Biostable conductive nanocomposite for implantable subdermal antenna

Implantable bioelectronics are becoming increasingly important for diagnostics and treatments of diseases, including neural electrode arrays for functional brain–computer interfaces,¹ deep brain stimulators,^2,3 intracranial seizure advisory systems,⁴ cardiac pacemakers,⁵ and cochlear implants.⁶ Wireless antennas are an essential component to implantable bioelectronic telemetry systems for monitoring the status of the medical devices and transmission of patient data.^4,7,8 To perform efficiently, an antenna must be approximately half a wavelength long at its operating frequency.^9–11 Common frequencies for implantable antennas are the MedRadio band (402 MHz–405 MHz), which corresponds to an antenna length of ∼6 cm while implanted in the body.¹² Many MedRadio antennas are less than half a wavelength long, which is suboptimal for the antenna, but often necessary given the available space. Medical devices are continually shrinking (the Utah neural array is about 4 × 4 mm²)¹⁰ and no longer have the space available to directly support this antenna size on the device itself. By remotely locating the primary antenna closer to the skin within the fat layer⁸ and installing a smaller, more accommodating antenna to act as a passively coupled feed, the size of the antenna is no longer limited to the size of the implantable device. The larger subdermal antenna receives signals from the smaller antenna feed on the medical device and re-radiates the signals outside the body. Locating the antenna within the fat layer [Fig. 1(a)] takes advantage of the fat’s electrically insulating properties and eliminates the need for encapsulation of the antenna in supplementary insulating materials.

Traditionally, antennas are made from solid metals like copper and aluminum. These metals exhibit high stiffnesses, characterized by their high Young’s moduli (∼100 GPa). The resultant mechanical mismatch between the antenna and its surrounding tissue will yield tissue damage, inflammation, and a decline in performance quality.^13,14 The minimum thickness of an antenna is limited by the depth an electromagnetic current flows within its surface, called the skin depth (δ). As the antenna thickness drops below the skin depth, the performance quality weakens because the conductor cannot carry enough current. This results in a decrease in the resonant frequency, a degradation (increase) of the reflection coefficient, and an increase in the signal bandwidth.¹⁵ At the MedRadio band frequency (∼402 MHz), silver, gold, copper, and aluminum all exhibit electrical skin depths (minimum acceptable thicknesses) upward of 3 μm. Metal films thick enough to support the current (>skin depth) will be too thick to be flexible.

Although many different materials have been used for flexible antennas, including microfluidic systems,¹⁶ a liquid crystal polymer,¹⁷ a silver nanowire and alginate nanocomposite,¹⁸ and conductive inks and pastes consisting of nanoparticles and nanoflakes,¹⁹ to our knowledge, there has not been a systematic optimization and evaluation of material properties for in vivo applications. The material requirements for an implantable antenna include: (a) excellent electrical conductivity (>10⁴ S/m)¹⁵ to support electromagnetic signals, (b) mechanical flexibility to be compliant with soft tissues, (c) biostability for chronic performance in vivo, and (d) biocompatibility to minimize toxic effects. It is possible to achieve this combination of material properties by matching a soft polymer with a nanoscale conductive filler. Suitably high conductivities and softnesses have been shown for nanocomposites containing conductive fillers like gold (Au) nanoparticles,²⁰ silver (Ag) nanowires and nanoflakes,^18,21–26 carbon nanotubes (CNTs),^27,28 Ag flake decorated CNTs,^29–31 Ag nanowires with an encapsulating gold shell,^32,33 and conductive polymers.^34–36 Excluding the Ag–Au core–sheath nanowire composite,³² these materials have been explored solely for their applications outside of the body. Ag and CNT based materials exhibit exceptional conductivities, but the potential cytotoxicity of these materials greatly limits the biocompatibility of the antenna when placed inside the body.^37,38 Conductive polymer based materials are biocompatible,³⁹ but potential biodegradation of conductive polymers can hinder chronic performance.^40,41 One-dimensional Ag nanowires with an encapsulating Au shell offer biocompatibility, low percolation thresholds, and biostability, but present setbacks from challenging synthesis and extensive purification procedures as well as issues matching with biostable polymers to make thin enough samples (<15 μm).^32,33,42 Even though zero-dimensional Au nanoparticles do not offer low percolation thresholds like one-dimensional nanowires, Au nanoparticles display low cytotoxicity,⁴³ follow a remarkably simple and scalable one-pot synthesis procedure,⁴⁴ are easy to match with biostable polymers, and have been studied extensively for biological applications.⁴⁵ Previously, Au nanoparticle based nanocomposites have been explored for neural interface applications⁴⁶ and stretchable conductors.²⁰ However, there has not been a comprehensive study on the correlation of the conductivity, mechanical properties, and biostability of the Au nanoparticle based composite.

