Synthesis and Study of Biocompatible Materials Interfaces

Before such microfluidic structures can be used in a practice, a formidable scientific and technical barrier must be overcome. The barrier is the fact that many of the materials under consideration for biosensing and biomedical platforms are incompatible with the biochemistry of cells and molecular species contained in aqueous solution. There are several major hurdles that must be cleared before these materials that form the micro- and nanostructures can become useful in medical applications. For example, modified material surfaces must be developed for reliably transporting biofluids across materials surfaces at high speed while minimizing the effects of surface tension and nonspecific protein adhesion. These fluids include pH buffered saline solutions, cell growth media, antibodies, cytokines, enzymes, as well as the molecules, cells and extracellular matrix attending the cells. The fluids must not degrade the electronic, optical, or mechanical performance nor corrode material surfaces. Proteins and cells must traverse material surfaces without sticking to the surface and limiting the flow of the biofluid. The materials surfaces must be passivated with inert materials that will not alter the physiology of the cells yet be robust enough to withstand enzymes that may attack them. Further, methods for sealing the dissimilar materials to form micro- and nanocavities must be developed to isolate fluids and prevent cross contamination from adjacent flow channels. The integrity of these seals must be retained over the lifetime of the structure such that the physical dimensions of the micro- or nanocavity remain constant.

Fundamental questions arise as to how to tailor the interface between the material and biosystem. Is it possible to use successive transitional layers between the material surface and a cell lipid membrane to gradually change the material characteristics from inorganic to organic? What materials should be chosen? What is the nature of the atomic level binding at the interface? How could the integrity of each material interface be preserved when exposed to various kinds of biofluids? Can the interface be functional as well as structural? What kind of transduction mechanisms can be produced in the substrate material, interfaces, or surface material to study the biosystem? Can the transduction mechanism be made specific and highly sensitive to a particular molecule or cell? If such interfaces can be developed, it would enable platforms that integrate a microfluidic-optical detection system on a chip, for example, for whole blood measurements that can probe the immune system response to pathogens.

3.1.

Approaches for the Ideal Material/Biosystem Interface

One approach by which to tailor the interface between the semiconductor material and biosystem is illustrated in Figure 3. It is an atomic/molecular schematic of successive transitional layers between the semiconductor and a cell represented by a lipid bilayer membrane. An initial overlayer of silicon dioxide is applied to the semiconductor. For compound semiconductors this is usually a deposited layer. Alternately, it is a thermally grown oxide in the case of silicon. The oxide layer serves several purposes. First, it is electrically insulating (if the oxide layer has uniform coverage, a condition not always met by deposited layers) to inhibit electronic conduction into the biofluid. Second, it is optically transparent over the spectral range of emission from the semiconductor (visible through near IR for compound semiconductors). Third, the oxygen rich surface provides good bonding to siloxane terminal groups of organic polymers molecules that are useful for the biocompatible interface.

Figure 3

Atomic/molecular view of the interface between a cell membrane and a surface functionalized semiconductor material.

Those molecules are applied by self-assembly or spin-casting layers that make a transition from inorganic to organic chemistry. The terminal group is preferentially hydrophilic for compatibility with the biofluid. Other overlayers like polydimethylsiloxane are spin cast and oxygen plasma treated to make them hydrophilic as well. Aliphatic and siloxane groups are usually transparent over the spectral range of interest, can be kept thin (tens of nm to μm range), are uniform in coverage, and insulating. Further, thicker organic overlayers are flexible and elastic, allowing contact sealing through interfacial bonding with other surfaces.

Thus, the interface between a semiconductor and biosystem is created with a tailored, insulating, optically transparent multilayer that makes transitions from inorganic to organic, dry to wet, crystalline to assembled molecules, toxic to biocompatible, rigid to flexible, and light emitting to light transmitting.

3.2.

