Thursday, February 2, 2012

Teaching thermo without stat mech

For the past ten years I have been involved in teaching a course "Thermodynamics and Condensed Matter Physics" to second year physics majors. In the past I have posted some of lectures which cover the second half of the course, applying thermo to condensed matter. I did not design the course but inherited it. [For department historical/political/personal reasons the condensed matter part was actually originally crystallography!]

I quite like the course because I believe (unlike many people) that students should learn and master macroscopic thermodynamics before they learn statistical mechanics. Entropy should be introduced as a measure of the relative accessibility of the equilibrium (macroscopic) states in an adiabatically isolated system (i.e. irreversibility) rather than as related to the possible number of microstates in an system.

However, finding a textbook is not easy. Students find Callen, the one originally used for the course, too difficult, and it is very expensive. We have tried using the first few chapters of Atkin's Physical Chemistry. The publisher provides .ppt slides of all the figures which is useful. However, it is a chemistry book, is expensive, very heavy, and the Solutions Manual is full of errors [in spite of being in the Ninth Edition].

So here is our plan (due to my co-lecturer Joel Corney) for this year: to follow an Introduction to Thermal Physics by Schroeder, chapters 1-5, but skip the sections which give a microscopic treatment of entropy. Some of that material will be substituted with some readings from Atkins.

The best part of Schroeder's book are the extensive problems. They are very interesting, covering a diverse range of topics from black holes to meteorology, and there is great Solutions Manual! I have used the book before to teach Stat. Mech.

I welcome suggestions about other texts.


  1. The problem that I've had with the idea of entropy as accessibility of equilibrium states is that sort-of invokes dynamics in a field that explicitly does not deal with dynamics. This never made sense to me as a student. Strangely I think that I became more comfortable with how to think about that definition through learning about the information-theoretic concept of entropy, because relative entropy is a useful quantitative model of "accessibility". But even now, I usually need a lot of sleep before I can think clearly about it (which I am not getting now.. It's 2 am and I'm brain dead and exhausted so maybe I shouldn't take up this thread now...).

  2. I was taught undergrad. thermo with Kittel & Kroemer -- it is not so bad. My favorite condensed matter book now is "Introduction to the Physics of Electrons in Solids" by H Alloul, Springer 2011. First year graduate level but good refresher upper level, fascinating problem sets.

  3. I should have mentioned that when Ben Powell taught the first half of the course he used C.J. Adkins, Equilibrium Thermodynamics, 3rd edition (1983). It does present thermo from a macroscopic point of view. However, it essentially has the look of a 1950s book which is not too appealing to the latest generation of students.

    Kittel and Kroemer is rather expensive and takes a microscopic approach. It used to be used for our stat. mech. course but I replaced it with Schroeder.

    Thanks for the tip on Alloul's book.

  4. This year I've started looking at Gould and Tobochnik's "Thermal and Statistical Physics". Their presentation of the fundamentals of classical thermodynamics is quite self-contained, which would suit our purposes.

    In later chapters, they incorporate a lot computer simulation, which might make for an interesting approach to the follow-on stat mech course.