In this study, we fabricated a flexible and conductive nanocomposite antenna with excellent biostability and biocompatibility. We combined waterborne polyurethane and aqueous Au nanoparticles to optimize implantable antenna properties. A cationic polyurethane was chosen for the nanocomposite’s structural matrix, and anionic Au nanoparticles were chosen for the conductive filler. Polyurethanes have remained one of the most popular biomaterials for over 50 years, due to their outstanding biocompatibility, tunable mechanical properties, and versatility.⁴⁷ The polyurethane selected displays a high biostability and softness due to its two-phase microstructure of hard and soft segments.^48,49 A strong resistance to hydrolytic degradation is attributable to its high degree of hydrogen bonding.^48–51 The exhibited ratio of soft polyol segments to hard isocyanate segments within the polyurethane contributes softer properties giving the films more tissue compliant features. Its strong positive charge allows for a robust ionic network to be formed with the negatively charged Au nanoparticles.²⁰ Au nanoparticles were chosen for their high conductivity, stability, biocompatibility,⁵² and ease in processing. The Au nanoparticles are stabilized by a thin citrate layer that exhibits a strong negative charge. The strong ionic attraction between the selected Au nanoparticles and polyurethane allows for a simple one-pot fabrication procedure and is necessary for highly uniform dispersions of Au nanoparticles within the polyurethane matrix.

By following a vacuum assisted flocculation (VAF) method, we synthesized a nanocomposite made from aqueous Au nanoparticles and waterborne polyurethane.²⁰ The synthesis procedure provides strong versatility by allowing for numerous compositions of Au nanoparticles and polymer to be made. Included in our previous work¹⁵ is a comprehensive electromagnetic analysis of the strip dipole antenna made from the Au nanoparticle nanocomposite as well as several other conductive materials. We only briefly described the synthesis method of the Au nanoparticle nanocomposite. In this work, we systematically investigated the correlations of material composition with electrical conductivity, mechanical properties, and antenna performance. For a more detailed electromagnetic analysis, please see Ref. 15. The optimized nanocomposite antenna exhibits high conductivities (∼10⁵ S/m), low Young’s moduli (10² MPa–10³ MPa), exceptional biocompatibility proven by in vivo and in vitro work, and long lasting biostability proven by degradation studies.

In this study, our objective is to develop a material that has suitable electrical and mechanical characteristics for an implantable antenna that can be embedded in the fat layer under the skin. The electrical characteristics are controlled primarily by the conductivity of the material used^15,53 and the size/shape of the antenna.^9–11 We will test our designs on simple strip dipole antennas (30 mm long and 6 mm wide), which are less than a half-wavelength long at 402 MHz.^8,15,53 The minimum thickness of the antenna is controlled by the skin depth (δ), which is the depth of penetration of the current into the conductive antenna material,⁵⁴ 

where f is the frequency (in Hz), and σ is the conductivity (in S/m). Previous studies^15,53 have found that the conductivity of the material should be at least 10⁴ S/m, and conductivity at 10⁵ S/m will help to reduce both skin effect and loss.

Nanocomposite synthesis followed a one-pot precipitation based method by mixing spherical Au nanoparticles and polyurethane [Fig. 1(b)]. Au nanoparticles were produced by following the well-established aqueous citrate reduction method.⁴⁴ The sodium citrate used reduces the Au ions and forms a thin layer encapsulating the reduced Au atoms, allowing for the Au nanoparticles to reach a close proximity with one another in the polymer matrix. Use of Au nanowires as a conductive filler produces an appealing alternative, but nanowires are too thin (∼2 nm) rendering them brittle and contain long chain capping agents preventing electron transfer between wires.^55–57 The aqueous synthesis of citrate nanoparticles eliminates potential toxic organic molecules found in other Au nanoparticle synthesis protocols. The Au nanoparticles expressed diameters of 14 ± 2.1 nm, characterized by UV–Vis spectroscopy and Dynamic Light Scattering (DLS) (see the supplementary material). Cationic polyurethane was diluted in deionized water to 1 vol. % preceding rapid addition to the Au nanoparticle solution. Following, flocculation with Au nanoparticles occurred instantaneously, changing the solution from a dark wine red color to a dark blue. The polyurethane displays a strong positive charge allowing for a robust ionic interaction to be formed with the anionic Au nanoparticles.²⁰ With the addition of polyurethane, the nanocomposite adopts its softer properties, giving the films more tissue compliant features.

