Wednesday, February 06, 2008

Combined single-molecule electronics and optics

This'll be my last self-referential post for a while. Now that it's out online, I want to write a post about our latest result. As readers of this blog know, I am not a big fan of the hype that accompanies a lot of nano research. I cringe everytime someone claims that a minor development is a breakthrough, and it drives me crazy when people who know better feel compelled to imply that self-reproducing nanobots are going to build spaceships out of single-crystal diamond in five years. That being said, I really do think that this result is a major advance, both in molecular-scale electronics and in ultrasensitive chemical sensing.

The two-sentence summary: we can do simultaneous electronic and optical measurements on single molecules (!) by using our electrodes as optical antennas. This opens up lots of science to be done as well as some very intriguing technological possibilities.

Over the last decade, a number of techniques have been developed to measure electronic conduction through single molecules. There are lots of basic physics and physical chemistry questions that still need to be answered in such systems (e.g., how does dissipation work at these scales? What happens when electronic correlations are strong and the system is driven out of equilibrium?). One long-standing problem, though, has been the lack of any independent (non-transport) way to confirm that conduction is taking place through a particular molecule of interest. Except for scanning tunneling microscopy (great for science, but impractical for some measurements and definitely not scalable for devices), there are no good imaging techniques to see the object (molecule of interest? contaminant? accidental nanoparticle?) through which the current is passing. The resulting approach to these devices has been essentially statistical, requiring the fabrication of large numbers of devices with many control experiments, etc.

Over the same period, as I discussed here in reference to an earlier paper from our group, surface-enhanced Raman spectroscopy (SERS) has been studied extensively. Raman spectroscopy is a very common physical chemistry technique to probe the vibrational spectrum of materials. Light comes in at some frequency, dumps some energy into the vibrational modes of the material (this is called Stokes scattering), and leaves with less energy. By measuring the energy shift between incoming and outgoing light, it's possible to pick out a material's characteristic vibrational modes - a kind of chemical fingerprint. In SERS, nanostructured metal surfaces act like little optical antennas when illuminated, creating so-called hotspots where the local electromagnetic intensity can be as much as a million times greater than the incident intensity, leading to greatly enhanced Raman emission. People have reported SERS capable of measuring single molecules, but demonstrating that conclusively is extremely difficult.

In our new paper, we've been able to kill two birds with one stone. We have been able to perform simultaneous electronic transport and Raman spectroscopy on individual molecules. The same metal electrodes used to push current through the molecules also function as a plasmonic antenna, giving enormous SERS enhancements. Conduction between the electrodes is known to be by tunneling, and tunneling depends so steeply on distance that the total volume through which the current is passing can only contain at most one or two molecules. (This steep distance dependence is the reason STMs work!) At room temperature (and in air), we see that the conduction from one electrode to the other bops around a bit as a function of time. This is due to molecular motion and the changing molecular environment, and isn't surprising. However, we can simultaneously measure the Raman signal from the region between the electrodes. We find that the time variation in the Raman emission correlates extremely well with the time variation in the interelectrode conduction. Since the conduction occurs via tunneling and probes about one molecular volume, the Raman emission must be from the same single molecule in question.

This demonstrates that we can mass-fabricate single-molecule sensitive SERS hotspots in high yield in pre-defined locations. At the same time, this multimodal single-molecule sensing shows via the Raman signature that we are pushing current through the specific molecule of interest in a given device.

We've got lots of ideas on where to go with this - it's very exciting.


Henry said...

Hi Doug,

what about decoherence near T=0 in nanoscale devices? Is a hot debated topic yet? Found an answer?



sylow said...

Well, where do you see molecular electronics going from here in the next 5 to 10 years, doug? Is it going to boom?

Doug Natelson said...

Henry - I think that the debate on this is pretty much over, though that doesn't mean that everyone agrees; it just means that there are two camps that have fixed views. The one thing that I think all would agree on: if you can properly describe the eigenstates of the system, there is no such thing as T=0 decoherence. At T=0, the system does find a quantum mechanical ground state that, by definition, is stationary. Now, there are still some subtleties out there about the temperature dependence of weak localization, particularly in the presence of dilute magnetic impurities, especially those with higher spin than 1/2.

Sylow - I think that there is a great deal of science that will be done with such structures over the next 5-10 years. I think that there are real opportunities for ultrasensitive chemical sensors. I think it's unlikely that we'll use single-molecule devices for high speed computing, but I do think that some of the lessons we learn in studying those structures will be applicable to ultrascaled Si devices.

Anonymous said...

Your results sound great. It is very exciting to have a technique that allows to characterize what is measured between the electrodes. I´m looking forward to read more details in the nano letters paper.

You wrote that you have a lot of ideas. Could you explain some of them?

Anonymous said...

" it drives me crazy when people who know better feel compelled to imply that self-reproducing nanobots are going to build spaceships out of single-crystal diamond in five years."

Does that mean you and Jim Tour fight a lot, or are you excluding nanoputians and nanocars from this discussion? :)

Some obvious questions: Is it possible to do these experiments at 4K? (is the room temp vs. low temp geometry the same for that system?) Could you do sequential or simultaneous Raman/IETS measurements? Here I'm thinking sequential might be interesting, because the Raman spectrum and IETS spectrum should show similar peaks. Simultaneous might be interesting to see how the vibrational excitations from current flow and those from optical excitation interact with each other. Also, if you hit the system w/ laser pulses at the most substantial IETS frequency, can you generate a switch with a ridiculously small on/off ratio? No practical use in this, but it would be fun to try!

Doug Natelson said...

