximately 50% and 70%, respectively, in single-layer porous silicon waveguides. There are three primary conclusions that can be drawn from the hybridization experiments. First, when comparing hybridization efciency in pores with similar diameters (20-nm double-layer and 15-nm single-layer waveguides), the pores with lower probe density experience larger hybridization efciency. This trend is consistent with prior reports on at surfaces, where hybridization efciencies have been observed as low as 10% for high-density DNA probe coverage of approximately 1.2 1013 molecules/cm2 and 50% for low-density DNA probe coverage of approximately 3 1012 molecules/cm2 [35]. By comparison, the DNA probe densities in porous silicon have been reported as three orders of magnitude higher per unit surface area [36], [37]. However, given the large surface area to volume ratios associated with porous silicon, the DNA probe densities on the internal pore wall surfaces of the waveguides are still relatively low. Second, unlike planar surfaces on which consistent hybridization efciencies are achieved as long as the probe density remains constant (regardless of the method utilized to obtain the particular probe density) [35], when comparing pores with similar probe density but different pore size, we nd that hybridization efciency is greater in the larger pores. We attribute this increased hybridization efciency in part to the extra space available for the target molecule to move within the pores. Third, although the hybridization efciency is signicantly lower in the waveguides with high probe density, the overall resonance shift in response to target inltration is larger. Note that for this comparison, the detection sensitivity of the waveguides at different initial resonance angles must be considered. We further note that nanomolar level detection sensitivities have been shown in both single- and two-layer porous silicon waveguides [21], [24]. Hence, when considering the most appropriate pore diameter and probe coverage for sensors utilizing porous materials, a modest probe coverage and slight compromise in hybridization efciency along with sufcient pore diameter to enable molecular inltration and binding is likely to yield the highest detection sensitivity. V. CONCLUSION There are signicant advantages in using porous materials as a platform for signal transduction in biosensing devices. The benets of size-exclusion and large internal surface area, however, may be offset by challenges in molecule inltration into the pores and, as a result, poor utilization of the internal surface area. We report a clear size-dependent inltration and surface coverage of molecules in porous silicon waveguides. Incremental increases in pore diameter allow for selectivity toward target molecules of a particular length, as demonstrated in porous

silicon waveguides with controllable pore diameters between 15 and 60 nm that are exposed to nucleic acids of length between approximately 1.76 and 5.28 nm. The pore diameter may be selected to allow the inltration of a molecule of interest, while excluding slightly larger contaminating molecules. Different probe attachment techniques were shown to offer tunable surface coverage of molecules. In situ synthesis of nucleic acids yields a surface coverage of 40% for sequences up to 24 bases, while direct inltration of the nucleic acid sequences yields a surface coverage of only 25%. While the hybridization efciency for complementary nucleic acids is compromised in the pores with higher molecular surface coverage, the overall response of these sensors remains at nanomolar detection levels. Moreover, the in situ synthesis method could provide distinct advantages for sensors using probe DNA as aptamers where the target molecule is of smaller size. This work provides a model system for balancing size-exclusion capabilities, ease of inltration, and sensitivity that may be utilized in applications beyond oligonucleotide detection, such as small molecule detection or aptamer-based sensing. By appropriately controlling both pore size and surface coverage, porous sensors can be optimally tuned for best sensing performance toward a wide size range of molecules. ACKNOWLEDGMENT The authors would like to thank J. Ryckman, Z. Xu, and P. Laibinis at Vanderbilt University for technical assistance and useful discussions. Resources in the Vanderbilt Institute of Nanoscale Science and Engineering were utilized in this paper. REFERENCES
[1] M. Vallet-Regi, L. Ruiz-Gonz lez, I. Izquierdo-Barba, and J. M. Gonz leza a Calbet, Revisiting silica based ordered mesoporous materials: Medical applications, J. Mater. Chem., vol. 16, pp. 2631, Oct. 2005. [2] Y. Y. Li, F. Cunin, J. R. Link, T. Gao, R. E. Betts, S. H. Reiver, V. Chin, S. N. Bhatia, and M. J. Sailor, Polymer replicas of photonic porous silicon for sensing and drug delivery applications, Science, vol. 28, pp. 2045 2047, Mar. 2003. [3] H. Ko, S. Chang, and V. V. Tsukruk, Porous substrates for label-free molecular level detection of nonresistant organic molecules, ASC Nano, vol. 3, pp. 181188, Dec. 2008. [4] J. Zhang, J. Dong, M. Luo, H. Xiao, S. Murad, and R. A. Normann, Zeolite-ber integrated optical chemical sensors for detection of dissolved organics in water, Langmuir, vol. 21, pp. 86098612, Aug. 2005. [5] M. P. Stewart and J. M. Buriak, Chemical and biological applications of porous silicon technology, Adv. Mater., vol. 12, pp. 859869, Jun. 2000. [6] V. Y. Yerokhov and I. I. Melnyk, Porous silicon in solar cell structures: A review of achievements and modern directions of further use, Renew. Sustain. Energy Rev., vol. 3, pp. 291322, Dec. 1999. [7] P. N. Ciesielski, A. M. Scott, C. J. Faulkner, B. J. Berron, D. E. Cliffel, and G. K. Jennings, Functionalized nanoporous gold leaf electrode lms for the immobilization of photosystem 1, ACS Nano, vol. 2, pp. 24652472, Nov. 2008. [8] D. R. Rolison, Catalytic nanoarchitecturesthe importance of nothing and the unimportance of periodicity, Science, vol. 299, pp. 16981701, Mar. 2003. [9] A. Jain, S. Rogojevic, S. Ponoth, N. Agarwal, I. Matthew, W. N. Gill, P. Persans, M. Tomozawa, J. L. Plawsky, and E. Simonyi, Porous silica materials as low-k dielectrics for electronic and optical interconnects, Thin Solid Films, vol. 398/399, pp. 513522, Nov. 2001. [10] S. M. Weiss, M. Haurylau, and P. M. Fauchet, Tunable photonic bandgap structures for optical interconnects, Opt. Mater., vol. 27, pp. 740744, Feb. 2005.