Following synthesis of the nanocomposite, detailed characterizations were performed to understand nanocomposite properties. Nanocomposites exhibit a metallic appearance and mechanical flexibility [Fig. 1(c)]. A SEM image of the nanocomposite may be seen in Fig. 1(d) and displays the densely packed Au nanoparticles within the polymer network. The close proximity of nanoparticles provides a pathway for electrons to travel.

We performed a systematic analysis to observe the effects of nanoparticle loading on the nanocomposite’s mechanical and electrical properties. Thickness measurements were taken with SEM [Fig. 2(a)], Zygo optical profiler [Fig. 2(b)], and Tencor profilometer, displaying reliable consistency between measurements. Various mixtures of nanoparticles and polyurethane were used with amounts confirmed by thermogravimetric analysis (TGA). A TGA plot is displayed in Fig. 2(c), while under a nitrogen atmosphere, the polyurethane began to vaporize at 200 °C and by 800 °C only the gold content remained. Subsequently, mechanical properties were characterized by tensile measurements [Fig. 2(d)]. The material’s Young’s modulus can be extracted by taking the slope from the initial linear portion of the stress/strain curve, where stress is the force per unit area of each sample and strain is the proportional deformation. Nanocomposites expressing ideal conductivities for antenna performance exhibit Young’s moduli from 0.5 GPa to 0.9 GPa. From Fig. 2(d), we observe the yield strength of the nanocomposite is ∼12 MPa, which is comparable to typical polymeric materials. Electrical conductivities of different nanoparticle loadings were measured with a four-point probe. Bulk resistivity values were calculated by inputting film thicknesses, following, the conductivity was calculated by taking the inverse of the bulk resistivity. In Fig. 2(e), we plotted the relationship between the electrical conductivity and Au content. Three segments can be related to increasing volume fractions of filler and increasing electrical conductivities. For low filler concentrations, highlighted in magenta, the conductive particles are electrically isolated from one another and the conductivity is that of the insulator. For the next segment highlighted in green, the nanocomposite transitions from insulator to conductor. The start of this segment illustrates the percolation threshold and is due to the formation of inter-particle contacts. Following, a sharp increase in conductivity over a short range of filler concentration can be observed. The third segment highlighted in yellow shows a conductivity plateau being reached.^58,59 For our nanocomposite, we observed the percolation threshold at ∼24 vol. %. Complementing percolation studies, we measured tensile properties of the nanocomposites to find the Young’s modulus followed a linear trend with increasing nanoparticle content, as shown in Fig. 2(f). Nanocomposites selected for antenna testing with corresponding properties are listed in Table I. For our highest conductivity (89.9 wt. % Au), our nanocomposite exhibited a conductivity of 7.9 × 10⁵ S/m, a thickness of 10 ± 0.4 µm, and a Young’s modulus of 0.93 GPa. Young’s moduli displayed values almost two orders of magnitude lower than solid Au (79 GPa).

Following characterization of the nanocomposite, antennas were fabricated and measured for varying nanocomposite conductivities. Simple half-wave strip dipole antennas were constructed (30 mm long and around 6 mm wide, for use at 402 MHz), and their reflection coefficients were measured. Images of the fabricated antenna can be found in Fig. 3(a). Tests¹⁵ were performed with each antenna sandwiched between a layer of skin and fat (excised from a rat) on pork loin, as shown in Fig. 3(b).^15,60 This provided a realistic representation of human skin-fat-muscle for the purpose of testing. The reflection coefficient, S₁₁, is a standard indicator for antenna performance, expressed as the ratio of the incident voltage and reflected voltage. Ideally, an antenna will transmit its incoming voltage with minimal reflection. An S₁₁ < −10 dB is commonly required for communication system design. S₁₁ was measured at the antenna feed point [the small gap at its center, as shown in Fig. 3(b)] using a vector network analyzer with a handmade 2-wire probe (made from bent stainless steel pins) connected across the gap. The network analyzer transmits an incident signal to the antenna and receives the antenna’s reflected signal. In Table I, the bandwidth over which S₁₁ < −10 dB and minimum reflection coefficient values are listed, along with corresponding antenna material properties. Full details of the S₁₁ curve can be found in Ref. 15. Nanocomposite conductivities of 10⁵ S/m produce high quality antennas, with a lower S₁₁ and broader bandwidth, while conductivities as low as 10³ S/m are functional in limited cases. Antennas with conductivities of 10⁴ S/m display more than adequate performance and should be selected for use as implantable antennas due to their lower Young’s modulus when related with higher gold loadings. Au nanoparticle antennas with conductivities of 10⁴ S/m presented lower Young’s moduli (491.8 ± 49.0 MPa) than 10⁵ S/m antennas (926.1 ± 62.2 MPa). Subsequent antenna design with these and other materials, including measurements of antennas in air, confirms these observations.¹⁵ 

In addition to displaying exceptional antenna properties, biocompatibility and biostability are necessary for chronic in vivo performance. We examined the biocompatibility of the nanocomposites by in vitro cell viability/differentiation studies with NG108 cells as well as in vivo histology evaluation in mice. The in vitro biocompatibility study served as a pre-screening test before the in vivo histology study. NG108 cell lines were selected so we could observe both the cell viability and differentiation process. The NG108 cells were cultured for one week with the nanocomposites (n = 5). A live/dead assay was also performed and indicated that 94% of the cells were alive on the surface of the glass directly adjacent to the nanocomposite [Fig. 4(a)]. A control live/dead assay performed on glass slide is shown in Fig. 4(b), with 83% of the cells alive. This proves the nanocomposite exhibits a biocompatible environment for cell growth. In addition to cell viability, a SEM image in Fig. 4(c) shows cells attach and differentiate into neuron-like morphology with considerable dendrite growth. This indicates the nanocomposite has minimal interruption to the cellular functionality of the cells.

For the in vivo study (n = 5), upon gross examination, there were limited adhesions between the device and the surrounding skin and fat pad. The device was removed, and histological sections were analyzed after fixation, embedding, sectioning, and staining with hematoxylin and eosin. Figures 4(d) and 4(e) are representative images from two of the devices that were implanted for one week. Figure 4(d) has a less dense fibrotic layer and more inflammatory cells present, while Fig. 4(e) is a denser fibrous layer and fewer inflammatory cells. These two types of phenotype were present on all samples at differing percentages of length of surface. All the devices had a fibrous tissue layer although the thickness of that layer differed along and between devices. In addition to the fibrous layer, there are areas of inflammatory cells infiltration including lymphocytes, neutrophils, and macrophages. This confirmed the nanocomposite is not toxic to the cells and can support cell growth/differentiation. Following the in vivo study, the nanocomposite was imaged by SEM [Fig. 4(f)] and was shown to maintain its densely packed Au nanoparticle network, for before comparison, see Fig. 1(d).

To test the films’ biostability, accelerated degradation studies were carried out to observe the temporal release of Au nanoparticles. Hydrolysis is one of the dominant degradation pathways found to take place in polyurethanes when implanted.⁵¹ To calculate the accelerated rate, the degradation time was fit to a first order chemical reaction assuming the rate of aging is increased by a factor of 2^∆T/10, roughly translating to a doubled rate every 10 °C.⁶¹ Films (n = 5) were incubated at 55 °C for two months (eight months equivalent at 37 °C) in phosphate buffered saline (PBS) and cell culture medium containing 10 wt. % fetal bovine serum. The released Au content of the incubation medium was measured by Inductively Coupled Plasma Mass Spectrometry (ICP-MS), and the results are shown in Table II. We observed less than 0.002% of Au nanoparticles released into the cell culture medium over eight months at 37 °C, while the Au nanoparticle release in the PBS was negligible. Concurrently, nanocomposite films displayed no decrease in conductivity after in vivo and in vitro studies. From these results, it can be seen the nanocomposites present compatible surfaces for cells to grow and proliferate. The films have also been shown to acquire polyurethane’s significant barrier to hydrolysis, providing the opportunity for a chronically implanted antenna.