Preparation of Polydimethylsiloxane Functionalized Surfaces

In previous studies, we have examined various organic layers for biocompatibility including bare SiO ₂, polymethylmethacrylate, polyethylene glycol, hexamethyldisilizane, silane, octadecyltrimethoxysilane, and N -(triethoxysilylpropyl)-O-polyethylene oxide urethane which are described elsewhere.^{10 11} The present studies have used polydimethylsiloxane (PDMS) polymer layers¹² to protect the compound semiconductor while interfacing with the biofluid. PDMS is cured by an organometallic cross-linking reaction. The siloxane base oligomers contain vinyl groups. The cross-linking oligomers contain at least three silicon hydride bonds each. The curing agent contains a proprietary platinum-based catalyst that catalyzes the addition of the SiH bond across the vinyl groups, forming Si–CH ₂–CH ₂–Si linkages. The multiple reaction sites on both the base and cross-linking oligomers allow three-dimensional crosslinking. One advantage of this type of addition reaction is that no waste products such as water are generated. If the ratio of curing agent to base is increased, a harder, more cross-linked elastomer results. Heating will also accelerate the cross-linking reaction.

Sylgard® 184 silicone elastomer and curing agent were purchased from Dow Corning. A mixture of elastomer and 10 wt  curing agent was prepared and allowed to cure at room temperature for 2 h. After 2 h, decane was used to dilute the polymer solution at a 2:1 polymer to decane ratio. Next, in a laminar flow hood, the polymer solution was mixed and filtered through a 1 μm syringe filter into a clean, dust-free container. SiO ₂ -coated wafers were cleaned in Piranha solution (1:1 hydrogen peroxide/sulfuric acid), rinsed thoroughly with de-ionized water, and dried. In the laminar flow hood, the wafers were dusted with a CO ₂ duster, placed on the spin coater, then 80 μL of polymer solution was placed on the wafer and the wafer was spun at 6000 rpm for 30 s. The wafer was removed and dried flat at 90 °C at 15 mm Hg for 24 h. The more porous PDMS films were made in a similar way, but with the following differences. They were diluted with tetrahydrofuran before curing and dried under vacuum at room temperature. These films were noticeably softer and did not perform as well as the former, as evidenced by the data described below.

5.1.

Experiment Setup

To examine the role of protective overlayers against optical degradation, we performed light emission experiments as a function of the current injection and concentration of ionic species as illustrated in Figure 5. A flow cell comprising a semiconductor light-emitting structure was injected with biofluid. The light emission from the semiconductor was collected with a fiber optic and relayed to an array spectrometer. The spectrum was studied as a function of ion concentration in the biofluid and of current injection into the semiconductor light emitter shown on the left side. The array of light emitters comprises a resistive metal contact that graduates the current injected into the emitters over two orders of magnitude.

Figure 5

Material microarray and microflow apparatus to study degradation of light emission from a semiconductor material overflown with a biofluid of controlled concentration of ionic species.

5.2.

Experimental Measurements of Light Emission Degradation

Figure 6(a) shows the measured light output as a function of time for overlayers of SiO ₂ only, PMMA on SiO ₂, and porous and hardened PDMS on SiO ₂. The first overlayer is ineffective in insulating the device from degradation by de-ionized water alone. PMMA offers some resistance to degradation, but it is not long lasting. PDMS layers are much more effective and show little degradation over the first half hour of use. The hardened layer of PDMS is very effective in preserving the material for many hours of operation.

Figure 6

(a) Time dependence of the light intensity degradation in a semiconductor light emitter by exposure to a biofluid. The different curves represent different protective overlayers on the semiconductor surface. The layer material and solution concentration are labeled on each curve. (b) Time dependence of the light intensity degradation for a semiconductor protected by an overlayer of porous PDMS and exposed to various concentrations of ions in a biofluid. The different curves represent different ionic concentrations of the buffered salt solution. The layer thicknesses were approximately 20 microns.

Figure 6(b) shows the effects of varying the ionic concentration of the buffered salt solution from de-ionized to full strength for the case of the porous PDMS layer. The de-ionized water has little effect on the light emission over the 30 min time period investigated. With the 0.1 solution, a slight 5 degradation can be observed in the data. With a full strength solution, the degradation is quite rapid until the device emits no light after 20 min. These data suggest that the ions in solution cause the degradation and not the de-ionized water alone. However, it may be the additional exposure time to water that induces the degradation. The spectral changes from de-ionized water to 1×PBS are shown in the inset of Figure 6(b). The light intensity decreases while the spectra shape remains relatively unchanged, indicating that the type of radiative recombination mechanism is constant but decreasing in efficiency. This suggests that the defect-induced recombination mechanism illustrated in Figure 4 is either nonradiative or is emitting light of wavelengths beyond the sensitivity limit of the diode array spectrometer (900 nm).

