What the Second Law Actually Says
The slogan “entropy always increases” is too blunt.
A safer macroscopic statement is:
The total entropy of an isolated system does not decrease.
For an ordinary spontaneous irreversible process,
For an ideal reversible limit,
So the law does not demand that every object, every room, every organism, or every small region increase its own entropy at every moment.
The bookkeeping boundary matters.
An isolated system exchanges neither matter nor energy with its surroundings. A closed system can exchange energy but not matter. An open system can exchange both.
When people seem to find a local violation of the second law, the usual first question should be:
What did we choose as the system, and what happened in the surroundings?
System vs. Surroundings
Suppose the entropy of a subsystem decreases:
That alone tells us nothing about whether the second law has been violated.
If the surroundings gain more entropy than the subsystem loses, then
and therefore
This is the central idea of the page.
Entropy is not a substance that must rise inside every individual object. Thermodynamics keeps track of the complete process across the relevant boundary.
That distinction also explains why visible organization is not enough to judge entropy. For that separate issue, see Why Isn't Entropy Just “Disorder”?.
A Refrigerator Lowers Entropy Locally
A refrigerator is the clearest everyday example.
As food and air inside a refrigerator cool, their entropy can decrease. If that were the whole story, refrigeration would look suspicious.
But a refrigerator does not cool its interior for free.
Its compressor requires work, usually supplied by electricity. The refrigeration cycle removes heat from the colder interior and rejects heat into the warmer room.
The heat dumped into the room is greater than the heat removed from the cold interior because the work input also ends up as energy that must be rejected.
So the process is roughly:
cold interior loses heat
+ electrical work enters the refrigerator
→ more heat is dumped into the room
For a real refrigerator, the entropy gained by the warmer surroundings exceeds the entropy lost by the cooled contents. The refrigeration process therefore produces positive total entropy.
A refrigerator therefore does not “destroy entropy” and it does not “create cold.” It uses work to move heat from a colder region to a warmer one.
One small boundary matters: once a refrigerator has reached a steady operating temperature, the entropy of its contents is not continually falling without limit. The machine then works mainly to offset heat leaking in from the room while continuing to produce entropy overall.
Freezing Can Lower a Material's Entropy
Freezing gives a second example.
When liquid water becomes ice, the entropy of the water itself decreases. The solid phase has a more restricted set of microscopic possibilities than the liquid phase under the same comparison conditions.
But freezing also releases latent heat to the surroundings.
Under conditions where freezing is thermodynamically spontaneous, the entropy gained by the surroundings is larger than the entropy lost by the water, so
That is why a process can produce a more visibly ordered material without violating the second law.
This does not mean water always freezes spontaneously whenever a thermometer reads below 0°C. Pressure, purity, supercooling, nucleation, and other conditions can affect what actually happens.
The thermodynamic statement is narrower:
when freezing occurs spontaneously under the relevant conditions, the total entropy accounting still satisfies the second law.
What About Living Organisms?
Life is another common source of confusion.
Living organisms build and maintain highly organized, far-from-equilibrium structures. Cells assemble proteins, maintain membranes, repair damage, and regulate internal chemistry.
None of that makes an organism an isolated system.
Organisms are open systems. They take in matter and usable energy and release heat, waste products, and other forms of energy and matter to their surroundings.
Maintaining local structure therefore comes with entropy production in the larger organism-plus-environment system.
It is too simple to say:
“Life has low entropy, therefore life fights the second law.”
“Organization” and thermodynamic entropy are not identical concepts. The safe point is that biological organization is maintained through continuous exchanges with the environment, and the total thermodynamic accounting remains consistent with the second law.
Historical phrases such as “negative entropy” or “negentropy” are sometimes used informally when discussing life. They should not be pictured as a literal negative-entropy substance that organisms consume.
Can Total Entropy Ever Stay Constant?
Yes—in the ideal reversible limit.
A reversible thermodynamic process is an idealization in which both the system and surroundings could be returned to their original states without leaving a net change elsewhere.
For such an ideal process,
Real processes are usually irreversible. Friction, electrical resistance, diffusion, viscosity, finite temperature differences, and other dissipative effects produce entropy.
So the second law is better remembered as:
Total entropy does not decrease.
Not:
Total entropy must always increase by a positive amount in every conceivable process.
The equality case matters.
What About Small Statistical Fluctuations?
There is a second, very different way entropy can temporarily decrease.
In sufficiently small systems, short-time entropy changes or entropy production can fluctuate downward. This is a statistical fluctuation, not the same mechanism as a refrigerator lowering a subsystem's entropy while transferring heat to its surroundings.
Statistical mechanics does not treat entropy increase as a microscopic force that absolutely forbids every reverse fluctuation.
Instead, entropy-producing trajectories are overwhelmingly favored on macroscopic scales.
For tiny systems—such as nanoscale or molecular systems—the fluctuations can become large enough to observe. Experiments and fluctuation-theorem studies directly examine such behavior.
This should not be confused with a cup of coffee spontaneously separating into a hot half and a cold half. The probability of a large macroscopic entropy decrease becomes fantastically small as the number of particles grows.
For the statistical basis of that asymmetry, see Why Does Entropy Increase?.
So there are two distinct cases:
- A subsystem loses entropy while its surroundings gain more.
- A very small system undergoes a rare statistical downward fluctuation.
They are not the same mechanism.
Is “Negative Entropy” a Real Thing?
An entropy change can certainly be negative:
That simply means the subsystem's entropy decreased during the process.
It does not mean a physical substance called “negative entropy” flowed into the system.
Entropy can be transferred along with heat and can be produced by irreversible processes, but entropy and heat are not the same quantity.
The phrase “negative entropy” can be useful historically or informally, but it should not replace ordinary thermodynamic bookkeeping.
Common Misconceptions
“Entropy can never decrease.”
Too broad. A subsystem's entropy can decrease. The second law constrains the appropriate total entropy accounting.
“A refrigerator violates the second law.”
No. It uses work to move heat from cold to hot and rejects even more heat to the room, producing positive total entropy.
“Freezing creates order, so the second law fails.”
No. The water can lose entropy while released heat raises the entropy of the surroundings by more.
“Life defeats entropy.”
No. Organisms are open systems that continuously exchange matter and energy with their environment.
“Negative entropy is a substance organisms consume.”
No. Entropy changes can be negative; “negative entropy” is not a literal thermodynamic material.
“A reversible process must still increase total entropy.”
No. The ideal reversible limit has ΔStotal = 0.
“Small entropy-decreasing fluctuations prove the second law is false.”
No. They are part of the statistical description. Large macroscopic reversals remain overwhelmingly improbable.
Visual Explanation
Arrows show heat transfer and electrical work, not entropy flowing as a substance.
One Thing to Remember
The second law does not say that every part of the world must increase its entropy separately.
A subsystem can cool or become more constrained while its entropy decreases. Visible organization by itself does not determine entropy; what matters is the complete thermodynamic accounting.
For ordinary irreversible processes, a local entropy decrease is paid for by a larger entropy increase elsewhere, so the total increases.
Go Deeper
Related Questions
- Can entropy ever decrease?
- What does the second law actually forbid?
- Can a refrigerator reduce entropy?
- Does freezing decrease entropy?
- Does life violate the second law?
- What is total entropy?
- Can entropy decrease in an isolated system?
- What is a reversible process?
- Can entropy fluctuate downward?
- What is negative entropy?