from a paper I found when I was creating some order in my bookshelves:

Peter Dittrich, Jens Ziegler, and Wolfgang Banzhaf:

Artificial chemistries – a review. Artificial Life, 7(3):225–275, 2001

the quote runs:

”

Mechanical Artificial Chemistry

Hosokawa et al. [64] show that a mechanical self-assembly system can be regarded as a
chemical system. This allows to build a model for the yield of the desired target structure
by means of chemical kinetics theory. In [64] this approach is demonstrated by deriving
an analytical model for a simple mechanical self-assembling system shown in Fig. 9. The
basic units of the mechanical systems are regular triangular units which may form bonds
by permanent magnets. This work demonstrates that artificial chemistry research is not
limited to formal or algorithmic systems.

Another physical model of a chemical system has been developed by Grzybowski,
Stone, and Whitesides [55]. It consists of millimeter-size discs rotating at a liquid-air
interface. An important difference to the work of Hosokawa et al. [64] is that the emergent
structures are dissipative. This means, that the structures are only present as long as
they are supplied with energy. The energy is fed into the system by a permanent magnet
rotating underneath the liquid.

”

the images above are two pictures from the paper by Hosokawa referred to:

Kazuo Hosokawa, Isao Shimoyama, and Hirofumi Miura. Dynamics of self-assembling systems –
analogy with chemical kinetics. In R.A. Brooks and P. Maes, editors, Artificial Life IV, pages
172–180, Cambridge, MA, 1994. MIT Press. (not online for non-subscribers unfortunately, I should get a subscription..)

looks very intriguing…