Silicon gets sensitive

 Porous silicon was accidentally discovered by scientists at Bell Laboratories in the mid-1950s while they sought techniques to electrochemically polish silicon wafers. To their astonishment, they found that certain conditions of applied current and solution composition drilled fine holes into the wafers. This did not provide the desired smooth polish, so these curious results were reported in a small Bell Labs technical note and pretty much forgotten, aside from the work of a few researchers.

Then, in the late 1980s, Leigh Canham at the Defense Research Agency (Malvern, UK) reasoned that the diaphanous silicon filaments generated when the pores become large and numerous enough to overlap might display quantum confinement effects. Canham's intuition turned out to be correct: The material fluoresces bright orange, a wavelength that is significantly blue-shifted from the bulk silicon band-gap energy. Interest in the material exploded because of the possibility that purely silicon-based optoelectronic switches, displays, and lasers were just around the corner.

However, problems with the material's chemical and mechanical stability, and the fact that no one could make an efficient electrical contact to the material, caused interest to wane by the mid-1990s. But in this same decade, the unique features of the material—large surface area within a small volume, controllable pore sizes, convenient surface chemistry, and compatibility with conventional silicon microfabrication technologies-- led to applications beyond the optoelectronics field.¹ The material's specific surface area of a few hundred square meters per cubic centimeter corresponds to about a thousand times that of a polished silicon wafer. This characteristic makes porous silicon a convenient material from which to fabricate biosensors, chemical sensors, bioresorbable materials for drug delivery, on-chip separators, micro-electro-mechanical-systems power supplies, and cellular microphysiometers.

Porous silicon is created by electrochemically corroding silicon in solutions containing hydrofluoric (HF) acid. The pores propagate primarily in the <100> direction of the crystal. By changing the electrochemical current, the electrolyte composition, and the dopant characteristics of the wafer, one can tune the average diameter of the pores from a few nanometers to several micrometers. Tuning the pore diameters and chemically modifying the surface allow developers to control the size and type of molecules adsorbed.² Researchers exploit these properties to develop porous silicon sensors to detect toxic gases, volatile organic compounds, explosives, DNA, and proteins.

One of the unusual features of the material is that the porous layers generated in the electrochemical etch can be smooth enough to create optical interferometers, Bragg filters, and other high-quality optical structures.³ The physical characteristics of reflectivity-based sensor devices are determined by the optical thickness of the films, which is the product of the refractive index (n) and the thickness (L). The electrochemical parameters used in the synthesis control both n and L, allowing developers to etch thin films with given optical parameters in a reproducible way. For example, one can produce a porous silicon Fabry-Pérot film with two planar and parallel interfaces. The device can produce high-contrast optical fringes. Shifts in these fringes occur when an analyte binds to the surfaces in the pores, providing a sensitive transduction modality.⁴

One can also construct more elaborate optical structures based on multilayers of porous silicon.⁵ The multilayers can be prepared by periodically varying the current density during the etching process. This transfers the current versus time profile to a porosity (refractive index) versus depth profile, resulting in a stratified structure. Such multilayer structures act as 1-D photonic crystals with reflectivity maxima that depend on the refractive index gradient and the periodicity of the superlattice.

The sharp resonance features in these structures can lead to more sensitive chemical or biological sensors. For example, Fauchet and coworkers demonstrated femtomolar-level detection of single-stranded DNA using a related optical microcavity structure modified with a complementary DNA fragment.⁶

The desire to miniaturize and reduce the power drain of a sensor motivated our team's early efforts in the porous-silicon smart-dust project. Inspired by the efforts of Kris Pister at the University of California at Berkeley to construct dust-sized computer chips for use in remote-sensor networks,⁷ we developed techniques to place the necessary chemistry and optics of a rudimentary remote sensor on a 50-µm particle of porous silicon (see figure). The particles are optically encoded by incorporating a periodic rugate filter structure into the porous matrix and taking advantage of microcapillary condensation effects to concentrate analyte within the pores of the material. We can detect the binding of analyte from a remote distance using a low-power laser.⁸

Tiny particles of porous silicon can be used as sensitive detectors. For example, copper ions (Cu²⁺) bound to the porous silicon could react in the presence of chemical warfare agent Sarin (GF), producing hydrofluoric acid (HF).

In order to make these particles, we first prepare multilayered porous silicon films by electrochemically etching a <100>-oriented, polished silicon wafer in an ethanol/aqueous HF-acid solution. Modulating the current density periodically with a pseudo-sine wave generates the sinusoidally varying porosity gradient that creates the rugate filter. The films are then removed from the substrate by applying a pulse of current. We turn the freestanding films into particles by mechanical grinding or ultrasonic fracture, producing particles ranging in size from several hundred nanometers to a few hundred micrometers (see image).

The optical reflectivity spectrum of such structures displays a sharp reflection maximum at a wavelength and a source-sample-detector angle that satisfies the Bragg equation. The spectrum shifts in a predictable fashion upon analyte absorption, allowing the encoded particles to report on their chemical environment.

The multilayer structures generated in porous silicon provide a simple means of optically encoding micrometer-sized particles. Techniques to label small particles or beads have received increased attention in recent years because of their potential for use in high-throughput screening and bioassay applications. Strategies for performing a large number of assays while minimizing the necessary sample volume have focused on either on-chip spatially differentiated arrays or random distributions of encoded polymeric beads.⁹ Attempts to encode large numbers of polymeric or glass beads in random arrays or in fluid solution have employed fluorescent molecules or quantum dots as bits, in schemes that use each chromophore to represent a separate bit.¹⁰ This approach requires the user to incorporate and fix n different types of chromophores in a bead to achieve 2n different codes and is limited by chromophore photochemical stability and spectral linewidth. Researchers have also used submicrometer-size metallic rods containing stratified compositional variations as encoded supports.¹¹ These materials don't suffer from photo- chemical degradation effects, although they are more difficult to read in a high-throughput or parallel fashion.

