For a closed system at constant temperature and volume, the criterion for spontaneity is $dA < 0$. However, for a system at constant composition, the total differential of $A$ is given by $$dA = -PdV -SdT.$$ This seems to imply that $dA = 0$ for a system at constant volume, temperature, and composition. The Wikipedia page for Helmholtz Free Energy claims that this relationship holds even for non-spontaneous processes. The derivation of $A$ as criterion for spontaneity at constant volume and temperature is $$dA = dU - TdS - SdT = \delta q - TdS + \delta w - SdT.$$ If we consider only $PdV$ work, we obtain $$dA = \delta q - TdS - PdV - SdT$$ and at constant $T$ and $V$, we get $$dA = \delta q -TdS.$$ The criterion $dA <0$ for spontaneous processes then follows from the Clausius inequality. These two equations seem to be contradictory, so I was hoping someone could explain how this can be reconciled. Thank you so much!

Clarification: The way I derived the $dA = -PdV - SdT$ for an arbitrary process is as follows:

For a function $f(x_1,\dots,x_n)$, the complete differential of $f$ is $$df = f_1 dx_1 + f_2 dx_2 + \dots + f_n dx_n$$ where $$f_i = \left(\frac{\partial f}{\partial x_i}\right)_{x_j \: j\neq i}.$$ Consider the function $$g = f - f_1 x_1.$$ The differential of $g$ is $$dg = df - f_1 dx_1 -x_1df_1$$ so that $$dg = -x_1 df_1 + f_2 dx_2 + \dots + f_n dx_n.$$ Now, applying this to the special case of internal energy $U(S,V)$, if we define $$A = U - \left(\frac{\partial U}{\partial S}\right)_{V} S = U - TS$$ then the above implies that $$dA = -SdT - PdV$$ since $$dU = TdS - PdV.$$

  • $\begingroup$ Consider review of search results for site:stackexchange.com OR site:libretexts.org thermodynamics closed system spontaneity criteria $\endgroup$
    – Poutnik
    Oct 10 at 20:57
  • $\begingroup$ Provide you reasoning what you think they are contradictory. as dS >= dq/T, dq - TdS <=0. 0 for reversible processes, <0 for spontaneous ones. $\endgroup$
    – Poutnik
    Oct 10 at 21:03
  • $\begingroup$ For a spontaneous process at constant temperature and volume, the fact that $dQ < TdS$ implies that $dA < 0$ with a strict inequality. However, the fact that $dA = - PdV -SdT$ seems to imply that for any constant volume and temperature (and constant composition), we have $dA = 0$ since $dV = 0$ and $dT = 0$. This is the main source of my confusion. $\endgroup$ Oct 10 at 22:39
  • $\begingroup$ dA=-p.dV -SdT is derived with assumption dq=T.dS, i.e. for reversible, non spontaneous changes, so TdS and -TdS mutually cancel. $\endgroup$
    – Poutnik
    Oct 11 at 6:11
  • $\begingroup$ The link I provided to the Wikipedia article states that this relationship holds for non-reversible processes as well. That is where I am confused $\endgroup$ Oct 11 at 10:37

1 Answer 1


$$A = U -TS$$ $$\mathrm{d}A = \delta Q - p\mathrm{d}V + \delta W_\text{nonexp} - T\mathrm{d}S - S\mathrm{d}T$$ $$dS \ge \delta Q/T \implies T\mathrm{d}S \ge \delta Q $$ Assuming $\delta W_\text{nonexp}=0$: $$\mathrm{d}A = \delta Q - p\mathrm{d}V - T\mathrm{d}S - S\mathrm{d}T$$ As $T\mathrm{d}S \ge \delta Q$: $$\mathrm{d}A \le T\mathrm{d}S - p\mathrm{d}V - T\mathrm{d}S - S\mathrm{d}T$$ $$\mathrm{d}A \le - p\mathrm{d}V - S\mathrm{d}T$$

  • $\begingroup$ This makes sense. Does the derivation of $dA = -PdV - SdT$ using Legendre transformations assume reversibility? I thought it holds in general $\endgroup$ Oct 11 at 11:46
  • $\begingroup$ Because the statement that "the natural variables of A are V and T" seems to imply that A(V,T) is constant for any process at constant V and T so that dA = 0 $\endgroup$ Oct 11 at 11:49
  • $\begingroup$ It uses the A definition, the Clausius inequality TdS>=dQ and ordinary mathematical operations. // AFAIK, the Leg. tr. was used and the source for decision of the way to define A and G as state functions. $\endgroup$
    – Poutnik
    Oct 11 at 11:54
  • $\begingroup$ Aren't all of those true in general? I don't see why that derivation is only valid for reversible processes $\endgroup$ Oct 11 at 11:58
  • $\begingroup$ But I as a chemist am not familiar with Legendre transformations by the name, rather just Legendre orthogonal polynoms. $\endgroup$
    – Poutnik
    Oct 11 at 12:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.