Patent Application
20160075883
Interest in superhydrophobic surfaces (defined as having a water contact angle (CA) greater than 150° and contact angle hysteresis (CAH) less than 10°) has grown rapidly in recent years due to unique characteristics such as self-cleaning, antifouling and fluid drag reduction. However, for applications such as self-cleaning windows, optical devices, and solar panels, high optical transparency is additionally required, as well as resistance to mechanical wear. For example, typical requirements for an automotive windshield are visible transmittance>90%, haze<1%, 10,000 cycles of wiper sliding, and 250 car wash cycles, the last two requirements representing 10 years of life.
A maximum CA of about 120° can be obtained for a nominally flat surface with a low surface energy coating. In order to produce a superhydrophobic surface, roughness is required. On a rough surface, a deposited water droplet will reside in either the Wenzel or Cassie-Baxter wetting regime. In the Wenzel regime, water fully penetrates roughness features, creating a continuous liquid-solid contact. In the Cassie-Baxter regime, the droplet rests on the peaks of roughness features, with air pockets filling the gaps in between. This air pocket formation leads to low CAH for self-cleaning ability, and thus in addition to roughness, a high liquid-air fractional area (fLA) is important for superhydrophobic surfaces when self-cleaning is desired. However, the dual requirements of superhydrophobicity and transparency pose a challenge. The surface must be sufficiently rough to obtain high CA and low CAH, but the dimensions of the roughness features must be small enough to preserve high transmittance of light. It is usually suggested that the size of surface features should not exceed roughly one quarter of the wavelength of visible light (around 100 nm or less).
While glass is the most common optical material for lenses, architectural windows, etc., transparent polymers such as polycarbonate (PC) and poly(methyl methacrylate) (PMMA) are also of great engineering importance. PC and PMMA are used for wide-ranging applications such as aircraft canopies, bullet-proof windows, solar cell panels, laptop computer screens, and many high-performance optical, electronic and medical devices. SiO2, ZnO, and ITO (indium tin oxide) thin films are of interest for varying applications. These three metal oxides have high transmittance for visible light due to low refractive indices (minimizing reflectance) and band gap wavelengths shorter than the visible range of 400-700 nm (minimizing visible-range absorption). SiO2 in particular has extremely high visible transmittance. ZnO thin films can have a UV-protective effect, and have been shown to reduce photodegradation of PC. When doped with other metals such as AI or Ga, ZnO can also be used for transparent conducting films. ITO is the most commonly used material for transparent conducting films due to its combination of high visible transmittance and low electrical resistivity. In addition, these particles have high hardness. Thus, SiO2, ZnO, and ITO nanoparticles would seem to be suitable candidates for wear-resistant, transparent, superhydrophobic surfaces.
In the case of glass, nanostructuring has generally been achieved through dip coating or spin coating of nanoparticles. For polymers, plasma etching techniques have also commonly been used. Several studies have reported optical transmittance approaching 100%, with a few even reporting enhanced transmittance compared to the uncoated substrate due to an antireflective effect. However, many of the so-created surfaces required post-fabrication treatment with fluorosilane or other low surface energy substance to achieve superhydrophobicity. In some cases, CAH and/or tilt angle (TA), which are important for self-cleaning ability, are not reported. In many cases, mechanical wear experiments are either absent or lack quantitativeness. Relatively few studies have used polymer substrates as compared to glass. Notably, SiO2 nanoparticles have been the overwhelming favorite to provide a nanostructure, while studies using other particles such as ZnO have been less common. The use of ITO nanoparticles to create superhydrophobic surfaces has not been found in the literature. In order to capitalize on the unique properties that different nanoparticles offer, as well as expand potential applications, a need exists for fabrication techniques that are suitable for a variety of nanoparticles and optical substrates.
Aspects of the present disclosure relate to a method for fabricating a superhydrophobic, optically transparent surface on a substrate. In some embodiments, a dip coating technique is disclosed in which a solution comprising hydrophobic nanoparticles, a resin binder and a solvent is provided. The substrate is dipped and then withdrawn from the solution. As the substrate is withdrawn, a precursor coating of the solution is formed on a surface of the substrate. The solvent in the precursor coating is allowed to evaporate (or is otherwise removed), immediately resulting in a superhydrophobic, optically transparent coating on the substrate surface.
In some embodiments, methods and solutions of the present disclosure are useful for creating a superhydrophobic, optically transparent surface or coating on multiple different substrate types including glass and polymer substrates such as polycarbonate (PC) and polymethyl methacrylate (PMMA). The hydrophobic nanoparticles can be metal oxide nanoparticles (such as SiO2, ZnO, and ITO) that are surface functionalized to be hydrophobic. Optional solvents useful with the methods and solutions of the present disclosure include tetrahydrofuran (THF), or mixtures of THF and other solvents such as isopropyl alcohol (IPA). With solution concentrations and dip/withdrawal speeds of the present disclosure, the desired superhydrophobic transparent coating is complete immediately following evaporation of the solvent, and no chemical post-treatment of the prepared surfaces is required to render them superhydrophobic.
As hydrophobized nanoparticles are often available commercially, the elimination of the need for surface post-treatment simplifies the fabrication process and reduces costs, particularly for substrate surfaces with large areas. In addition, for many polymer substrates, some post-treatment techniques such as vapor deposition or plasma may be undesirable. The surface coatings provided by the methods and solutions of the present disclosure are characterized as being superhydrophobic in terms of wettability (CA/CAH) and optically transmissive in the visible spectrum (e.g., at least 90% transmissive to visible light). Further, wear resistance experiments using an atomic force microscope and a water jet apparatus to examine sliding wear and impingement of water jet confirm that the surface coatings of the present disclosure are wear resistant.
Aspects of the present disclosure relate to methods for forming a superhydrophobic, optically transparent coating on a substrate, and solutions from which the coatings can be formed. With this in mind,
With reference to
The hydrophobic nanoparticles can assume various forms, and in some embodiments are metal oxide particles. As a point of reference, many metal oxide particles are inherently hydrophilic. These particles can be hydrophobized using silanes or other treatment before combining into the solution. For metal oxide particles (such as SiO2, ZnO, or ITO), phosphonic acids (such as octadecylphosphonic acid (ODP)) can be used to easily modify the particle surface to be hydrophobic.
Functionalize hydrophobic metal oxide nanoparticles useful with the methods and coating solutions of the present disclosure include functionalized SiO2, ZnO, and ITO. The metal oxide nanoparticles can have an average particle size in the nanoscale range. In some embodiments, the metal oxide nanoparticles provided in the coating solutions of the present disclosure are surface-functionalized SiO2 nanoparticles having an average particle size in the range of 1-100 nm, alternatively in the range of 20-80 nm, and optionally on the order of 55 nm (+ or −15 nm). In other embodiments, the metal oxide nanoparticles provided in the coating solutions of the present disclosure are surface-functionalized ZnO nanoparticles having an average particle size in the range of 1-150 nm, alternatively in the range of 20-130 nm, and optionally on the order of 70 nm (+ or −30 nm). In yet other embodiments, the metal oxide nanoparticles provided in the coating solutions of the present disclosure are surface-functionalized ITO nanoparticles having an average particle size in the range of 1-100 nm, alternatively in the range of 20-80 nm, and optionally on the order of 45 nm (+ or −25 nm). In other embodiments, the hydrophobic nanoparticles can be comprised of other materials and/or have other average particle sizes.
The resin binder component of the coating solutions of the present disclosure can assume various forms, and in some embodiments is epoxy or silicone. The resin binder is generally formulated to bind the hydrophobic nanoparticles to the surface of the substrate being coated. In some embodiments, a useful resin binder is methylphenyl silicone resin, as it has low surface energy (˜25 dyne/cm), dissolves in a variety of solvents, and has high hardness (˜1.3 GPa). Low surface energy for the resin binder may improve superhydrophobicity and low roll-off angle of surfaces, as gaps between particles will contain a hydrophobic resin layer. In addition, methylphenyl silicone resin may improve dispersibility in solution for some hydrophobic nanoparticles.
