DU = TdS - p dV is valid for irreversible process -- Proof

24 replies · 10K views · last reply
Pushoam
Messages
961
Reaction score
53
Printed textbook page with equation 14.17, dU = T dS - p dV, boxed in magenta, and equation 14.18, dU = T dS - p dV, boxed in black, with explanatory text between them about reversible and irreversible changes.


The book says that all quantities in eq. 14.17 are state- functions. How do we know that?
Earlier I read that both W and Q are not state functions.
Now, since S is state-function, it means that tem. is not a state- function, isn't it?
Does it mean that for a reversible process, both W and Q are state-functions? How to prove it?
How can we have eq. 14.18 always as entropy is not defined for irreversible process?
 
A state function is a function of the state the system is in. Work and heat are not state functions because they depend on how the system is changed (the path the system takes in the state space).
 
Orodruin said:
A state function is a function of the state the system is in. Work and heat are not state functions because they depend on how the system is changed (the path the system takes in the state space).
dS is a state function for reversible process. It is not defined for irreversible process. So, how can the eqn 14.18 hold true for irreversible process, too?
This is what I want to know.
 
Pushoam said:
dS is a state function for reversible process.
dS is not a state function. It is the differential of a state function.
 
Orodruin said:
dS is not a state function. It is the differential of a state function.
Yes, sorry for that. What I want to say was:
S is a state function for reversible process. It is not defined for irreversible process. So, how can the eqn 14.18 hold true for irreversible process, too?
This is what I want to know.
 
Pushoam said:
S is a state function for reversible process.
No, a state function does not depend on any process. It is just a function of the state of the system, not of how the system develops.
 
Orodruin said:
No, a state function does not depend on any process. It is just a function of the state of the system, not of how the system develops.
Yes, but S is defined only for a reversible process. As only for a reversible process, ##\oint\frac{dQ_{rev}}T = 0##. So, the concept of S is defined only for reversible process.
 
Pushoam said:
Yes, but S is defined only for a reversible process.
No, S is defined only for a state. It has nothing to do with any process of any sort.
 
Orodruin said:
No, S is defined only for a state. It has nothing to do with any process of any sort.
But for an irreversible process, ##\oint \frac{dQ} T <0 ## . So, S is not a state- function for an irreversible process.
Textbook page titled 14.1 Definition of entropy, defining dS = dQ_rev/T (eq. 14.1), with the integral form S(B) - S(A) = ∫ dQ_rev/T (eq. 14.2) and dQ_rev = 0 for an adiabatic process (eq. 14.3).

Textbook page section 14.2 titled Irreversible change, stating entropy S is a state function so the closed-loop integral of dQ_rev/T is zero, shown as equation (14.4).
Textbook page with the Clausius inequality derivation (equations 14.5 to 14.8) and a p-V diagram of an irreversible and a reversible change between points A and B.


What I have understood is:

The concept of entropy is defined :
dS = ##\frac {dQ_{rev}} T## such that ##\oint dS = 0## i.e. S is a state-function.

Even though the system is going through an irreversible process we can talk of entropy of the system as ## Q_rev## could be defined assumng that the system is in some reversible process. Of course, for this purpose, all reversible processes become equivalent as S depends only on the state of the system .

In the following eqn.
Inequality dS = dQ_rev/T ≥ dQ/T, with a red arrow pointing to the inequality sign.

if the system is in the reversible process, then ##dQ = dQ_{rev}##.
For example, if the system has dQ =0, i.e. thermally isolated and
a) it is in the reversible process, then dS = 0 ,
otherwise i.e.
B) the system is in the irreversible process, then ##dQ_{rev} > [dQ =0]~,dQ_{rev} >0##.

Now, my question is this ,
If the system is thermally isolated, then dQ =0 (from the definition of thermal isolation).
So, whether the system goes through reversible or irreversible process, dQ remains 0
i.e. dQ remains 0 independent of the process the system undergoes through. So, in this case, ##dQ_{rev} = dQ_{irrev}##. Right?
Is this correct?
 