In summary, we present highly conductive and soft Au nanocomposites developed through a vacuum assisted flocculation method for use to fabricate implantable antennas. The nanocomposites exhibit properties more compatible for implantable antennas compared to traditional antenna materials as well as CNTs, or Ag based nanocomposites. An optimized functional antenna for the MedRadio band was produced by systematically investigating the effects of the Au nanoparticle content on the electrical conductivity, mechanical properties, and antenna performance. The material has high conductivity (∼10³ S/m–10⁵ S/m), reduced Young’s moduli (10² MPa–10³ MPa), very low toxicity, and exceptional biocompatibility and biostability. These results present a promising material for nanocomposite antennas with the potential to establish chronic in vivo communications with medical devices for transfer of patient and medical device health data.

HAuCl₄, sodium citrate, and all chemicals were purchased from Sigma-Aldrich. Cationic waterborne polyurethane was acquired from Hepce Chem, South Korea (30 vol. % molecular mass ∼92 000). Vacuum filtration apparatus and filter papers were purchased from Fisher Scientific. The filter paper has 0.8 μm pore size with 47 mm diameter. All cell culture media and chemicals were purchased from Gibco, part of Thermo Fisher Scientific.

The method for Au nanoparticle synthesis was adapted from Ref. 20, which is a modified version of Ref. 44; using a 2 l glass beaker, a mixture containing 950 ml of deionized water and 180 mg of Au(III)chloride trihydrate was vigorously stirred and heated until boiling, where it was held for 20 min. To the boiling mixture, 50 ml of a 34 mM sodium citrate solution was added. The mixture was returned to a boil where it was held for 20 min. The solution became a deep wine-red color and later was allowed to cool to room temperature while being stirred. The prepared solution was next used for the formation of VAF films.

To the already prepared Au nanoparticles, varying amounts of 1 vol. % aqueous polyurethane were added. The diluted polyurethane concentrations were measured by the weight of polyurethane to deionized water. After addition of polyurethane, the mixture was stirred vigorously for 15 min. Subsequently, filtration was performed by adding the mixture to a filtration assembly consisting of a vacuum filtration device and filter papers with 0.8 μm sized pores. For smoother films, two intermediate filter supports were added between the filter paper and the polytetrafluoroethylene (PTFE) faced support on the funnel of the filtration system. Following filtration, the films were allowed to dry for 1 day–3 days. The films were later separated from the filter paper by being first wetted in DI water then submerged in acetone to dissolve the filter paper. The films were again submerged in DI water to wash away acetone and then taken out to allow for drying. Characterization followed once the films were fully dried.

Elastic properties were determined by using an Instron Universal Testing System model 4411 with a 50 N load cell. Films were cut into 4.4 mm wide and 30 mm long rectangular strips and pulled at a rate of 0.08 mm s⁻¹. The film thickness was measured by a Zygo Nexview 3D optical surface profiler, a Tencor P-10 profilometer, and SEM.

Filler fractions and differential scanning calorimetry data were determined by using a Q600 TGA from TA instruments. Samples were exposed to a temperature regime of 25 °C to 800 °C at 10 °C min⁻¹ under a N₂ atmosphere at a flow rate of 20 ml min⁻¹. Conductivity measurements were carried out by using a 4-point Probe Microtech RF-1.

Three Au nanoparticle dipole antennas with varied conductivity (7.9 × 10⁵ S/m, 1.0 × 10⁵ S/m, and 5.0 × 10³ S/m) (Table I) were formed into strip dipoles and tested on a body model made from pork loin (150 × 90 × 30 mm³). Each strip dipole had two arms (30 × 6 mm²) with slight thickness variation due to the manufacturing process. The thicknesses of the 7.9 × 10⁵ S/m, 1.0 × 10⁵ S/m, and 5.0 × 10³ S/m materials were 10.0 µm, 12.2 µm, and 13.4 μm, respectively.¹⁵ 

A passive, two-prong probe made from a modified SubMiniature version A (SMA) connector jack was used for measurement of reflected power (S₁₁) on an Agilent 8753C vector network analyzer. Antenna radiation was confirmed based on low S₁₁ values across the frequency range of 0.4 GHz–1 GHz as shown in Table I. Smith chart impedance measurements are also characteristic of power radiation.

Losses seen in the measurement were likely due to a combination of high relative dielectric values of muscle (57.9 at 402 MHz^62–64), and skin effect losses.

