So how do you actually make electrical contact to tiny microchips? Wirebonding is the standard industrial technique. For large chips (e.g. complex processors) fabricated in foundries, the process is fully automated. However in small research labs, a manual wirebonder is used, as each chip tends to be different.
I was rather enamoured by various artsey and atmospheric pictures available on the internet of people’s wirebonding endeavours, so I thought I’d try and capture a few of my own:
The process is as follows: First the chip (the dark coloured rectangle) is glued down onto a chip carrier (the large light coloured surrounding piece). The carrier may vary in design depending upon how many contacts are needed to be made and what apparatus that the chip is to be mounted within. Specifically, for very low temperature experiments it is important to have the chip in good themal contact with a metallic chip carrier (preferably Copper). A more specialised chip carrier for use in a dilution refrigerator is shown below:
The chip carrier is held firmly in the wirebonder vice, and then the wirebonder tip (which is like a needle with the metallic wire threaded through) is brought down into contact with the surface of the bonding pads on the carrier. The tip undergoes ultrasonic agitation, and is sometimes heated, which mimics a welding of the wire to the the metallic tracks below. As the tip is then moved by the user the wire is pulled through the needle and brought down again to form the second bond to the chip, at which point the bonder also cuts the remaining end of the wire. You end up with a small, neat loop of wire between the chip and your chip carrier.
Most of the time Sometimes, the wire comes unthreaded from the needle-tip.
It’s awkward to rethread, the hole through which you pass the thread is around the back of the needle, inclined at a 45 degree angle, and virtually invisible. It is necessary to use extremely fine tweezers to grab hold of the tiny thread. Here is a picture showing the rethreading process:
Demonstration of two-qubit algorithms with a superconducting quantum processor
Is being reported in several places at the moment, see here and here for example.
Whilst I’m confident that it works well in the few qubit limit, I’m not at all convinced that cavity QED is a scalable QC architecture. We’ve done some work on this architecture ourselves. There’s only so many qubits you can fit into one of these microwave resonators, and that’s without thinking about bias lines, error correction etc… I’d like to see a proposal for a scalable version.
So we’ve had a week of University Admissions open days and activities days here in Physics land. Here are a couple of pictures where I’m giving a talk entitled ‘Supercool Computers’ to AS level students (16-18).
We typically show some demonstrations with Liquid Nitrogen, verify that gases do not follow the ideal gas law using balloons, show electrical resistance at low temperatures, explore Lenz’s law at different temperatures using magnets and copper tubes and show some of the lovely properties of (High-Tc) superconductors such as the Meissner effect by levitating magnets.
The talk also introduced some fairly advanced concepts such as some fundamental aspects of Quantum Mechanics and practical Quantum Computing. In fact we have a HTC SQUID set up in the talk demonstrating the wave properties of electrons, analogous to the Young’s double slit experiment. The idea is to whet the appetite of potential students and make them realise that Physics isn’t all astronomy and the LHC, and that device physics/QC can also be a damn sexy thing to study 🙂
It’s all rather fun. I had this great toy to play with called a ‘visualizer’ which is basically a camera hooked up to the projector. Great for showing demos. I even zoomed in on some SQUID chips to show the audience the results of the microfabrication process.
Eventually I’m hoping to get a video of the talk up online when I’ve harrassed someone into filming the event and done it enough times to get all the demos working smoothly. We had issues with a dodgy multimeter this time. Lesson of the day: Never work with children or electronics.
If you are passionate about fine wine, cheese, and superconducting digital electronics, you may wish to attend the EUROFLUX 2009 conference taking place in beautiful Avignon in the South of France.
Although Euroflux is based on a European framework, the conference is international, and people are welcome to attend. You can read my review of EUROFLUX2008 here.