Attachments
  • Excerpt from a thermodynamics textbook showing equation (14.9), dS ≥ 0, with an arrow to the note 'For a thermally isolated system', followed by explanatory text and footnote 1 on reversible processes.
    upload_2017-10-7_15-35-42.png
    28.9 KB · Views: 785
Textbook page excerpt, section 14.4 The Joule expansion, with the title boxed and the phrases irreversible process and Joule expansion highlighted, describing one mole of ideal gas expanding into an evacuated half of a thermally isolated container.
Two-panel schematic of a Joule expansion: panel (a) shows a gas in a left chamber at pressure p_i and volume V_0 separated by a closed valve from an empty right chamber; panel (b) shows both chambers at volume V_0 with pressures p_i and p_f.
Textbook page on Joule expansion showing equations for entropy change ΔS = R ln 2, with a p-V diagram below showing the Joule expansion and a reversible isotherm between volumes V0 and 2V0.


Since, Joule expansion itself is an irreversible, the path followed by the system is undefined.
So, for calculating entropy change, we simply find out the previous and the final state i.e.(##p_i, V_i, T_i) ~ and ~ (p_f,V_f,T_f).##
We define a reversible process which leads the system from the initial state to the final state. Any reversible process would work as ##\Delta##S depends only on the initial and final state.
Here, isothermal expansion acts as the suitable reversible process as the tem. remains the same (in both final and initial state).
Then, using the reversible process and 1st Law of thermodynamics with entropy, we calculate ##\Delta S##.
Is my understanding correct?
 
Textbook page, Example 14.3: entropy changes for a Joule expansion. Gas: ΔS_gas = R ln 2; surroundings: ΔS_surroundings = -R ln 2; Universe: ΔS_Universe = ΔS_gas + ΔS_surroundings = 0 (14.30).


Textbook page excerpt on entropy: highlighted text says the entropy of surroundings goes down and the Joule expansion system is thermally isolated; below, equations DS_gas = R ln 2, DS_surroundings = 0, and DS_Universe = DS_gas + DS_surroundings = R ln 2, labelled (14.31).


After calculating the ## \Delta S## , we have to come back to our system, which is going through Joule Expansion; not the isothermal expansion.
Isothermal process is just a tool calculate ##\Delta S##.
Is this correct?
 
Attachments
  • Textbook page text on entropy with highlighted sentences, followed by equations: ΔS_gas = R ln 2, ΔS_surroundings = 0, ΔS_Universe = ΔS_gas + ΔS_surroundings = R ln 2 (14.31).
    upload_2017-10-7_17-45-15.png
    28.5 KB · Views: 855
Pushoam said:
So, S is not a state- function
Look at the text after eq 14.2. S is a function of state.

Figure 14.1 explains the situation. Between any two states there are an infinite number of paths. For some paths you can calculate the change in entropy using the formula. For other paths you cannot calculate the change in entropy that way, but it is the same since it is an equation of state.
 
I'm going to respond only to the last question of your original post, regarding the applicability of dU=TdS-PdV to an irreversible process. This equation describes the relationship between the changes in U, S, and V only between two closely neighboring thermodynamic equilibrium states of a system. So it can't be integrated along an arbitrary irreversible path and be expected to give the correct answer, as you have correctly noted. However, if a system starts out at a given thermodynamic equilibrium state and then, after suffering some wildly tortuous irreversible process path, ultimately finds itself situated in a closely neighboring thermodynamic equilibrium state to its original state, then the mutual differences in U, S, and V will indeed be correctly described by this equation.
 
Dale said:
Figure 14.1 explains the situation. Between any two states there are an infinite number of paths. For some paths you can calculate the change in entropy using the formula. For other paths you cannot calculate the change in entropy that way, but it is the same since it is an equation of state.
What I have understood is:
For calculating entropy - change, I have to consider only the final and initial state.
Then, I have to take a suitable reversible process which leads to the system to the final state from the initial state.
Then, I have to calculate entropy change using this reversible process.
It doesn't matter whether the system reaches to the final state through this reversible process, or some other reversible process, or some irreversible process.

Is this correct so far?
 
