The definition of spontaneous is often briefly glossed over in most of the thermodynamics texts that I own. Peter Atkins in Physical chemistry defines spontaneous as follows

Some things happen naturally, some things don’t. Some aspect of the world determines the spontaneous direction of change, the direction of change that does not require work to bring it about.

Later in the text and in many others you then find that

The entropy of an isolated system increases in the course of a spontaneous change: $\Delta S_{tot} > 0$

These two definitions of spontaneity lead me to two conclusions:

1: That net work must be done on a system in order to bring about a non-spontaneous process. Alternatively, systems may only do net work on their surroundings through spontaneous processes.

2: That there is no such thing as a non-spontaneous process in this universe.

My question pertains to my second conclusion. Is this conclusion valid? The thought process that leads me to this conclusion is as follows: The universe is an isolated system. Any spontaneous change in an isolated system increases the entropy of that system. The second law dictates that the only processes possible are those that increase the entropy of the universe. Thus, only spontaneous processes are possible. As an example, a fridge which transfers energy from a cold sink to a heat sink must be spontaneous process. This is because in order for this to occur, we must perform work on the fridge and that work is coming from a heat engine(i.e a power plant) which is spontaneous. The ultimate result is that the refrigerator increases the entropy of the universe and thus is spontaneous. Is this correct?

I guess my question could be more succinctly stated as follows: Is the set of possible spontaneous processes equal to the set of all thermodynamically allowable processes? That is, if a process is non spontaneous, is it forbidden to occur by the laws of thermodynamics?

Any help on this issue would be greatly appreciated!

  • $\begingroup$ The fridge example is really confusing - it actually decreases the entropy. $\endgroup$
    – Roger V.
    Feb 25, 2021 at 9:36
  • $\begingroup$ @Vadim A fridge surely does not decrease the entropy of the universe? $\endgroup$ Feb 25, 2021 at 10:29
  • 2
    $\begingroup$ It seems that the logical issue here is what you define as a system, to which you apply the terms like entropy increase/decrease, spontanuity, etc. Sure, the whole of the unievrse can be thought of as a system spontaneously evolving towards equilibrium, increasing its entropy. But when you focus on a fridge, you are far from considering the whole universe. $\endgroup$
    – Roger V.
    Feb 25, 2021 at 10:37

4 Answers 4


You can see it like this: if you consider the universe as a whole, "of course" each process is "spontaneous" in the sense that the total entropy of the universe entropy always increases. After all, if something happens somewhere in your universe, then it must be allowed to happen: otherwise, it just would not happen!

However, what you ususally care about is that your fridge stays cold, and that is why thermodynamics focuses on sub-systems: it is not enough to say "somewhere in the universe there is the possibility of a fridge". You want to know about your fridge (and your electrical bill at end of the month and the natural resources of your planet!).

A question about a spontaneous process would be: if I buy a fridge, would it get cold? And the answer is: no (unless you plug it to an external power source)! If you turn the question to "do cold fridges exist in the universe without any external assumption except the existence of the universe" then the answer is, of course, yes.

So depending on the scale you look at things (fridge < electrical bill < planet < universe) things can be considered spontaneous or not. You need the current for your fridge, you need a power plant for the current, you need the sun for the power plant, you need the big bang for the sun, etc.). So you need to choose: at what scale do I look at the system: would a fridge work without current? Would current work without the sun? Would the sun work without the univse?

Can you say this in more scientific words?

The universe as a whole only has processes which increase the total entropy. Now, let us focus on the entropy of a sub-system! Here is the tricky point: a sub-system of the universe, if you consider it as an individual object, could locally decrease its entropy if you supply work to it, i.e. it could to something you would NOT expect to happen unless you considered it as part of a bigger system.

So, let's say I have a subsystem, I see its evolution. Then the question is:

do I have to resort to the presence of an outside energy source in my system to explain its behavior?

If yes, then the process is not spontaneous. If not, then the process is.

A spontaneous process inside a system the you are observing is one that you can explain without resorting to an outside energy source.

Extended discussion with a simple example

Imagine you have a system which is not in an equilibrium state. How did it get there is irrelevant.