The solvent component of the coating solutions of the present disclosure can assume various forms, and optionally exhibits a high evaporation rate for uniform coatings. Alcohols (such as ethanol or isopropyl alcohol (IPA)) may be less preferred as they may not provide an adequate evaporation rate, and additionally cannot dissolve epoxy or silicone resin binders. Tetrahydrofuran (THF) evaporates very rapidly and evenly from substrates after dip coating, leaving uniform coatings, and easily dissolves epoxy and silicone resins. However, THF dissolves many polymer substrates of interest to methods of the present disclosure, such as polycarbonate (PC) and polymethyl methacrylate (PMMA). In some embodiments, the solvent component is a mixture of THF and an alcohol such as IPA, with the alcohol component being at least 51% (by volume) of the mixed solvent. For example, in some embodiments, solvents of the coating solutions of the present disclosure are a mixture of 30-49% THF and 51-70% IPA (or other alcohol) by volume; alternatively approximately 40%/60% THF/IPA by volume (+ or −5%). It has surprisingly been found that when THF was mixed with IPA at a ratio of approximately 40%/60% THF/IPA by volume, PC and PMMA substrates showed no visible damage after 1 minute immersed in solution. In addition, a 40%/60% THF/IPA mixture evaporated quickly and evenly enough for uniform coatings. For glass substrates, however, pure THF may be used.
With the coating solutions of the present disclosure, a concentration of the hydrophobic nanoparticles in the solvent can vary, for example as a function of the materials employed as the hydrophobic nanoparticles. For example, with embodiments in which the hydrophobic nanoparticles are surface functionalize SiO2 nanoparticles, a concentration of the SiO2 nanoparticles in the solvent can be in the range of 1-20 mg/mL, alternatively in the range of 5-15 mg/mL, and optionally about 10 mg/mL (+ or −1 mg/mL). With embodiments in which the hydrophobic nanoparticles are surface functionalize ZnO nanoparticles, a concentration of the ZnO nanoparticles in the solvent can be in the range of 5-65 mg/mL, alternatively in the range of 25-45 mg/mL, and optionally about 35 mg/mL (+ or −1 mg/mL). With embodiments in which the hydrophobic nanoparticles are surface functionalize ITO nanoparticles, a concentration of the ITO nanoparticles in the solvent can be in the range of 20-80 mg/mL, alternatively in the range of 40-60 mg/mL, and optionally about 50 mg/mL (+ or −1 mg/mL). It has surprisingly been found that too low a concentration can result in loss of superhydrophobicity in the resultant substrate surface coating, while too high of a concentration can result in visible agglomeration of particles on the substrate surface, thus negatively affecting transparency of the resultant substrate surface coating. In other embodiments, the concentration of hydrophobic nanoparticles in the solvent can have other values.
With the coating solutions of the present disclosure, a concentration of the resin binder in the solvent can vary, for example as a function of the materials employed as the resin binder. For example, with embodiments in which the resin binder is methylphenyl silicone resin, a concentration of the methylphenyl silicone resin in the solvent is in the range of 1-5 mg/mL, alternatively in the range of 2.75-4.75 mg/mL, optionally approximately 3.75 mg/mL (+ or −0.5 mg/mL). It has surprisingly been found that in the case of a silicone resin, a concentration significantly below these preferred levels can result in poor adherence of the hydrophobic nanoparticles to the substrate surface, whereas a concentration significantly above these preferred levels can result in loss of superhydrophobicity (likely due to the complete engulfing of particles in the resin layer).
With the above characteristics of the coating solution in mind, the substrate 12 is dipped into the coating solution at step 32. Prior to, or commensurate with this step, the coating solution can be subjected to sonification or other mixing.
At step 34, the substrate 12 is then withdrawn, in some embodiments immediately withdrawn, from the coating solution. The rate of withdrawal of the substrate 12 can be uniform or controlled, and is selected to create a substantially uniform coating of the coating solution on the substrate surface. For example, in some embodiments, the substrate 12 is withdrawn from the coating solution at a rate in the range of 5-15 cm/min, optionally approximately 10 cm/min (+ or −1 cm/min).
As generally reflected in
Immediately following evaporation or removal of the solvent, the precursor coating 42 transitions to the final, superhydrophobic, optically transparent coating 14 of
The superhydrophobic, optically transparent coatings 14 generated by methods and solutions of the present disclosure can have a thickness in the range of 50-150 nm, optionally approximately 100 nm (+ or −10 nm) in some embodiments. Further, the superhydrophobic, optically transparent coatings 14 of the present disclosure are highly wear resistant as described below.
In the examples described below, superhydrophobic, transparent coated surfaces were formed on various substrate samples using methods and coating solutions of the present disclosure. Different hydrophobic nanoparticles and different substrates were employed for various ones of the examples surfaces. Contact angle, contact angle hysteresis, and optical transmittance were measured for samples using all particle-substrate combinations. Wear resistance testing was also performed.