Attachments
  • Boxed text titled 'A paradox?' listing Joule expansion reasoning: Delta Q = 0, Delta W = 0, Delta U = 0, then dQ = TdS question, concluding dQ <= TdS and Delta Q <= T Delta S with Delta S = R ln 2.
    upload_2017-10-7_17-53-51.png
    34.2 KB · Views: 854
Boxed text titled 'A paradox?' listing four points on the Joule expansion: dQ = 0, dW = 0, dU = 0 so dT = 0 for an ideal gas, and dQ = 0 implying dS = 0, followed by a note that dQ ≤ TdS and ΔS = R ln 2.

I have not understood the answer to the paradox.
Since the system is thermally isolated, the dQ is 0 for both reversible and irreversible process. So, entropy change should be 0.
What is wrong with the above argument?
 
Chestermiller said:
However, if a system starts out at a given thermodynamic equilibrium state and then, after suffering some wildly tortuous irreversible process path, ultimately finds itself situated in a closely neighboring thermodynamic equilibrium state to its original state, then the mutual differences in U, S, and V will indeed be correctly described by this equation.
So what I understood is :
The eqn. holds true locally, not globally.
That is if the change in U,S,V are very small (i.e infinitesimally small) between the initial and final state, then the eqn. holds true.
Smaller the change in U,S,V, more correct it is to use the eqn.14.18.
Is this correct?
But how can one show that the eqn. hold true in this case?
 
Pushoam said:
t doesn't matter whether the system reaches to the final state through this reversible process, or some other reversible process, or some irreversible process.
Yes
 
Dale said:
Why? What equation leads to this conclusion?
Since the system is thermally isolated,dQ =0, irrespecctive which process( reversible or irreversible) it is going through.
##dS = \frac {dQ_{rev}} T = 0##
 
Pushoam said:
Since the system is thermally isolated,dQ =0, irrespecctive which process( reversible or irreversible) it is going through
Yes, sort of. In Joule expansion the system is thermally isolated so dQ = 0. However, if you are considering a different path to a given state, then dQ=0 may not hold.

Pushoam said:
##dS = \frac {dQ_{rev}} T = 0##
Joule expansion is irreversible so this equation does not hold. Instead ##dS \ge \frac {dQ} T = 0##
 
Last edited:
Pushoam said:
So what I understood is :
The eqn. holds true locally, not globally.
That is if the change in U,S,V are very small (i.e infinitesimally small) between the initial and final state, then the eqn. holds true.
Smaller the change in U,S,V, more correct it is to use the eqn.14.18.
Is this correct?
But how can one show that the eqn. hold true in this case?
The equation is correct for all reversible paths (which consist of a continuous sequence of thermodynamic equilibrium states), so it must be true in going differentially to a neighboring thermodynamics equilibrium state.
 
Pushoam said:
View attachment 212520
I have not understood the answer to the paradox.
Since the system is thermally isolated, the dQ is 0 for both reversible and irreversible process. So, entropy change should be 0.

In an isolated system dQ = 0 because no heat is exchanged between the system and the environment. However, depending on the initial state immediately after isolation, even in isolated systems "internal things" can happen between the various parts of the system, e. g., reactions between components, redistribution of components by means of diffusion, phase transition etc., until the overall system has attained its absolute thermodynamic equilibrium state. That’s the general meaning of ΔSsystem >= 0 for isolated systems. Whatever happens inside an isolated system results in an increase of the system's entropy.
 
Pushoam said:
Since the system is thermally isolated,dQ =0, irrespecctive which process( reversible or irreversible) it is going through.
What you don't seem to realize is that the system is not thermally isolated in the reversible process which is used for the calculation (check it again).

The idea of the calculation is this: We don't know how to calculate things for the Joule expansion so let's pretend we have a different physical process with the same initial and final state as in the Joule expansion. So instead of the thermally isolated expansion (where no heat is exchanged), we calculate things for the reversible isothermal expansion (where heat is exchanged).
 
Last edited:
  • Like
Likes   Reactions: Dale
Pushoam said:
What I have understood is:
For calculating entropy - change, I have to consider only the final and initial state.
Then, I have to take a suitable reversible process which leads to the system to the final state from the initial state.
Then, I have to calculate entropy change using this reversible process.
It doesn't matter whether the system reaches to the final state through this reversible process, or some other reversible process, or some irreversible process.

Is this correct so far?
Yes.
 
  • Like
Likes   Reactions: Chestermiller