602

IEEE TRANSACTIONS ON NANOTECHNOLOGY, VOL. 9, NO. 5, SEPTEMBER 2010

[11] J. Kost and R. Langer, Responsive polymeric delivery systems, Adv. Drug Del. Rev., vol. 46, pp. 125148, Mar. 2001. [12] C. C. Striemer, T. R. Gaborski, J. L. McGrath, and P. M. Fauchet, Chargeand size-based separation of macromolecules using ultrathin silicon membranes, Nature, vol. 445, pp. 749753, Feb. 2007. [13] J. D. Ryckman, M. Liscindi, J. E. Sipe, and S. M. Weiss, Porous silicon structures for low-cost diffraction-based biosensing, Appl. Phys. Lett., vol. 96, pp. 171103-1171103-3, Apr. 2010. [14] M. M. Orosco, C. Pacholski, and M. J. Sailor, Real-time monitoring of enzyme activity in a mesoporous silicon double layer, Nature Nanotechnol., vol. 4, pp. 255258, Apr. 2009. [15] S. C. B. Gopinath, K. Awazu, M. Fujimaki, K. Sugimoto, Y. Ohki, T. Komatsubara, J. Tominaga, K. C. Gupta, and P. K. R. Kumar, Inuence of nanometric holes on the sensitivity of a waveguide-mode sensor: Labelfree nanosensor for the analysis of RNA aptamerligand interactions, Anal. Chem., vol. 80, pp. 66026609, Sep. 2008. [16] G. Rong and S. M. Weiss, Biomolecule size-dependent sensitivity of porous silicon sensors, Phys. Stat. Sol. A, vol. 206, pp. 13651368, Jun. 2009. [17] Y. Jiao and S. M. Weiss, Design parameters and sensitivity analysis of polymer-cladded porous silicon waveguides for small molecule detection, Biosens. Bioelectron., vol. 25, pp. 15361539, Feb. 2010. [18] H. Ouyang, M. Christophersen, R. Viard, B. L. Miller, and P. M. Fauchet, Macroporous silicon microcavities for macromolecule detection, Adv. Funct. Mater., vol. 15, pp. 18511859, Sep. 2005. [19] J. L. Lawrie, Z. Xu, G. Rong, P. E. Laibinis, and S. M. Weiss, Synthesis of DNA oligonucleotides in mesoporous silicon, Phys. Stat. Sol. A, vol. 206, pp. 13391342, Jun. 2009. [20] C. Steinem, A. Janshoff, V. S.-Y. Lin, N. H. Volcker, and M. Reza Ghadiri, DNA hybridization-enhanced porous silicon corrosion: Mechanistic investigations and prospect for optical interferometric biosensing, Tetrahedron, vol. 60, pp. 1125911267, Nov. 2004. [21] J. L. Lawrie, Z. Xu, P. E. Laibinis, and S. M. Weiss, An aptamer-based approach to label-free biosensors in porous silicon waveguides, presented at the Chem. Biol. Defense Sci. Technol. Conf., Dallas, Texas, 2009. [22] J. Saarinen, S. M. Weiss, P. M. Fauchet, and J. E. Sipe, Optical sensor based on resonant porous silicon structures, Opt. Exp., vol. 13, pp. 3754 3764, May 2005. [23] G. Rong, A. Najmaie, J. E. Sipe, and S. M. Weiss, Nanoscale porous silicon waveguides for label-free DNA sensing, Biosens. Bioelectron., vol. 23, pp. 15721576, Jan. 2008. [24] G. Rong, J. D. Ryckman, R. Mernaugh, and S. M. Weiss, Label-free porous silicon membrane waveguide for DNA sensing, Appl. Phys. Lett., vol. 93, pp. 161109-1161109-3, Oct. 2008. [25] P. K. Tien and R. Ulrich, Theory of prism-lm coupler and thin-lm light guides, J. Opt. Soc., vol. 60, pp. 13251337, Oct. 1970.