The simplest example is two identical boxes with a perfect gas inside, one at temperature $T_1$, one at temperature $T_2$ with $T_1>T_2$. If they are put somehow in contact, then, without any additional external work, heat will flow from the hot to the cold bath ($T_1\to T_2$) until the two temperatures are the same at the equilibrium temperature $T_e=(T_1+T_2)/2$.

You did not do anything: just put the two gases in contact and they will spontaneously change their temperature to the equilibrium state.

Now, imagine you want the cold gas to stay cold (a fridge) despite its hot sourrounding: to counter the spontaneous process you need to remove heat from the cold gas to keep it cold. To do that, you need some work (the energy given by the current to the fridge).

In this latter case, the entropy of the universe will increase, but the entropy of the sub-system "fridge" will not, because you are artificially keeping it constant. You are basically removing entropy from the fridge at the expense of the universe. But you need to assume that there is an outisde universe that is providing the extra energy required!

Let's reverse the reasoning. You see two gases together, at $T_1$ and $T_2$. You observe them and see that their temperature does not change. Then it must mean that there must be somewhere some work given to the system to maintain them as they are. On the other hand, if you work at the two gases and see that they equilibrate, then you can assume the system is behaving as it should be if there is no work done on it. That a process is spontaneous means you can describe it without resorting to an external energy source: that is how two isolated gas should behave.

If now you include the whole universe then of course you can explain why the fridge stays cold: the universe already contains the energy source, so there is no inconsistency. And because the universe is isolated, you don't see any process inside the universe that requires an energy source outside your system: it's already somewhere in your system.

Summing up, (not-)"spontaneous" actually can only be applied to sub-systems. If you consider only a subsystem, which can exchange heat with the rest of the universe (so not an isolated system) you can make it evolve towards states which the same subsystem, if isolated, would never evolve to. We call spontaneous processes those that would happen also if the system was completely isolated.

In order to have a non - spontaneous process, you need some sort of energy flux (work, heat..) from the outside. In the big picture of the universe, on the other hand, everything is spontaneous, in a way.

It really depends on your point of view when describing the system.

  • $\begingroup$ (1/2) Thanks for the excellent response! Okay, it pretty much all makes sense now. Just one thing before I accept the answer. When you state : "do I have to resort to the presence of an outside energy source in my system to explain its behavior? If yes, then the process is not spontaneous..." , do you not mean outside work ? Suppose our system is the inner compartment of a fridge and it gets colder than the surroundings. This is clearly not spontaneous and requires work done on the system to occur. Now suppose our system is a gas cylinder in a basic heat engine which undergoes a cycle. $\endgroup$ Feb 25, 2021 at 11:04
  • $\begingroup$ (2/n) After the cycle is complete, $\Delta S_{uni}>0$ and $\Delta S_{sys}=0$. Would this be an example of a spontaneous process? We do have to resort to an external energy source to explain its behaviour (the heat source) but we dont require an outside source of work to explain its behaviour. So depending on whether we rely on outside work or outside energy to determine spontaneity, this may or may not be a spontaneous process. $\endgroup$ Feb 25, 2021 at 11:05
  • $\begingroup$ Outside energy can be work but also heat (and at some level of course to generate heat you usually need work in the first place as in a power plant ;) ) $\endgroup$
    – JalfredP
    Feb 25, 2021 at 12:42
  • $\begingroup$ In a coclea you have to put some work in the system at some point. Then you emit heat and the net energy flux can in principle be 0, but in some part of the coclea you need work! So that part of the cycle is not spontaneous. $\endgroup$
    – JalfredP
    Feb 25, 2021 at 12:45
  • $\begingroup$ This is a very misleading answer because one cannot resort on reasoning only on energy or work balances to decide about the spontaneity of a process, else the first law would be sufficient. Such reasoning might work in some very specific, well chosen examples, but not in general. $\endgroup$
    – The Quark
    Jun 24, 2023 at 15:10

The two definitions you cite do not mean that non-spontaneous processes decrease entropy, if that is what you are implying. In other words, the fact that all spontaneous (natural) processes increase entropy does not mean all non-spontaneous processes should decrease entropy.

The refrigeration/heat pump cycle is a non-spontaneous transfer of heat from cold to hot, but as you know a net work input is required to do so and that results in an increase in entropy for a real cycle (zero change for an ideal reversible cycle), not a decrease. If no net work input were required, there would be a decrease in entropy and a violation of the Clausius's Statement of the Second Law:

No refrigeration or heat pump cycle can operate without a net work input.