6.1.

Secondary Ion Mass Spectroscopy Analysis of Atomic Species in a Semiconductor

After the optical degradation experiments were performed, the materials forming the microcavity were disassembled for analysis of the material surfaces. A postmortem analysis of cation-like species by secondary ion mass spectroscopy of the degraded semiconductor light-emitting material is shown in Figure 7. Here, three different structures corresponding to different levels of current injection were analyzed for Na and K penetration into the active region of the p–n junction using an oxygen sputtering source to profile the material. The graphs also include Al and As compositions to help locate the active region of the structure (indicated by the vertical dashed line through each graph). The upper graph is the result for a virgin wafer with no injected charge and no exposure to biofluid. The middle graph is the result for 1 mA injected for 1 h in the presence of biofluids. The lower graph is the result for 10 mA injected for 1 h. The virgin wafer shows only background noise levels near 10¹³ cm ⁻². The latter two profiles show increasing cation density (from mid- 10¹⁴ to mid- 10¹⁵ cm ⁻²) in the active region. The Na penetrates more deeply than the K, as expected from the smaller ionic radius of K shown in the inset of Figure 4.

Figure 7

Secondary ion mass spectroscopy profiles of semiconductor structures as a function of the total charge injected.

A separate postmortem analysis of anion-like species by secondary ion mass spectroscopy is shown in Figure 8. Here, a virgin wafer and an expired device structure were analyzed for Cl, C, O, and P penetration into the active region of the p–n junction using a cesium sputtering source to profile the material. These graphs also include an Al composition profile to help locate the active region of the structure (the region where the Al composition dips a few tenths of a micron from the surface). The upper graph is the result for a virgin wafer with no injected charge and no exposure to biofluid. It reveals a surface region (within 100 nm of the surface) where all elemental contributions are artificially high due to contamination. Beyond the surface region, the C profile is high at 10¹⁹ cm ⁻³ until 0.3 μm, because it is an intentional p -type dopant. It is about 10¹⁷ cm ⁻³ through the active region and modulates with the Al composition in the subsurface layers. This modulation is unintentional and apparently arises from the growth conditions of the subsurface layers. The O, P, and Cl contents are nearly constant at mid- 10¹⁷, mid- 10¹⁶, and 10¹⁵ cm ⁻³, respectively, throughout the structure. The slight modulations in these elemental profiles are artificial and arise from the data normalization procedure.

Figure 8

Secondary ion mass spectroscopy of a virgin wafer (upper plot) and an expired device (lower plot). Here, elements O, C, P, and Cl from the right side of the periodic table are analyzed using a cesium sputtering source. The uppermost curve is the Al composition and it defines the active region (dip area) and other regions of the structure.

The lower graph is the result for an injected device that degraded to the point of no light emission. The active region of this structure (dip region) is nearer the surface at 0.2 μm. The profiles show that, near the active region, the elemental species exhibit an exponential tail and have all increased at least a factor of 10 in the active region during operation of the light-emitting diode. A notable exception is the O profile which appears to have saturated in the active region at a value nearly 100 times that of the virgin wafer. It is likely that this saturation represents a conversion of the AlGaAs materials to a partially oxidized material that inhibits the electron-hole recombination process. The very high O content in the action region is many orders of magnitude above other species. It is likely that O plays a critical role in the degradation process, either alone or in forming a complex with other species. The source of the O is probably decompositon of water molecules as opposed to dissolved O ₂ gas.

6.2.

Atomic Diffusion into a Solid

Interstitial atoms/ions must squeeze through openings between interstitial sites to diffuse into a crystal. The activation energy for diffusion is the energy required for these atoms to squeeze through the small openings,

The diffusion profile described by Eq. (1) predicts a Gaussian-like tail from the surface into the semiconductor. On a semilogarithmic plot this distribution would appear as an inverted parabola. The data of Figures 7 and 8 in the expired devices show profiles that, away from the surface contamination spike region and background noise region, have negative curvature. We can fit these regions to extract a diffusion constant for the various atomic species. The results are Na 13, K 7.2, Cl 4.5, and P 1.1 in units of 10⁻¹³ cm ²/s. Both cationic and anionic species moved into the semiconductor lattice, suggesting a lack of electric field dependence and movement in the form of atomic rather than ionic species. However, we expect that the electric field force on an ion is higher than the diffusive force on atoms in a gradient of 10¹⁷ cm ⁻³/μm. It is interesting to note that the rank order of the diffusion constants is inversely related to the rank order of ionic radii for these species. The data for O and C are more difficult to analyze since their profiles are more complicated. As mentioned, C was an intentional dopant, and O has probably reacted with the semiconductor atomic species to form metal oxides.

It is constructive to measure the total atomic accumulation in the active region as a function of the charge density that passes through the material. These data are displayed in Figure 9 for cations Na and K, and reveal the accumulation of ions in the active region as the injected charge increases. As expected, the Na diffusion constant is larger than that for K and differs by a factor of about 2.

Figure 9

Measured cation density in the active region of the semiconductor structure vs injected charge density. The fitting lines determine an effective ionic diffusion constant for potassium and sodium.

Future Applications of Biomedically Implanted Optoelectronics

A very new area of research is the development of light-emitting materials and structures for implantation into animals or humans for the purpose of detecting disease, monitoring progression, and therapy. Because bioMEMS devices can be made so small, it is feasible to consider implanting them into the body at strategic sites like lymph nodes, near blood vessels, prostate or cervical tissues, or even in the pancreas or gastro-intestinal tract as illustrated in Figure 10. Because the devices can be microscopic, they can be minimally invasive to tissues.

Figure 10

Schematic diagram of strategic locations for optoelectronic implants into the human body, including lymph nodes, ex vivo or in vivo blood vessels, gastrointestinal tract, pancreas, intestines, or prostate gland.

Applications include in vivo flow cytometry and ex vivo dermal contact cytometry of blood. Internally implanted devices might be coupled with molecular recognition sensors to trigger optical readout through transparent tissue. The light beams would be intense, and exhibit much light scattering, so it is preferable to locate them close to the dermis. Issues arise about how to anchor them in place and avoid migration in the body, or how to supply power (implanted battery, rf coupling). It is even possible to consider using the material as an internal power source by coupling external light into a photovoltaic junction. It may be possible to include both photovoltaic and photo-emitting action in the same implanted microdevice. In each of these cases, biocompatibility becomes a major consideration and it is necessary to determine if the materials induce harmful effects when they are in intimate contact with living cells. Some of the possible negative effects may include activation of an immune response, hemolysis, gene mutation, carcinogenicity, tumorogenicity, inhibition of cell growth, irritation, and toxicity.

This thrust represents the advent of a fertile new uncharted area of research that interfaces between materials science and living systems at the micro- and nanoscale levels. Biocompatibility of micro- and nanostructured materials would need to be studied in some detail from both the materials and biological perspectives to discover ways in which to mitigate detrimental responses of the body to foreign atomic species. Much of the initial research would be done in vitro with cell cultures to understand the basic mechanisms of cytotoxicity and genotoxicity. Questions arise as to the type and form of atomic/ionic/molecular species that detach from material surfaces and become incorporated into the cell membrane, nucleus, cytoplasm, or organelles and destabilize the metabolic cycle. On the biological side, research might develop new assays of chromosomal aberration to assess the impact of atomic/ionic/molecular species. Other issues that arise from the biological perspective are carcinogenicity, the role of bacterial endotoxins from material surfaces, and hemocompatibility including material effects on coagulation, hemolysis, and thrombosis. Ex vivo dermal contact materials would stimulate the need to study irritation and acute toxicity. As this field develops, implantation studies in suitable animal species would unfold additional domains of material biocompatibility.