In contrast, the multilayer structures generated in porous silicon provide a simple means of optically encoding micrometer-sized particles. By modulating the current density during electrochemical etch, one can generate a complicated porosity gradient in the film. Since the optical spectrum is essentially the Fourier transform of the spatial porosity variation, an arbitrarily complicated spectrum can be generated. Like encoding methods that use metal nanorods, the codes are fixed in the particle and cannot diffuse out or photobleach. Like fluorescently encoded beads, they can be read at a distance using fast and inexpensive optical-detection methods. In addition, because the particles present a silicon-dioxide surface to the environment, they don't readily quench luminescence from organic chromophores, and they can be handled and modified using the chemistries developed for glass-bead bioassays. Simple tagged antibody-based bioassays have been performed with these coded particles, demonstrating the feasibility of the approach.¹²

Silicon is receiving increased attention for use as a biomaterial. Crystalline silicon in particular has been used as a textured surface to guide cell alignment, to encapsulate cells for implantation, and as an electroactive substrate to stimulate excitable cells.^13-15 Porous silicon has been less extensively studied. The range of tunable pore sizes (2 to 2000 nm) in porous silicon spans a range of sizes important in biology—the size of a small DNA fragment is on the order of a few tens of nanometers, while proteins are generally in the 100-nm size range, and bacteria and cells can be as large as a few micrometers in diameter.

Researchers have begun to explore the use of porous silicon as a biodegradable material for the slow release of drugs or essential trace elements to cells or for in vivo diagnostic tests.¹⁶ Canham's research group observed hydroxyapatite nucleation on porous silicon in vitro, suggesting that porous silicon, in contrast to crystalline silicon, could be a bioactive surface.¹⁷

Two aspects of porous silicon are of particular relevance for bioimplant applications: It can be used as a sensitive biosensor for proteins, antigens, and DNA, and it can be modified with a wide range of biological or organic molecules. These two features should allow porous silicon to serve as a versatile biomaterial. Although effort in this area is still in early developmental stages, combining the biocompatibility of the material with its highly sensitive biosensing capabilities should lead to new applications in tissue-based bioassays, drug delivery, and health-monitoring applications. oe

Acknowledgements

The author benefited tremendously from discussions with Sangeeta Bhatia, Ronald Betts, Erkki Ruoslahti, and Tomas Mustelin. Most of the work described in this report has been funded by the National Science Foundation and the National Institutes of Health.

References

1. L. Canham, Properties of Porous Silicon, EMIS Datareviews, ed. B. Weiss,Vol. 18. (1997).

2. J. Buriak, Chem. Rev. 102[5]: p. 1272 (2002).

3. V. Mulloni, Z. Gaburro, et al., Phys. Stat. Solidi A 182[1]: p. 479 (2000).

4. M. Ghadiri, Porous Semiconductor-Based Optical Interferometric Sensor, USA (2001).

5. G. Vincent, Appl. Phys. Lett. 64[18], p. 2367 (1994).

6. S. Chan, et al., Phys. Stat. Sol. A 182[1], p. 541 (2000).

7. B. Warneke, et al., Computer 34[1], p. 44 (2001).

8. T. Schmedake, et al., Adv. Mater. 14, p. 1270 (2002).

9. A. Lueking, et al., Anal. Biochem. 270[1], p. 103 (1999).

10. W. Chan and S. Nie, Science 281, p. 2016 (1998).

11. S. Nicewarner-Peña, et al., Science 294[5540], p. 137 (2001).

12. F. Cunin et al., Nature Materials 1, p. 39 (2002).

13. S. Turner, et al., J. Vac. Sci. Technol. B 15[6], p. 2848 (1997).

14. T. Desai, et al., Biotechnol. Bioeng. 57[1], p. 118 (1998).

15. M. Colicos, et al., Cell 107[5], p. 605 (2001).

16. L. Canham, et al., Phys. Stat. Sol. A 182[1], p. 521 (2000).

17. L. Canham, et al., Thin Solid Films 297[1-2], p. 304 (1997).

nraveling the dynamics and mechanisms of protein folding may hold the key to understanding and treating diseases ranging from cancer to Alzheimer's, as well as improving overall quality of life. Researchers in Sandy Asher's group at the department of chemistry at the University of Pittsburgh (Pittsburgh, PA) have pioneered the use of ultraviolet resonance Raman (UVRR) spectroscopy to study protein structure.

"For the first time, we have applied nanosecond time-resolved UVRR spectroscopy for kinetic studies of the first stages in protein folding," says Asher. The group initiates the unfolding process with a rapid IR laser-induced temperature jump of the aqueous solution. The intermediate states of the relaxation are probed by time-delayed UV pulses of about 200 nm, which excite the transient UVRR spectra. By using tunable nanosecond UV lasers, Asher and colleagues excite amide resonance Raman spectra that give detailed information on protein structure.

"Now that we have mapped the genome, we can relate disease to a specific protein alteration," says Asher. "Many diseases are caused by protein misfolding. Insight to the details and consequences of that alteration may be useful for treating that disease." —Jeremy Weston, Positive Light