Hope this helps.

  • $\begingroup$ Thanks for the response. Would it be accurate to summarize as follows: A spontaneous process is any process that can occur within an isolated system without work being done on the system by the surroundings. Similarly, a non spontaneous process is any process that cannot occur in an isolated system without being work done on the system by the surroundings. Having said this, would the cycle of a heat engine be spontaneous? It results in a net work output and hence doesn't require net work input and hence is spontaneous? $\endgroup$ Feb 26, 2021 at 7:32
  • $\begingroup$ @SalahTheGoat Though I'm wary of such generalizations I suppose what you say about spontaneous processes is accurate provided it is understood that spontaneous processes can occur due to work performed (and heat exchanged) within the isolated system. Regarding the heat engine cycle I'm not sure what you are getting at. It always requires a net heat input not a net work input. The refrigeration cycle requires a net work input. In any case, if either is reversible then there is no overall increase in entropy in which case they can't be spontaneous. $\endgroup$
    – Bob D
    Feb 26, 2021 at 15:51
  • $\begingroup$ @SalahTheGoat This way of reasoning gives the feeling of bringing clarifications, but it is a false perception: a spontaneous process is a process that can occur, period. This is actually what the term "spontaneous" recapitulates. Yet giving the meaning of the term doesn't help to decide if any described process (whether or not it takes place in an isolated system) is spontaneous. And as I commented under the reply of JalfredP, one cannot resort on reasoning only on energy or work balances else the first law would be enough. $\endgroup$
    – The Quark
    Jun 24, 2023 at 15:44

Two types of processes:

  • There are thermodynamics processes that change the state of some variables like temperature and entropy.

  • In statistical mechanics there are microscopic processes which move particles between microstates, this change may move the system closer to the equilibrium macrostate or not.

There is no such thing as a spontaneous thermodynamic process, in the sense of a system spontaneously doubling its volume.

There are spontaneous and stimulated (non-spontaneous) microscopic processes.

Spontaneous processes tend to move the system to the equilibrium macrostate, because they allow energy to be distributed ergodically across all degrees of freedom of the particles involved.

For example, a laser operates by the competition of spontaneous emission and simulated (non-spontaneous) emission processes.

In a laser, energy flows into a system, initially, this causes the spontaneous processes to dominate, as optical modes are populated more or less equally. After a certain time stimulated processes will dominate: photons begin to populate a single optical mode. This is a highly ordered macrostate and is only maintained by energy flowing into the system.

I’ve tried to add how I would approach thinking about this, it was too long for a comment.


These are interesting questions.

1: That net work must be done on a system in order to bring about a non-spontaneous process. Alternatively, systems may only do net work on their surroundings through spontaneous processes.

The first statement might be considered true* but it does not rank as a definition; the second is wrong.

(* Note that if a process "has been made" spontaneous, it cannot strictly be considered as the same process anymore. So this rather common way of stating that "a non-spontaneous process becomes spontaneous", even though frequently encountered, is actually improper.)

Think about a system at a lower pressure than the exterior. It will spontaneously contract, and thereby receive work: $dW = -P_{\rm{ext}}dV > 0$. Inversely, the non-spontaneous expansion would correspond to the system delivering work.

And more generally, since one would get the reverse results by following a similar reasoning with a system at a higher pressure than the exterior, it follows that whether work is received or delivered is not a criterion of spontaneity.

2: That there is no such thing as a non-spontaneous process in this universe.

This is actually the logical negative ($\neg$) equivalent form of the definition of spontaneous, it is excessive to call it a "conclusion", rather a reformulation of the definition of the term.

Is the set of possible spontaneous processes equal to the set of all thermodynamically allowable processes? That is, if a process is non spontaneous, is it forbidden to occur by the laws of thermodynamics?

Yes, by definition. With the caveat regarding whether reversible processes qualify as "spontaneous" and "allowable" since they are limiting cases...

(A side note: "possible spontaneous" can be seen as a pleonasm here depending on the meaning of the term "possible" at this place in the sentence...)

And yet, all this still doesn't give an operational definition (i.e. a criterion) that would allow one to determine whether a given process should be qualified as spontaneous or non-spontaneous...


Your Answer

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

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