The abstract deadline is Monday!! See you there 😉
I’m really confident someone will have fusion technology working pretty soon. There are so many people now getting interested in this area. Just in the past few weeks I’ve found this article covering progress at General Fusion, and this follow up article from Next Big Future. There’s also this article about progress with Polywell-style reactors, and that’s just what’s popped up in my newsfeeds, I wasn’t even looking 🙂
Picture © General Fusion
Speaking of cleaner energy, this story was all over the main news channel yesterday, discussing climate change in the UK in 60-70 years time. What? Why do people insist on extrapolating current trends without even considering technological advancements (exponential ones at that)? I can’t believe that people have such a pessimistic doom and gloom attitude towards this issue. They should be running news stories about research and progress towards future methods of active, local and global climate control and advances in green technologies. I seriously believe that by 2080 we’ll be thinking it was rather quaint that people were so worried about the fate of the planet from the point of view of traditional, naturally occurring weather systems. And our ability to produce large amounts of CO2 detrimental to our own survival will be more likely seen as a bygone era once we are able to control the global climatic conditions.
I do still believe we should make what changes we can now to reduce further compunding of the energy crisis. Hence why I’m saving up for a hybrid car. But think about the probability that someone, somewhere will engineer and commercialize (Note: not even invent – there are plenty of good ideas already) a green energy technology which can be implemented globally within the next 50 years. I don’t think that seems unlikely. I get really irritated with the large proportion of technologically traditionalist and Luddite views in this country, especially when they dominate news headlines!
Courtesy of Quantum Bayesian Networks, an article entitled “The Quantum Leap of Quantum Computing” on Penny Sleuth. It’s great to see a wider business and market audience becoming interested in QC.
However, this is slightly irritating:
“This means computers would become exponentially more powerful because each “quantum bit” (qubit) could store a much greater range of numbers than the two that binary math restricts us to. Imagine a laptop with the computing power of the world’s 10 most powerful supercomputers. Then you begin to grasp the potential of quantum computing.”
In the spirit of a very popular television program:
Let me explain for any readers who are slightly confused at this point: Quantum computers will be very good at solving certain types of hard problems somewhat faster than classical computers. This should become some sort of mantra. (If anyone can think of a catchy version that would be cool).
They won’t be general purpose machines. The best way to think of a QC is more like a co-processor (say like a hardware graphics accelerator).
The types of problems that they will be good at solving are exciting and interesting in themselves. Quantum computers are cool enough without the overhype 🙂
On Thursday I attended a seminar by Simon Webster from the Ion trap QC group in Oxford. I didn’t realise that ion trap QC was so advanced. Having spent so much time in the happy world of LSI Josephson logic, I had a prehistoric picture of an ion trap being a large metallic cavity surrounded by huge electrodes similar to plasma confinement systems. But no, it can be done to micrometer precision on chip with ions trapped in tiny channels. You can shuttle individual ions, or little chains of them around the chip, allow them to interact and evolve to perform the computation, and then move them elsewhere, or read them out.
Photo © Oxford Ion trap group
Alas, it is still only possible to manipulate a couple of qubits at the moment, as is the case with most QC realizations. The qubits themselves are formed by manipulating transitions between energy levels of the ions. In this case, Ca+ ions. Entanglement can occur for example between the ion and the photon emitted during a relaxation from an excited state. Therefore one advantage of Ion trap QC is that it is natively good at handling static (ionic) and flying (photonic) qubits with the same technology, and quantum information can therefore be transferred over long distances and on/off chip quite easily.
Exciting stuff. I’ve still got my money on Josephson junctions, but competition in experimental QC is healthy 🙂
Overview of the Oxford Ion trap
Nature paper on Ion trap QC
Info from MIT
Here’s a measurement showing the response of a Josephson Junction as it switches from the zero voltage (superconducting) state to the voltage state, corresponding to escape of the phase from a potential well in the junction’s ‘washboard’ (energy) potential. The switching is a probabilistic process, so many measurements are compiled into a histogram at each temperature to get an average, stochastic response. The width of this histogram is then monitored as a function of temperature:
Classically, the phase gets excited out of the metastable minimum due to thermal fluctuations (the state gets a ‘kick’ out of the well from the thermal energy available in the system). As the temperature is decreased, this is less likely to happen, and so the state stays in the well for longer, and the histogram gets narrower. If the junction is small enough it may be possible to see escape due to quantum tunneling of the phase, a competing escape mechanism. This is a temperature independent process and the width should saturate at low temperatures if quantum tunneling occurs. Unfortunately, the prescence of external interference gives a very similar effect. So here we measure a large junction, which should behave classically to a very low temperature. Any saturation of the width would demonstrate noise limitation.
