Thursday, August 21, 2014

Should I join this professional scientific society?

Why are they important?
Why should you join? not join?
Why are the membership numbers of some societies declining (some dramatically)?

It seems every month the American Chemical Society (ACS) sends me a letter asking me to join. I am not sure who recommended me for membership. I find it ironic because I once tried to join the Royal Australian Chemical Institute but was rejected because they did not seem to think I was a real chemist. [ouch!] Over the years I have belonged to several societies. But, some of these memberships have lapsed. Recently, I was personally asked by one, "What do we have to do to get you to rejoin?"
I did not have an answer, stimulating this post.

First, let me say why these societies can be incredibly important. They can
  • Publish good journals that are owned and run by scientists. These can avoid the problems of commercial outfits such as Nature [sensationalism over substance] and Elsevier [quantity over quality, dubious business practises].
  • Organise useful conferences.
  • Give prizes and awards to recognise excellence.
  • Provide career services, particularly for younger members.
  • Represent science and scientists to government, industry, and the community. This is not just lobbying for more funding but making important public statements on issues such as climate change.
If we don't join, we end up with the Tragedy of the Commons, whereby our long-term collective interests suffer because we prioritise our individual self-interest.

So, why not join?
  • Membership is expensive, particularly if you belong to several.
  • Your mail box (both hard and soft) will be clogged with magazines, newsletters, fund-raising appeals, announcements, elections, ...
  • You may be asked to serve on committees.
  • There are many societies to choose from, particularly if you live outside the USA and you work  at the interface of two or more disciplines [physics, chemistry, biophysics, materials science, ...]. APS, ACS, RACI, AIP, IoP, MRS, ...
  • Smaller national societies are struggling for viability in an era of internationalisation. It is not clear why some still publish journals.
  • Society conferences compete with a multitude of other conferences. Some national society conferences may not have a critical mass of people or seem a magnet for mediocrity.
  • If you don't go to the society conferences and can read their magazine online via a library subscription there is less personal incentive to join.
So, how do you decide who to join? or not join? or let your membership lapse?
What would a society have to do to convince you to join?

Wednesday, August 20, 2014

Belated recognition for early work on superconducting organic charge transfer salts

In the mid-1990s, through the influence of Jim Brooks, I became interested in organic charge transfer salts. I read a very helpful paper by Kino and Fukuyama that considered a Hubbard model for the family kappa-(BEDT-TTF)2X. This led to me writing a review article and a short piece in Science, comparing the organics to the cuprates.

Aside: Being young and naive, and living before impact factor obsessions, I made the mistake of publishing the review in Comments on Condensed Matter,  which is not even listed on ISI Web of Science. I chose that journal because I knew it had published an influential review on heavy fermions by the stellar cast of Lee, Rice, Serene, Sham, and Wilkins.
Fortunately, I put the paper on the arXiv and a lot of people read it, and the Science paper often gets cited, by association.

The review stimulated a lot of work, particularly on the Hubbard model on the anisotropic triangular lattice at half filling, and the associated Heisenberg model for the Mott insulating phase. A recent review is here, with Ben Powell.

Only recently I became aware of some related work from around the same time that is never cited, literally.

κ-(BEDT-TTF)2X organics, as seen for Hubbardists 
V.A. Ivanov

Electronic structure and superconductivity of κ-(BEDT-TTF)2X salts
V. A. Ivanov, E. A. Ugolkova, M. E. Zhuravlev

The relevant Hubbard model, the importance of correlations, and the possibility of d-wave superconductivity are all discussed. These papers may not be as clear, or crisp, and comprehensive as my review but some of the key physics is there.
Ivanov should get credit for that.

What is the sociology here? Why was I influential but Ivanov was not?
Here may be some contributing factors.
I published the Perspective in Science, but back then not many physicists actually read it.
More importantly, I put my papers on the arXiv. Back then [pre-web] most people subscribed to and read the daily email listing recently posted papers. I timed my posting so my review would be the first of the daily list.
I followed up with more papers. I gave a lot of talks, both at conferences and universities around the world. I talked to experimentalists. I encouraged theorists to use their favourite technique to study the relevant Heisenberg and Hubbard models. I had postdocs work on the subject. They then went around talking about it.