For antennas with thickness less than the skin depth, there is a loss in performance. This was seen here as the skin depth at 402 MHz for conductivities of 7.9 × 10⁵ S/m, 1.0 × 10⁵ S/m, and 5.0 × 10³ S/m is 28.2 µm, 79.3 µm, and 355 μm, respectively. At 2 GHz, the skin depths are 10.0 μm, 12.2 μm, and 13.4 μm.

The feasibility of the Au nanoparticle antenna as an implantable antenna was further confirmed by overlaying the 7.9 × 10⁵ S/m material with excised rat skin (2.6 mm) and fat (1.7 mm), such that the antenna is above the pork loin and below the rat fat and skin layer as shown in Fig. 3(b). Measurement pins [shown in Fig. 3(b)] were inserted through rat skin and fat for measurement. The low S₁₁ of the antenna in this realistic body model also confirmed the viability of the Au nanoparticle material.

All procedures were approved by the University of Utah IACUC committee (20–010030). Adult male C57BL/6 mice (n = 5; 10–12 weeks) were anesthetized with isoflurane, surgical area shaved and prepared with alcohol and betadine. A cutaneous incision, ∼1.5 cm, was made on the back of the animal above the supraspinal fat pad.⁶⁵ A 1 × 1 cm² sample was attached to the fat tissue with a single suture, 6-0 Vicryl (Ethicon, Johnson and Johnson, NJ), and the skin was closed with the same suture. One week post-implantation, the animal was euthanized, the device exposed and removed with attached tissue, fixed in formalin, embedded in paraffin, sectioned, and then stained with hematoxylin and eosin. The devices were then evaluated histologically for indications of inflammation and cellular response to the device.

We cultured NG108, neuroblastoma–glioma hybrid cells on the nanocomposites (n = 5) to understand their adhesion and viability on the nanocomposite pieces. The nanocomposite was pressed on a glass substrate and sterilized by UV-light for 2 h. Cells were seeded on the sterilized nanocomposite and cultured in high glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 10% (v/v) FBS, 1× (v/v) HAT supplement, 1% (v/v) penicillin/streptomycin and incubated in a CO₂ incubator at 37 °C for 2 days. The cells were further incubated in the same culture medium with reduced FBS concentration (1% FBS) for two days to allow differentiation.

Prior to cell imaging, the cells on the nanocomposite were fixed with 4% paraformaldehyde for 12 h. After fixing, the samples were washed with buffer solution, 50%, 60%, 70%, 80%, 90%, and 100% ethanol, and hexamethyldisilazane (HMDS) subsequently for 20 min each. The cells on the nanocomposite in the HMDS were dried under a chemical hood for HMDS evaporation. The dried cell samples were sputter coated and imaged in the SEM.

SEM images of the nanocomposite and neuron attachment were taken by using a Hitachi SU5000. The nanocomposite was imaged under an acceleration voltage of 20 keV, and the cells were imaged at 3 keV.

Biostability tests were performed by placing 4.4 mm wide by 15 mm long samples (n = 5) in both cell culture media with 10% fetal bovine serum and 1M phosphate buffered saline. Samples were incubated in cell culture media and phosphate buffer at 55 °C for 2+ months. Films were extracted for the incubation solution, and the incubation solutions were analyzed using ICP-MS to determine the amount of Au in the solution.

See the supplementary material for UV–Vis spectroscopy and dynamic light scattering data.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

This work made use of the University of Utah USTAR shared facilities (Surface Analysis Lab) supported, in part, by the MRSEC Program of NSF under Award No. DMR-1121252. The authors thank Dr. Jeff Kessler (Mechanical Engineering, University of Utah) for assistance in gathering mechanical properties of the nanocomposites. The authors also thank Dana Overacker (Chemical Engineering, University of Utah) and Robert Cox (Chemical Engineering, University of Utah) for their aid in characterizing compositions of nanocomposites. The authors thank Dr. Diego Fernandez (Geology and Geophysics, University of Utah) for his assistance in obtaining ICP-MS results. The authors also thank Dr. Robert Soto (Chemical Engineering, University of Utah) for his assistance in DLS measurements. The authors also thank Dr. Terry Ring (Chemical Engineering, University of Utah) for discussion of nanoparticle synthesis and percolation theory. The authors are grateful for the support from National Science Foundation Grant No. 1310642, the NSF EAPSI Summer Graduate Program, and the Utah Science Technology and Research (USTAR) fund, Grant No. UTAP-172163.f.