Soda-lime glass (2.2 mm thick), polycarbonate (Lexan, SABIC Innovative Plastics, 2.4 mm thick), and PMMA (Optix, Plaskolite Inc., 2 mm thick) were used to create 1 cm×1 cm substrates. Silane-modified hydrophobic SiO2 nanoparticles with average diameter of 55 nm (±15 nm) were obtained from Evonik Industries (AEROSIL RX 50). ZnO nanoparticles with average diameter of 70 nm (±30 nm) were obtained from Alfa Aesar (NanoTek Zinc Oxide). ITO nanoparticles (90:10 In2O3:SnO2) of average diameter 45 nm (±25 nm) were obtained from US Research Nanomaterials (US3855 Indium Tin Oxide Nanopowder). Octadecylphosphonic acid (ODP) was purchased from Aldrich, and methylphenyl silicone resin was obtained from Momentive Performance Materials (SR355S Methylphenyl Silicone Resin).
While the obtained SiO2 particles were already silane-modified, the ZnO and ITO particles were not surface-modified as received. In order to hydrophobize them, they were treated in solution by octadecylphosphonic acid (ODP). ODP can be used to functionalize metal oxides from hydrophilic to hydrophobic. The process by which functionalization occurs is illustrated in
Particles were dispersed in a 40%/60% THF/IPA (by volume) mixture to form the dip coating solution. While pure THF rapidly dissolves PC and PMMA resulting in complete loss of transparency, it was found that when THF concentration was kept below approximately 50% by volume in IPA, substrates could be dipped for over one minute without visible damage or loss of transparency. A dip coating solution of pure IPA, however, does not evaporate quickly or evenly enough to leave a homogeneous coating on the substrate. Optimal concentrations of nanoparticles in the solvent were found to be approximately 10 mg/mL for SiO2 particles, 35 mg/mL for ZnO particles, and 50 mg/mL for ITO particles. Too low of a concentration resulted in loss of superhydrophobicity, while too high of a concentration resulted in visible agglomeration of particles on substrates, substantially reducing transparency.
The nanoparticles were added to 30 mL of the THF/IPA solvent in a 100 mL glass beaker and sonicated for 4 min with a Branson Sonifier 450A (20 kHz frequency at 35% amplitude). Then, 150 mg of methylphenyl silicone resin was added and the mixture was sonicated for an additional 4 min. In the case of the silicone resin, a concentration significantly below this optimal level resulted in poor adherence of the particles to the substrate. Concentration significantly above this level resulted in loss of superhydrophobicity, likely due to the complete engulfing of particles in the resin layer. In addition to ultimately acting to bind nanoparticles to the substrates, the silicone resin worked excellently as a dispersant in the dip coating solution. For ITO particles in particular, settling was noticeable within seconds when silicone resin was not added, but particles remained homogeneously dispersed with resin included. After sonication, approximately 10 mL of fresh solvent was added at 40%/60% THF/IPA ratio. Substrates were dipped into the solution and immediately removed at a speed of 10 cm/min. Coated samples were then heated at 40° C. for 10 min to remove any remaining solvent. The samples required no chemical post-treatment or modification after dip coating.
For wettability measurements, water droplets of 5 μL, volume (˜1 mm radius) were deposited onto samples using a microsyringe. Reproducibility of all CA/CAH data is reported as (±Σ) as determined from measurement on five samples using a model 290-F4 Ramé-Hart goniometer (Ramé-Hart Inc., Succasunna, N.J.). Values for fLA were estimated using SPIP™ imaging software (Image Metrology). Transmittance measurements were performed using an Ocean Optics USB400 spectrometer (Ocean Optics Inc., Dunedin, Fla.) with a 200 μm aperture width. All transmittance data is reported for a one-sided coating as a percentage of the transmittance of the uncoated substrate in the visible spectrum (400-700 nm).