Sorry about the delay in posting. I've been working on getting papers out the door.

Anon1: One of the things I'd like to do is to work at lower temperatures, to slow down a lot of blinking and spectral diffusion caused by thermally activated changes in the molecule's environment. Then we can use the gate electrode to look in detail at things like the charge-state dependence of the vibrational mode energies, and the charge-state dependence of the Raman enhancement of the various modes. This should let us study the still-murky issue of "chemical enhancement" in SERS. I'd also like to look at Stokes/antiStokes intensity ratios, both to see if we're optically pumping the molecule, and to see what happens to mode populations when we push current around. That's a sampling of a few ideas....

Anon2: Jim Tour and I agree to disagree about some things :-) To be fair to him, he's never claimed that we'd have nano-assemblers, as far as I know. On the science side, I think it will be very interesting to do measurements at lower temperatures, as I said above. If you review my IMR proposal to buy the necessary hardware, please give me an "E" :-)
We already see some other interesting optical effects on transport, and I'm sure that there's much more to learn.

Doug Natelson said...

One more point.... Studies like the ones I mentioned, particularly looking at vibrational populations, are going to be important in the long term. They are a way of looking at dissipative processes at the molecular scale, and similar physics will be relevant even in ultrascaled silicon devices.

Anonymous said...

Hi, Doug

What you are talking about is not a single molecule Raman spectroscopy.
Read latest papers on SM Raman by Eric Le Ru and Pablo Etchegoin. Large fluctuations in spectra does not mean that it is single molecule. At present there are no direct prove of Single Molecule Raman spectroscopy

Doug Natelson said...

Anon3 - Read our paper before knocking it. I know that blinking doesn't necessarily imply single-molecule spectroscopy.

The point is, conduction in our devices is a single molecule effect. The conduction and the Raman track each other. The simplest explanation is that both are due to the same individual molecule. With coauthors we've also done modeling to rule out other possibilities. I know Etchegoin's work, and he and I have exchanged email about this.

Anonymous said...

Hi, Doug
I red carefully yours article and still not satisfied about Single Molecule Raman claims made in yours article.
According to Eric Le Ru direct prove of SM Raman need two things: 1. very clear physical condition to prove that just one single molecule event (as opposed to few or many) is contributing into specific Raman spectra, 2. To present clear Raman spectra from this single molecule even with at least Mln CPS per milliWatt (as oposed to marginal fluctuating 2.5 CPS per milliWatt from recent JACS article of Richard Van Duyne.
Unfortunately I never seen in yours article clear Raman spectra from single molecule event with big counts.
There are a lot of articles including Bi-Analyte method of Eric Le Ru and many Meixner group articles that provide INDIRECT evidence of SM Raman, and yours does not offer too much beyond that

Doug Natelson said...

Anon3 - I agree with everything Le Ru says, if you're trying to conclude "single molecule" based on Raman spectra alone. Indeed, in our earlier paper (NL 7, 1396 (2007)), where all we had was localized Raman emission with blinking and spectral diffusion, we explicitly said that one cannot conclude single-molecule effects.
If you just have Raman spectra to work with, you need lots of statistics (the high count rates Le Ru discusses) and cleverness (e.g., the bianalyte approach).

In our experiment the conduction is a single molecule (or at most, two molecule) phenomenon because of the local nature of tunneling. The fact that the tunneling conduction (dependent on roughly a molecular volume) and the Raman emission track each other in time extremely well is very very suggestive. If you have an alternative explanation for how this can happen aside from our interpretation, I'd be happy to discuss it. We're not claiming single-molecule SERS based just on the Raman spectra; it's the fact that Raman emission tracks what is known to be a single-molecule phenomenon that's the essential idea. If you don't find that persuasive, then I guess you won't agree with our interpretation.

Anonymous said...

Hi, Doug
I understand idea about correlation of conductance data with Raman data .
However there are still no clean direct evidence that Raman spectra are generated by just single molecule event ( as opposed to few or many).
Still it will help to see clear Single Molecule Raman spectra with all metrics : counts per second per milliWatt and SM enhancement factor.
That will be usefull to compare across data of other groups

Doug Natelson said...

Anon3 - Could you explain the rationale for comparing CPS per milliwatt across different measurement platforms? In our system there are losses at fiber couplings and filters, for example, that would be different in a different system, and the number of counts will also depend on the grating and the CCD. Just looking at a Si wafer and counting on the 520 cm^-1 line, you would see different cps per mW in a WiTEC than in a Renishaw, for example.

That being said, typical SERS counts per second per milliwatt that we end up seeing in our system are on the order of a few hundred. SERS peaks and their fluctuations can both be significantly larger (10x, say) than the magnitude of the 520 cm^-1 Si Raman line for the substrate.

I would be happy to discuss this with you further via email, if you like. I'm not sure that putting lots of nitty gritty details is of interest to the overall readership.

sk said...

Hi Doug
Could you please show a few conventional Raman spectra of x-wavenumber and y-intensity from Figure3 and Figure4's waterfall plots? I would like to see the peak width, the intensity fluctuation, and the background signal. It's too much for my brain to process the color-coded images and get those values.
Thanks a lot!

Doug Natelson said...

sk - Sorry about the long delay in responding; the March Meeting ate some of my time. I have the data in question all set - I think the easiest thing for me to do is to put it up on the web, perhaps as a pdf. Please respond if you still want to see it.

Standard dishwashers said...

Interesting and important information. It is really beneficial for us. Thanks

Online electronics store said...

Another informative blog… Thank you for sharing it… Best of luck for further endeavor too.