Tuesday, August 19, 2014

Future directions for physical chemistry

At the American Chemical Society meeting last week J.T. Hynes gave a talk
Some modest proposals for 21st century physical chemists 
Here are his three main points.

(1) The most familiar problems/phenomena may in fact not be at all already understood, and can provide fertile areas for discovery;

(2) Just an experiment or a theory because it is 'old' (e.g. of a certain vintage) does not mean it is inferior/wrong despite the lack of novelty and modernity;

(3) Simple, well-constructed analytic models have a significant role to play in comprehending and advancing both theory and experiment.

Unfortunately, I was not at the meeting, but my colleague Seth Olsen was and told me I would have enjoyed the talk. These points certainly resonate with my own views.

Monday, August 18, 2014

Gratuitous graphs

Movies sometimes feature gratuitous graphic violence and sex.
Scientific papers sometimes feature gratuitous graphs.

Before desktop computers it was a lot of work to produce a single graph.
Now one can produce ten graphs in an afternoon!
Why not put them all in the paper?
The paper will be longer and give the impression of being more comprehensive.
I feel that some students want to show all their graphs to show just how productive they have been.
No! Not every graph you make should be in the paper.
Analysis, synthesis, and discipline is needed.

Figures should carry the weight of the paper.
Many people, myself included, will scan the figures looking for something potentially interesting and
important, in order to make a decision as to whether or not to actually read the paper.
Figures and their captions need to be clear, comprehensible, and have significant content.

Graphs should be like text: polished and repolished. Good authors work hard at writing, editing, polishing, re-writing, and re-writing text. Some graphs should be deleted or relegated to supplementary material. Some should be combined. Some should be split in two.

Some chemistry journals now put an author produced "summary graphic" or "graphical abstract" on the journal website for each paper. I find some of these very helpful. I also feel it is unfortunate that these are usually not in the paper. I think many papers would benefit from at least one simple schematic figure that illustrates the main point, key definitions, or approaches of the paper. I reproduce an example below from a recent paper by Seth Olsen and I.

Saturday, August 16, 2014

Facing the black dog of depression

Unfortunately, this past two months has seen the tragic death by suicide of Robin Williams,  Seth Teller [an MIT Computer Science Professor], and Yoshiki Sasai [co-author of two retracted Nature papers].

I thought the following video on depression was helpful.

Thursday, August 14, 2014

Scale of the Nernst effect in a bad metal

A science fiction fantasy is that we should be able to make "materials by design" that have any physical property (density, thermal conductivity, hardness, thermoelectric figure of merit, heat capacity...)  that we desire. However, it seems that there are certain physical constraints that determine the overall scale of many physical properties.

I find it helpful to have a feel for typical orders of magnitude. What is particularly interesting is that sometimes these magnitudes are related to fundamental constants [electronic charge (e), Boltzmann's constant (k_B), Planck's constant (hbar)] and basic length scales such as the lattice constant a of a crystal.

Here are three scales I have emphasised before

Resistivity ~ hbar a / e^2 ~ 100 microohm-cm  which is associated with the Mott-Ioffe-Regel limit.

Thermoelectric power,  S ~ k_B/e ~ 86 microvolt/K

Mobility, mu ~ e a^2/ hbar ~ 1 cm^2 V/sec

One can find these scales by dimensional analysis or by doing things like looking a formulas from transport theory and (assuming a bad metal) setting the mean-free path comparable to the lattice constant. One can debate whether one uses hbar or h, but for little purpose.

How about the Nernst signal, nu?

nu ~ k_B a^2 / hbar ~ 0.01 microV/KT

A few minor notes.

1. One can get this scale from the above expressions for S and mu if one uses the observation that in some strongly correlated materials
nu ~ S * Hall mobility.

2. One Volt/Tesla = m^2/sec  [One can see this easily from F = q(E + vxB)].

3. Given that the Nernst effect involves charge transport I find it surprising that the electronic charge does not appear.

The figure below, taken from a nice review by Behnia, shows that this is the right scale for bad metals such as cuprates, and heavy fermions above the coherence temperature.

One also sees this scale in recent DMFT calculations for a doped Hubbard model (see Figure 2d in this PRL ) and recent measurements (see Figure 4) on organic charge transfer salts.

Wednesday, August 13, 2014

Documenting harassment of women in science

There is a very disturbing article in the New York Times about sexual harassment and discrimination of female scientists and science writers.