To examine the wear resistance of the samples, wear experiments were performed using an AFM and water jet apparatus. In order to study sliding wear resistance, an established AFM wear experiment was performed with a commercial AFM (D3100, Nanoscope IIIa controller, Digital Instruments, Santa Barbara, Calif.). For wear experiments, investigation of single asperity contact is necessary to understand fundamental interfacial phenomena. An AFM tip can simulate single asperity contact for micro/nanostructured surfaces. Samples with SiO2, ZnO, and ITO nanoparticles on glass substrates were worn using a borosilicate ball with a radius of 15 μm mounted on a rectangular Si(100) cantilever (k=7.4 N/m) in contact mode. Areas of 50×50 μm2 were worn for 1 cycle at a load of 10 μN. To analyze the change in the morphology of the surfaces before and after the wear experiment, height scans of 100×100 μm2 in area were obtained using a rectangular Si(100) tip (f=76 kHz, k=3 N/m) in tapping mode. As a baseline, the wear results for the samples were compared to that of the silicone resin alone on a glass substrate.
An established water jet procedure was performed to examine macroscale wear resistance of the samples in water flow. For applications involving self-cleaning glass, resistance to impingement of water is of critical interest. A schematic of the water jet setup is shown in
The nine types of transparent superhydrophobic samples using three different nanoparticles (SiO2, ZnO, ITO) on three different substrates (glass, PC, PMMA) are discussed below. First, roughness values and surface morphology are presented. Then, the CA, CAH, and transmittance of samples are reported, discussing trends in the data. Lastly, the results of the wear resistance experiments are examined.
Table 1 displays RMS roughness, PV (peak-valley) distance, roughness factor (Rf), and coating thickness for samples with SiO2, ZnO, and ITO. Surfaces had nanoscale roughness formed by nanoparticles bound to the substrate with silicone resin. The values of Rf were calculated using AFM surface height maps. By using the Z-height of each data point in the AFM scan matrix, the real surface area can be approximated using simple geometry. Dividing this value by the two-dimensional scan area provides Rf. Coating thicknesses were measured with a Tencor® stylus profiler on the step formed by partially coating a substrate, and found to be nearly equal to PV distance.
Data for samples using three different nanoparticles (SiO2, ZnO, ITO) on three different substrates (glass, PC, PMMA) are shown in Table 2 (data shown graphically in
The Cassie-Baxter equation (Eq. 1) can be used to predict CA in the case where the droplet rests only on the highest asperities, with air filling gaps between:
cos θ=Rf cos θ0−fLA(Rf cos θ0+1) (1)
where θ0 is the CA on a flat surface of identical surface energy, and Rf (roughness factor) is the ratio of the real area of the interface to its two-dimensional projection. In the case of full wetting with no air pockets (Wenzel regime), CA can be predicted by simply setting fLA equal to zero in Eq. 1. AFM surface maps were analyzed and fLA values (shown in Table 1) were estimated to be 0.94, 0.94, and 0.91 for SiO2, ZnO, and ITO samples, respectively. ITO nanoparticles tended to form more evenly distributed microscale islands with lower height distribution compared to SiO2 and ZnO, possibly due in part to smaller primary particle size. In addition, CAH can be estimated by:
where θrec0 and θadv0 are the flat-surface receding and advancing contact angles, respectively.
Table 3 shows measured and calculated CA and CAH values, with measured values taken from samples on glass substrates. Flat-surface angles (θ0, θrec0, θadv0) were measured on a glass slide modified with the same ODP solution, and found to be θ0=103°, θrec0=75°, and θadv0=132°. Comparison of the measured vales to calculated Wenzel and Cassie-Baxter values strongly suggests a Cassie-Baxter regime, especially given the very low CAH values measured, which are typically associated with Cassie-Baxter wetting. Thus, the droplet predominately contacts the highest peaks of the hydrophobic-modified nanoparticles. For CAH, Eq. 2 predicts essentially identical values of 0.3°, 0.4°, and 0.3° for SiO2, ZnO, and ITO, respectively. The lower measured CA and higher CAH for ITO compared to calculated values may suggest that a small fraction of the droplet contacts the substrate for ITO samples, leading to partial Wenzel wetting and higher hysteresis. This may be a result of the topography of microsized islands for ITO samples, which tended to be smaller and with greater distance between. In the case that the droplet interface exhibits a small fraction of Wenzel wetting behavior, the hydrophobicity of the silicone resin (θ0=99°) may help to preserve superhydrophobicity (CA>150°) and low CAH (<10°) necessary for self-cleaning behavior.
The uncoated glass, PC, and PMMA substrates had visible-range transmittances of 92%, 87%, and 94%, respectively. Transmittances of the samples are reported as percentages of the transmittance of the uncoated substrate. Table 2 shows the average values for samples across the visible spectrum (data shown graphically in
The transmittance of the samples in the context of coating thickness is in rough agreement with other published studies that have reported transmittance of greater than 90% for coating thickness of 380 nm using SiO2/PDMS, and coating thickness of about 60 nm using SiO2, respectively. Data for SiO2 samples in these previous studies were intermediate, with transmittance values from 90-96% for coating thickness of 137 nm. Although very high transmittance values were achieved, an antireflective effect resulting in transmittance greater than 100% of the uncoated substrate, as reported in some studies, was not seen. In addition, the typical morphology of disconnected islands of particles, while beneficial for roughness and superhydrophobicity, disallows a path for electrical current. Further development of this technique would be necessary to prepare transparent, superhydrophobic coatings that are also electrically conductive. ITO-coated glass, PC, and PMMA samples with deposited water droplets can be seen in
The results of the AFM wear experiment for SiO2, ZnO, and ITO particles on glass as well as silicone resin alone on glass are shown in
The results of the water jet experiment can be seen in
The versatile dip coating techniques and coating solutions of the present disclosure were systematically shown to create transparent, superhydrophobic surfaces on glass and plastic substrates with SiO2, ZnO, and ITO nanoparticles. ZnO and ITO particles were hydrophobized with ODP, and the prepared samples did not require post-treatment with low surface energy substances. The nanoparticles showed different tendencies in the way they deposited onto substrates from dip coating, which may be partly due to differences in primary particle size. This caused variation in coating thickness and morphology between particles, which helps to explain differences in wettability and transmittance between samples. ITO samples had slightly lower CA and slightly higher CAH than SiO2, and ZnO, which is likely the result of a comparatively lower liquid-air fractional area (fLA). Roughness and coating thickness seemed to influence transmittance more than inherent optical properties of particles, which may be due to the proximity of roughness and thickness values to the 100 nm threshold for visible transparency. Samples on PMMA substrates performed modestly better than those on PC and glass in terms of wettability and transmittance. However, all samples exhibited a superhydrophobic CA (>150°), low CAH (<10°), and high transmittance of visible light (>90% in most cases). In addition, all surfaces showed wear resistance for potential commercial use in AFM wear and water jet experiments, indicating strong bonding of the silicone resin and sufficient hardness of nanoparticles and resin.
Transparent superhydrophobic surfaces with wear resistance can be fabricated in accordance with principles of the present disclosure with a broad range of materials to expand potential engineering applications. However, primary particle size, roughness, and coating morphology appear to be at least as important a factor in transparency as inherent optical properties of the nanoparticles when coating thickness is on the order of 100 nm.
Methods in accordance with the present disclosure entail the dip coating of glass or polymer substrates in a solution containing hydrophobic nanoparticles, a resin binder, and a solvent. Samples were successfully created using silicon dioxide (SiO2), zinc oxide (ZnO), and indium tin oxide (ITO) nanoparticles on soda-lime glass, polycarbonate (PC) and polymethyl methacrylate (PMMA) substrates with a methylphenyl silicone resin binder and solvent containing a mixture of tetrahydrofuran (THF) and isopropyl alcohol (IPA). Solution can be sonicated to improve dispersion. With appropriate solution concentrations and dip removal speed, superhydrophobic surfaces with high transmittance to visible light are obtained. Surfaces do not require post-fabrication treatment with low-surface-energy compounds, such as fluorosilanes, to achieve superhydrophobic effect.
Particle and resin binder concentrations and dip removal speed may vary based on particle type and size. Using silane-modified SiO2 nanoparticles with average diameter ˜55 nm, useful concentrations were found to be approximately 10 mg/mL SiO2 particles and 3.75 mg/mL methylphenyl silicone resin for a dip removal speed of 10 cm/min in some non-limiting embodiments.
The coated surfaces described here may be useful for applications involving self-cleaning windows/windshields, solar panels, or high performance optical devices, among others.
Although the present disclosure has been described with reference to preferred embodiments, workers skilled in the art will recognize that changes can be made in form and detail without departing from the spirit and scope of the present disclosure.
This invention was made with government support from the National Science Foundation, Grant Number CMMI-1000108. The government may have certain rights in the invention.
