Recent Changes · Search:

Dispense


Condensed Matter

Chapters:

Appendices


PmWiki

pmwiki.org

edit SideBar

PhaseTransitions

< Thermodynamics recap | Index | Crystals as Many Body? >


Solid, liquid, gas

Solid, liquid, gas (or vapour) phases can be qualitatively explained by Van der Waals equation of state.

Figure 1 shows its 3D phase diagram vs. {$V,T,P$}. Besides the three quoted phases certain loci correspond to coexistence of two of these phases. A triple point corresponds to the condition where the three phases coexist. A further special point, the critical point, marks the condition where liquid and gas cease to exist as distinct phases, leaving way to a supercritical fluid, that may display different physical and chemical properties.

Figure 2 shows the projection of Fig. 1 onto the {$P,T$} plane. The triple and critical points are evident, as well as lines of coexistence of two phases. Crossing these linesnormally involves a latent heat, i.e. absorption or release of heat at constant pressure and temperature.

Fig. 1: standard phase diagram of element, from Van der Waals

Top


Gibbs phase rule and classification of phase transitions

The Gibbs phase rule refers e.g. to a fixed volume (a flask where a reaction takes place, for example). It dictates the number of degrees of freedom {$D$}, or the locus in the {$P,T$} phase diagram of the system, characterized by {$C$$} components (e.g. {$C=1$} for a single element, {$C=2$} for a binary mixture) where {$P$} phases coexist. In the 3D phase diagram planes and curves are not flat.

{$$D=C-P+2$$}

It tells us that for the Van de Waals fluid {$C=1$}, each separate phase {$P=1$} is described by a plane with {$D=2$} degrees of freedom ({$P$} and {$T$}), the coexistence of {$P=2$} phases (liquid-valour, liquid-solid, solid-vapour) is a line, {$D=1$} and the coexistence of {$P=3$} phases is a point (the triple point).

The order of the phase transitions was first classified by Ehrenfest from the appearance of a discontinuity in the successive derivatives of the Gibbs free energy with respect to a given thermodynamic variable{$X$}. The first derivative is usually associated with an order parameter {$\eta$}, that is zero above the transition and non zero below. The Ehrenfest scheme runs according to the following table

Fig. 2: projection of the Van der Waals phase diagram onto the {$P,T$} plane

Order

{$$ G$$}

{$$\eta=\frac {\partial G}{\partial X}$$}

{$$\frac {\partial^2 G}{\partial X^2}$$}

I

continuous

discontinuous

II

continuous

continuous

discontinuous

This picture however is not entirely consistent, since derivatives of {$G$} may display cusps, instead of discontinuities, and even divergences. Not like that expected in the first line for the second derivative (a delta function for a single point, unmeasurable, and continuous elsewhere), but divergences developing slowly as the transition is approached. These divergence reveal the existence and relevance of fluctuations, neglected in the thermodynamic mean field approach.

Fluctuations characterize particularly second order phase transitions. A clear example is the critical point of water, {$T_c,P_c$}, where phases cease to be stable. At higher pressure or temparature water is a supercritical fluid with properties different from any of the other phases. The density differences among liquid and vapour (proportional to their order parameter) vanishes. At the critical point fluctuations diverge (i.e. dynamic clusters of the two phases appear on any length- and time-scales). This gives rise to strong light scattering, called critical opalescence (the fluid scatters a lot at optical wavelengths and becomes whitish).

Note that critical fluctuations, hence the corresponding derivative of the Gibbs free energy, diverge only in the thermodynamic limit. Any finite sze sample has an intrinsic limit in its length scale, hence in its time scale.

Lev Landau proposed a simple mean field theory that accounts better for many properties of phase transitions, although still neglecting fluctuations. However, thei can be reconciled with the theory by small modifications. We shall review some of its features. The order of the phase transition in this scheme is simply dictated by the appearance of a discontinuity in the order parameter, in I order phase transitions, or by its absence, in II order, or continuous phase transition.

Top


Examples

Beside solid-liquid-gas phases of elements and mixtures, phase transitions are important in many contexts. The solid is generally more complex than a single phase because new degrees of freedom appear, as internal fields develop.

Ice, seen before, is a remarkable case of many structural phase transitions, where symmetry reduction plays a role. The leading electrostatic dipolar interaction, with its long range nature, is compatible with very reduced symmetry at zero temperature. As thermal energy increases, approximate symmetries appear. This is a process worth coming back to and it gives rise to a very rich variety of phases in ice.

First order transitions are accompanied by hysteresis. Very often structural transitions are first order, but examples of second order ones exist (e.g. in Sr2CuWO6 at 870 K).

The ferromagnetic transition is second order in zero applied field and the order parameter, magnetization {$M$} goes continuously to zero at the Curie temperature {$T_C$}. Static uniform susceptibility {$\partial M /\partial T$} diverges for {$T\rightarrow T_C$}.

Also the Néel transition between the paramagnetic and the antiferromagnetic state is a second order transition. The order parameter is the staggered, or sublattice magnetization, {$M_s$} with diverging susceptibility at the wavevector characteristic of the order, typically at the Brilllouin zone boundary: if the primitive cell is doubled {$a^\prime = 2a$} by the alternating antiferromagnetic moments, the propagation vector is {$q_{AF}=2\pi/a^\prime = \pi/a$} and {$\chi(q_{AF})=\chi(\pi/a)$} diverges.

In both cases fluctuations become critical, i.e. the fluctuations corresponding to the order, {$q_c=0$} for the ferromagnet and {$q_c=q_{AF}$} for the Néel antiferromagnet, slow down and grow in correlation length in the paramagnet, as {$T\rightarrow T_c^+$}. This is analogous to critical opalescence. It corresponds the appearance of clusters of the new order on any length- and time scale at {$T_c$}, i.e. to the divergence of the response function {$\chi(q=q_c,\omega=0)$}. A similar thing happens for {$T\rightarrow T_c^-$} to the spin wave modes, with the appearance of dynamical disordered clusters in the ordered phase.

Liquid crystals made of lower symmetry elongated molecules (compared to globular molecules) undergo transitions from liquid to nematic phases, where translations are still disordered, like in a liquid, but orientations are nearly static, like in a solid: molecules align within a domain along a given direction (the nematic director). Depending on the details of the molecular symmetry ,ore complex phases appear. The nematic-liquid transition is first order.

Biomolecules undergo phase transitions of similar nature, since they are often long polymer chains. Lipids can thus form membranes and mycelles, very important for their functionality. Also folding of proteins and DNA, essential for their biological functions, is governed by phase transition: thus life cannot work outside pretty narrow thermodynamic conditions.

Top

Edit - History - Print - PDF - Recent Changes - Search
Page last modified on March 10, 2019, at 02:33 PM