University of California, San Diego Main PageSailor Group HomeDepartment of Chemistry and Biochemistry, UCSD

Home Announcements Research People Publications Contact information
Who coined the term "Smart Dust" Dr. Kris Pister, Professor of electrical engineering at UC Berkeley first coined the term in 1996. He was referring to "an autonomous sensing, computing, and communication system" that is "packed into a cubic millimeter mote (a small particle or speck) to form the basis of integrated, massively distributed sensor networks.”-Computer, 34, 44-51 (2001).
In your 2003 PNAS paper, I understand how chemical substances bond with the particle surface and create a reaction, and I understand how particles can be crafted to "stick" to a material. But how does this in turn lead to the possibility of self-assembling micro- or nanoscale robots?
Think of the particles as something akin to a virus: they can move through their environment by random diffusion (they don't have little arms, legs or propellers yet). When they find a surface they recognize (the chemistry on the surface is complementary to their surface chemistry), then they stick.

Self-assembly: They are not as sophisticated as transformers. They have the ability to assemble into a larger, more complex structure, but not into anything like an anthropomorphic being. The main thing we demonstrated was that the larger structure sets up a coherent mirror, to allow the complex assembly to signal to the outside world that it has found the surface more easily than would an individual particle. This can be thought of as operating like a choir; just as an individual voice cannot be heard as easily as all the voices in a choir singing together, an individual particle has a very weak "voice" (it is not easily seen by the eye or by a spectrometer because it is so small). When many particles assemble on the interface, all of the reflectivity properties combine into one big signal that is easily seen by eye or with a spectrometer.
Your 2003 PNAS paper shows how the reflectivity spectra of the dust changes in the presence of hexane. I assume that different compounds would change the reflectivity spectra in different ways. Let's take a hypothetical "real world" situation, like Smart Dust being used to look for pollutants in effluent from industry. You would likely want to monitor for more than one compound, and maybe you would even want to be alerted if there were something there you hadn't expected. Would you need different types of Smart Dust mixed together, say one for each compound, or would a single type of Smart Dust be used to detect a variety of compounds?

There are two types of smart dust that we use. For most of the biological and biomedical applications, we place very specific chemistry, usually in the form of an antibody, in the pores. This makes the material very specific for one target molecule, which is useful for many of the medical applications. In the other general approach (and in the PNAS paper), we use fairly non-specific chemistry, in the form of an aliphatic hydrocarbon. This is a greasy molecule (technical term is hydrophobic) that is attracted to many other organic molecules, such as fats, VOCs (volatile organic compounds), PAHs (polycyclic aromatic hydrocarbons), and the like, providing a sensor for a wide range of organic pollutants. So in some cases we would use a cocktail of many different types of smart dust and in others we would only use one. The strength of our encoding strategy is that we can distinguish between many different types of smart dust simultaneously.
In order to read your smart dust to determine what chemicals it is seeing, would you need to have some type of computerized database that compiled the different spectra you would expect to see if there were hexane, benzene, toluene, etc, and then compare your sample to the database? How would this work exactly?

This approach would be used in the biomedical applications, where we would have a library of thousands of different types of particles each targeting a different molecule. The library would be read out by scanning the ensemble with a machine similar to the bar code scanner used in grocery stores. This approach is useful for high throughput screening applications.
This might be a silly question, but why does poring create a mirror? I would have thought the more diffuse the surface, the poorer the reflectivity, but maybe this is because I am thinking of things on a macroscopic scale. Maybe the pores are so small that you are only changing the wavelength of the reflected light, rather than causing scatter?

That is correct. If the pores get too big (>500 nm), all they do is scatter light and we lose the photonic crystal properties.
In the 2003 PNAS paper, if all the dust particles are facing in a particular direction, along the air-water interface for example, or clustered around a hexane droplet, how can you measure the spectrum of the side that is facing away from you?

We can make them thin enough such that they are partially transparent so we can read both sides simultaneously. In the PNAS paper we didn't want to do this because it would make it difficult to determine the particle orientation.
In the study in Nature Materials, you could tell when a binding rxn occurred because of fluorescence. But you mentioned that the code of the silicon particle changes when it is bound by a specific chemical. Is THAT is how you detect a binding reaction, and thus the presence of a specific chemical?
We did two separate types of experiments. The other one, on standoff detection (Adv. Mater. 2002, 14, 1270-1272) was published in late September 2002. In that paper we take advantage of three properties of the dust: the codes themselves, the ability of this porous matrix to change it's code on binding an analyte, and the ability of the very small pores (2 nm) to concentrate the analyte based on a phenomenon known as microcapillary condensation (it is similar to the physics that causes water to be drawn uphill against the force of gravity in a capillary: the vapors liquify when they are confined to a very small pore). The Nature materials paper (Nature Materials 2002, 1, 39-41) just uses the codes, coupled to a conventional bioassay.
Does one silicon particle have the ability to react with only one specific chemical, and then its wavelength changes and you know that that chemical is present? Or does there have to be many probes that each react with only one of the many chemical/biological agents that exist? There are two approaches here. Recognition based on specific interactions, such as those between an antibody/antigen pair. This is the same methodology used in home pregnancy tests and a large variety of biomedical assays. In that case one particle does only one assay. It is highly specific, but the disadvantage is you need to have a different batch of particles for each agent (of course you can mix all the batches together to make a cocktail if you like). The second approach is based on non-specific interactions, such as in van der Waals or hydrogen bonds. An example is the interaction that occurs between your skin and oils. Motor oil, suntan lotion, or olive oil all stick to your skin and are hard to wash off; even though they are very different molecules they are similar as a class in that they have relatively strong binding interactions with your hydrophobic skin. The paper attached uses particles that are hydrophobic so they bind nonspecifically to a broad range of hydrophobic VOCs (volatile organic compounds). The disadvantage with this approach is that we can't tell the difference between members of a class. For example acetone (fingernail polish remover), toluene (an additive to gasoline to improve octane rating) and hexane (another component of gasoline) were all detected by our smart dust. Water vapor was not detected by these particles, because it is not hydrophobic and so doesn't enter our pores. That is good in the present paper because it shows that we can see VOC pollutants without interference from atmospheric humidity changes.
Could your smart dust be used for weapon inspections in places such as Iraq? The short answer is yes, in principle. Because they can be encoded and then read remotely, the particles can be used as microscopic taggants to track and identify objects similar to how a barcode works on a package of cereal at the grocery store.

The particles are really microscopic machines. Our goal has been to incorporate functionality into these machines. They have nanoscale concentrators incorporated into their structure that allows them to absorb vapors from their environment. The codes change in a predictable fashion when this occurs, so we can use them as remote sensors. By incorporating a catalytic recognition element into the nanostructure, we have demonstrated that the porous material can selectively detect some of the G-type nerve warfare agents such as Sarin. The concept would be to covertly sprinkle the smart dust in the vicinity of a suspect site and then read it out sometime later to see if it has detected anything.

Other chemistry on these particles allows us to detect VOCs (volatile organic compounds). This could be used to find gas leaks or pollutants in water. Perhaps more relevant to your question, it can also be used (in principle) to detect diesel powered generator exhaust fumes which may be present in the vicinity of the ventillation shaft of an underground bunker.
Designed by Andrea Tao.
Main address: Department of Chemistry, University of California, San Diego, 9500 Gilman Dr., La Jolla, CA 92093-0358 (858) 534-0227
Send questions, comments, and suggestions to:

Last modified Monday, February 17, 2003