Universality classes for unstable crystal growth

Universality has been a key concept for the classification of equilibrium critical phenomena, allowing associations among different physical processes and models. When dealing with nonequilibrium problems, however, the distinction in universality classes is not as clear and few are the examples, such as phase separation and kinetic roughening, for which universality has allowed to classify results in a general spirit. Here we focus on an out-of-equilibrium case, unstable crystal growth, lying in between phase ordering and pattern formation. We consider a well-established 2 + 1-dimensional family of continuum nonlinear equations for the local height h(x, t) of a crystal surface having the general form partial derivative(t)h(x, t) = -del . [j(del h) + del(del(2)h)]: j(del h) is an arbitrary function, which is linear for small del h, and whose structure expresses instabilities which lead to the formation of pyramidlike structures of planar size L and height H. Our task is the choice and calculation of the quantities that can operate as critical exponents, together with the discussion of what is relevant or not to the definition of our universality class. These aims are achieved by means of a perturbative, multiscale analysis of our model, leading to phase diffusion equations whose diffusion coefficients encapsulate all relevant information on dynamics. We identify two critical exponents: (i) the coarsening exponent, n, controlling the increase in time of the typical size of the pattern, L similar to t(n); (ii) the exponent beta, controlling the increase in time of the typical slope of the pattern, M similar to t(beta), where M approximate to H/L. Our study reveals that there are only two different universality classes, according to the presence (n = 1/3, beta = 0) or the absence (n = 1/4, beta > 0) of faceting. The symmetry of the pattern, as well as the symmetry of the surface mass current j(del h) and its precise functional form, is irrelevant. Our analysis seems to support the idea that also space dimensionality is irrelevant.