This is a textbook response: The straight line demonstrates that the system follows the thermal activation theory, and furthermore is not noise limited. So in future measurements of junctions, any saturation observed must be either due to quantum effects, or noise sources intrinsic to the junction itself.
So I can now believe that I’m seeing real quantum processes in the junctions.
This makes me happy.
The CF cryostat is a lovely piece of kit for quickly measuring samples down to 4.2K. Here is a picture:
The dewar containing the liquid Helium can be seen on the left. The setup is rather simple, you have a gas pump, a thin transfer tube (the silver tube joining the dewar to the cryostat) and the sample space (inside the brass-coloured cryostat body). When the pump is turned on, it draws helium from the dewar through the thin tube and the sample space and returns it to a pipe on the wall (part of the overall in-house helium gas system installed in all the labs). Because the transfer tube is thin, the liquid Helium comes through quite slowly (you can adjust the flow rate) and a single dewar of liquid Helium (which holds ~45L) can therefore go a long way.
The electronics to control the experiment can be seen on the rack of equipment to the right in the photograph. This particular setup allows current-voltage (IV) characteristics and differential conductance (dI/dV) measurements to be performed on Josephson Junctions, with a PC controlled DAQ (data acquisition) system. A patch panel on top of the experimental insert allows up to 20 DC lines to be chosen for the measurement, so that many junctions to be tested in the same run.
Controlling the temperature on this type of cryostat is notoriously tricky. The generally accepted method is to run the pump to cool the system, whilst simultaneously heating the sample until the temperature stabilises at the required value. Which is a bit wasteful, it’s like running the heater and the air-con in your car at the same time. It can be done automatically with a PID temperature controller. Another way to obtain a measurement as a function of temperature is just to record data whilst the system cools down to 4.2K whilst the pump is running, or warms back up to 300K after the pump is switched off. Doing this on the warming cycle is slightly more stable. Additionally you don’t have the additional electrical noise of the pump if you are trying to conduct a low-noise experiment. This is a good way of obtaining Resistance-versus-Temperature (RT) measurements, which allow you to see the point at which your sample/junction goes superconducting. From this data you can also obtain the Residual Resistivity Ratio (RRR) which gives a measure of the quality/purity of the material being used.
A couple of thermometers are usually placed near the sample to give a good idea of the temperature gradient inside the cryostat. If there is a high temperature gradient near the sample, the temperature you read on your thermometer (which is generally a few cm away) might not be the exact sample temperature.
One of the best features of the CF is that if you run the pump for long enough, once the temperature inside the cryostat gets down to 4.2K, liquid Helium starts to collect at the bottom. Once you have collected some, you can turn off the pump and the system will stay at 4.2K until the collected liquid has all boiled off. This is a nice way to ensure that your measurement stays at a definite temperature for about an hour. Of course we don’t turn the heater on when we have liquid Helium in the bottom of the cryostat, or else the top of your experiment blows out and hits the ceiling as the Helium gas tries to occupy 700 times its liquified volume 🙂
With Niobium based Josephson junctions (Tc~9K), 4.2K is a good temperature to take measurements of their superconducting IV curves.
Why would you want to measure IV curves? Well, that’s another post 🙂
I’m currently reading Distress by Greg Egan. The story is set at a Physics conference, so I’m actually really enjoying it (although I think overall I preferred Permutation City). It is however most unlike any Physics conference I’ve ever attended. With murder, mystery, intrigue, fanatical religious cults, shadowy biotech corporations, kidnapping, deadly bioweapons and potentially the end of the multiverse as we know it, maybe I should be frequenting TOE conferences instead of LT ones 🙂
However there are some descriptions which aren’t so far fetched, such as the conference venue being a picturesque tropical coral-reef island. That one does happen occasionally.