[26] J. J. Saarinen, S. M. Weiss, P. M. Fauchet, and J. E. Sipe, Reectance analysis of a multilayer 1-D porous silicon structure: theory and experiment, J. Appl. Phys., vol. 104, pp. 013103-1013103-7, Jul. 2008. [27] B. Y. Chow, D. W. Mosley, and J. M. Jacobson, Perfecting imperfect monolayers: Removal of siloxane multilayers by CO2 snow treatment, Langmuir, vol. 21, pp. 47824785, Apr. 2005. [28] D. Rekesh, Y. Lyubchenko, L. S. Shlyakhtenko, and S. M. Lindsay, Scanning tunneling microscopy of mercapto-hexyl-oligonucleotides attached to gold, Biophys. J., vol. 71, pp. 10791086, Aug. 1996. [29] J. E. Lugo, H. A. Lopez, S. Chan, and P. M. Fauchet, Porous silicon multilayer structures: A photonic band gap analysis, J. Appl. Phys., vol. 91, pp. 49664972, Apr. 2002. [30] S. Setzu, G. L rondel, and R. Romestain, Temperature effect on the e roughness of the formation interface of p-type porous silicon, J. Appl. Phys., vol. 84, pp. 31293133, Sep. 1998. [31] G. L rondel and R. Romestain, Reection and light scattering in porous e silicon, in Properties of Porous Silicon, L. T. Canham, Ed. London, UK: INSPEC, 1997, pp. 241246. [32] B. H. Clare and N. L. Abbott, Orientations of nematic liquid cyrstals on surface presenting controlled densities of peptids: Amplication of protein-peptide binding events, Langmuir, vol. 21, pp. 64516461, Feb. 2005. [33] L. M. Bonanno and L. A. DeLouise, Tunable detection sensitivity of opiates in urine via a label-free porous silicon competitive inhibition immunosensor, Anal. Chem., vol. 82, pp. 714722, Jan. 2010. [34] S. P. Song, L. H. Wang, J. Li, J. L. Zhao, and C. H. Fan, Aptamer-based biosensors, Trends Anal. Chem., vol. 27, pp. 108117, Feb. 2008. [35] A. W. Peterson, R. J. Heaton, and R. M. Georgiadis, The effect of surface probe density on DNA hybridization, Nucleic Acids Res, vol. 29, pp. 51635168, Dec. 2001. [36] N. H. Voelcker, I. Alfonso, and M. R. Ghadiri, Catalyzed oxidative corrosion of porous silicon used as an optical transducer for ligand receptor interactions, ChemBioChem, vol. 9, pp. 17761786, Jul. 2008. [37] I. Rea, G. Oliviero, J. Amato, N. Borbone, G. Piccialli, I. Rendina, and L. De Stefano, Direct synthesis of oligonucleotides on nanostructured silica multilayers, J. Phys. Chem. C, vol. 114, pp. 26172621, Feb. 2010.

Authors photographs and biographies not available at the time of publication.