Tuesday, August 07, 2007

This week in cond-mat

Another couple of papers that caught my eye recently....

arxiv:0707.2946 - Reilly et al., Fast single-charge sensing with an rf quantum point contact
arxiv:0708.0861 - Thalakulam et al., Shot-noise-limited operation of a fast quantum-point-contact charge sensor
It has become possible relatively recently to use the exquisit charge sensitivity of single-electron transistors (SETs) to detect motion of single electrons at MHz rates. The tricky bit is that a SET usually has a characteristic impedance on the order of tens of kOhms, much higher than either free space (377 Ohms) or typical radio-frequency hardware (50 Ohms). The standard approach that has developed is to terminate a coax line with an rf-SET; as the charge environment of the rf-SET changes, so does its impedance, and therefore so does the rf power reflected back up the coax. One can improve signal to noise by making an LC resonant circuit down at the rf-SET that has a resonance tuned to the carrier frequency used in the measurement. With some work, one can use a 1 GHz carrier wave and detect single charge motion near the rf-SET with MHz bandwidths. Well, these two papers use a gate-defined quantum point contact in a 2d electron gas instead of an rf-SET. See, rf-SETs are tough to make, are fragile, and have stability problems, all because they rely on ultrathin (2-3 nm) aluminum oxide tunnel barriers for their properties. In contrast, quantum point contacts (formed when a 2d electron gas is laterally constricted down to a size scale comparable to the Fermi wavelength of the electrons) are tunable, and like rf-SETs can be configured to have an impedance (typically 13 kOhms) that can be strongly dependent on the local charge configuration. Both the Harvard and Dartmouth groups have implemented these rf-QPCs, and the Dartmouth folks have demonstrated very nicely that theirs is as optimized as possible - its performance is limited by the fact that the current flowing through the QPC is composed of discrete electrons.

arxiv:0708.0646 - Hirsch, Does the h-index have predictive power?
*sigh*. The h-index is, like all attempts to quantify something inherently complex and multidimensional (in this case, scientific productivity and impact) in a single number, of limited utility. Here, Hirsch argues that the h-index is a good predictor of future scientific performance, and takes the opportunity to rebut criticisms that other metrics (e.g. average citations per paper) are better. This paper is a bit depressing to me. First, I think things like the citation index, etc. are a blessing and a curse. It's great to be able to follow reference trails around and learn new things. It's sociologically and psychologically of questionable good to be able to check on the impact of your own work and any competitor whose name you can spell. Second, Hirsch actually cites wikipedia as an authoritative source on how great the h-index is in academic fields beyond physics. I love wikipedia and use it all the time, but citing it in a serious context is silly. Ahh well. Back to trying to boost my own h-index by submitting papers.

No comments: