I am a graduate student at Boston University, studying for PhD.
After careful thoughts, I found that the research topics that best
fits my background and that's fun and challenging enough is nano-scale computing.
I have joined Professor Mohanty's research group.
Part of my past research can be found here.
Quantum computing
In the rush for computing power, the historical trend for the last 35 years
was to shrink the computation "building bricks" (typically transistors and resistors),
to pack more into the same volume. However, we've got to a point where this approach breaks down.
At nano-scales, quantum mechanics installs as the appropriate describing language. For now, circuit
designers fight to counteract these effects, as they manifest as undesirable leakage in MOS transitors.
This brings up the concept of quantum computing - why not using these new (from a
computing point of view) properties for computation?
To exploit quantum mechanics computing power, one has to re-think
about everything known from classical computing. There are ideas
suggesting that quantum computers can have an enormous processing
power, at least for some specific applications (factoring big numbers, for example). Some of these
quantum computing algorithms were actually tested experimentally - with NMR quantum computers for
example - and were proven to work.
However, designing efficient quantum algorithms seems to be a difficult task. Only
a handful of such algorithms were imagined so far.
And what's worse, from an experimental point of view, handling quantum states proves to be at, or slightly
beyond, the limits of today's experimental capabilities.
There are only few physical systems on which some control over quantum states has been demostrated:
NMR - Most advanced results, but non-scalable and thus not promising on the long run.
Superconducting junctions - One of the most promising. Quantum gates are close to be demostrated
experimantally. Might scale relatively well.
Optical cavity and optical photon computers - Not easily scalable. Lack of very non-linear
optical materials makes them difficult to implement.
Ion traps - Some of the most beautiful Physics experiments in recent years. Wonderful demonstrative results,
but difficult to scale up.
The main idea behind the research I'm doing is to enable yet another technique for quantum control,
exploiting coherent transport phenomena through diffusive metals.
At low enough temperatures (<1K), the phonons are "frozen", and the main decoherence mechanisms for
electrons diffusing through normal (non-superconducting) metals are electron-electron interactions and
phase-destructive scattering on deffects in the lattice. In particular, "magnetic" impurities are
very effective at destroying phase coherence, and are highly unwanted.
In such conditions, electrons can maintain phase coherence over macroscopic distances (microns), and thus
it is possible to design techniques to actually measure parameters characterizing the coherence of the
electron system.
One might think to exploit coherent diffusive transport for quantum control. Unfortunately, coherence
times are very small - between tens of ps and a few ns. The electronic equipment currently available is
not fast enough to implement quantum control over such small intervals of time.
Following the example of NMR, we are trying to use external electromagnetic fields to increase the
coherence time, as a first step towards making these systems practical for quantum control. However,
the interaction between electromagnetic fields and coherent electrons in diffusive metals is not very
well understood, and the first task is to study this interaction.
Coherent transport phenomena
The quantum coherent transport phenomena investigated are Aharonov-Bohm oscillations and weak localization.
From these measurements we extract coherent transport parameters, and we evaluate
the influence of external electromagnetic fields on these parameters.
Aharonov-Bohm oscillations in micron-sized rings
Electrons split wavefunctions when reaching the ring, and the two packets
propagating through the two branches interfere on the other side of the ring. By
applying a variable magnetic field, the phase difference between the two interfering
wavefunctions can be tuned up, and the sample exhibits oscillatory magnetoresistance. This
is how the so-called Aharonov-Bohm oscillations form:
Weak (anti)localization
Another effect of coherent transport is the slight increase (decrease) in resistance
around zero magnetic field because of the time-reversal symmetry breaking due to the magnetic field.
This is known as weak (anti)localization.
Weak antilocalization, Aharonov-Bohm oscillations (with period h/e) and weak antilocalization
oscillations (with period h/(2e)) can be observed here.
Setup and measurements
A 4-probe differential setup is used to measure the low frequency (17Hz) measurement of the resistivity of
the sample:
Two gates are capacitively coupled to the two branches of the ring, and the effect of various
electromagnetic fields applied with the gates on coherent transport through the ring is measured. Typical
signals applied to the gates are sine waves in the GHz range.
External fields were found to decohere the electron system for most configurations. Below are typical plots of
Aharonv-Bohm oscillations and their Fourier transforms (to clearly see the Aharonov-Bohm oscillation
amplitude), and their dependence on the power of the fields applied to the gates. The peak vanishes with
increasing power:
However, some configurations were found in which the measurements suggest the external fields might actually
increase coherence. In the plot below, increasing external field's power increases the amplitude of the
Aharonov-Bohm oscillations, for very specific ranges of frequency and power. However, the increase is
small (~25%) and comparable to experimental noise. More measurements are needed to confirm this.
An interesting behavior of the resistance of the sample was found as a function of the external fields.
For some specific ranges of frequency and power of external fields, the resistivity shows a minimum. The
fact that the position of this minimum is correlated with the Aharonov-Bohm amplitude suggests that this
behavior builds upon coherent phenomena, but how exactly this happens remains to be
investigated.
By varying both frequency and power of the external fields, regimes of most efective coupling between the
electron system and the external fields were identified.
The relative amplitude of the Aharonov-Bohm oscillations is very small, typically of the order of
1e-5 ... 1e-6 of the total resistance of the sample. On the other hand, the currents used are
small (~10nA) in order to avoid heating. These make the measurements relatively difficult. Low noise
measurement design is a must: shielded room, low noise equipment, proper ground design.
Here's a photograph of the equipment used:
One of the main problems encountered was that the
cryostat itself was generating noise that was easily masking the oscillations. "Hot" liquid helium at 4K
pouring in a cold pot with liquid helium at 1.8K was generating shock waves that made the pot vibrate.
As the pot was placed in magnetic field, and the sample was placed near it and suspended by it, the currents
induced in the vibrating metallic components were big enough to generate a noise with an amplitude of
one order of magnitude bigger than the Aharonov-Bohm oscillations. That's why the fridge needed to be
operated in a special, non-standard mode.
The samples
I am building these rings by electron beam lithography. Typical dimensions are
~1 micron on a side, 30-75nm wire width and thickness equal to width.
The typical metals of choice are Au or Ag. As for the substrate, I use SiO2.
Here's the typical fabrication process:
Here are some typical samples:
Ag/SiO2 1.5micron on a side, 70nm wire width No underlayer
Au/SiO2 500nm on a side, 35nm wire width No underlayer
20nm Au wire No underlayer
Here are some of the few hundred failed samples.
And here are some non-standard applications of an SEM :-).
The most difficult fabrication step was the liftoff. Both Au and SiO2 are
very reluctant at adhering, and the rings are easily destroyed during the last fabrication step. An
intermedate "sticky" layer (a few nm of Ti or Cr) vastly enhances the adhesion, but induces scattering sites that destroy
phase coherence, so this solution was avoided. After some practice, the success rate
for Au/SiO2 samples (without underlayer) was big enough to make measurements possible.
