Temperature, Ideal Gas and Statistics

What is temperature ?

Something that you measure ... Operational denition

Stops at some point. At THERMAL EQUILIBRIUM after some RELAXATION TIME Physical device

General idea Physical objects when they are in contact spontaneously exchange stuff (e.g. energy, volume, particles). After some relaxation time, they reach a (thermodynamic) equilibrium. When equilibrium is reached, some other associated quantities are the same for the exchanging physical objects. Temperature is one of these. Remark : equilibrium here does not mean that nothing moves within the physical objects. For instance, we will see indeed that Temperature is a measure of thermal agitation.

Exchanged quantity

Operational quantity

Energy (heat)

Temperature

Volume

Pressure

Particles

Concentration (actually chemical potential)

What does high Temperature mean ? An object with high T will spontaneously lose energy (heat) if in thermal contact with an object at low T. A balloon at high pressure (e.g. your lungs) will lose volume to a balloon at low pressure. A solution with high concentration will spontaneously lose particles if in diffusive contact with a solution at low concentration.

Thermal and statistical physics is by essence, the study and quantication of these transfers.

Two interesting aspects : - out of equilibrium : the study of the transfer itself (can be extremely hard) - equilibrium : the study once the transfer is completed (slightly easier, we will mostly study these cases)

Units All quantities have to be measured, to be properly scaled, we use units. Crazy unit : Fahrenheit degree 0 : coldest temperature in Danzig his birthplace - during winter 1708-09 96 (=12x8) : blood temperature of a horse
Daniel Fahrenheit (1686-1736)

Good idea : based on some natural variables. Bad idea : nobody can come back to Danzig in 1708 to double check ...

Then came ....

The French revolution

The greatest Revolutionary legacy : the metric system 1790 : French Assemble Constituante votes to dene universal units. June 22 1799 : the reference meter and kg are precisely designed and stored in National French Archives. (we still use that kg) Talleyrand

Almost all physical units you know derive from these.

Temperature : Celsius degree C 0 : icing temperature at sea level 100 : boiling temperature at sea level (slightly different now) Kelvin degree
Andre Celsius (1701-1744)

1 C = 1 K 0 is 1273.16 of the thermodynamic temperature of the triple point of water

William Thomson (1824-1907)

Absolute temperature scale Experimental fact : the volume of a given mass of gas increases at the function of temperature. This connects Temperature to Mechanics!
a unit of heat descending from a body A at the temperature T of this scale, to a body B at the temperature (T1), would give out the same mechanical effect [work], whatever be the number T. Such a scale would be quite independent of the physical properties of any specic substance.

Thermodynamics temperature, Kelvin scale

Speculation : Guillaume Amontons, 1702 There is temperature for which this volume drops to 0. This temperature is absolute zero. On Amontons (imprecise) scale, is was -240 C. Lambert later provided a better estimation of - 270 C. Kelvin found the right value of -273.15 C.

Some (obvious) consequences

if you double temperature and keep volume

constant, you double pressure

if you double pressure and keep temperature

constant, the gas squeezes into half the volume absolute zero

if T=0,V=0, this was our denition of order of magnitudes : 1 mol of gas, 300 K at
1 atm (=101.3 kPa ) gives 22.4 L Ideal : dilute gas, distance between molecules much bigger than typical size of molecules.

Large numbers http://xkcd.com/980/

Total economic production of the human race over ages :

2 10 $

15

Thermodynamical number

There are 10 millions more gas molecules in one litre than the total economic production in $ of the human race in history
There are 10^11 stars in the galaxy. That used to be a huge number. But it's only a hundred billion. It's less than the national decit! We used to call them astronomical numbers. Now we should call them economical numbers. Feynman (1918 - 1988)

NA 6.02 10

23

The fundamental law of statistical world : law of large numbers Consider a random process (e.g. rolling a dice). Run it N times with N very big, and consider the average (or the sum). Theorem : in the limit of innite N, we get a gaussian (normal law)

Example : one million random digits between 0 and 9

We needed huge table of perfectly random numbers for computer simulations in the old time

00000 00001 00002 00003 00004 00005 00006 00007 00008 00009 00010 00011 00012 00013 00014 00015 00016 00017 00018 00019 00020 00021 00022 00023 00024 00025 00026 00027 00028 00029 00030 00031

10097 32533 37542 04805 08422 68953 99019 02529 12807 99970 66065 74717 31060 10805 85269 77602 63573 32135 73796 45753 98520 17767 11805 05431 83452 99634 88685 40200 99594 67348 65481 17674 80124 35635 74350 99817 69916 26803 09893 20505 91499 14523 80336 94598 44104 81949 12550 73742 63606 49329 61196 90446 15474 45266 94557 28573 42481 16213 23523 78317 04493 52494 00549 97654

76520 13586 64894 74296 19645 09303 09376 70715 80157 36147 34072 76850 45571 82406 02051 65692 05325 47048 03529 64778 14905 68607 39808 27732 06288 98083 86507 58401 87517 64969 17468 50950 17727 08015 77402 77214 66252 29148 14225 68514 68479 27686 26940 36858 85157 47954 11100 02040 16505 34484 26457 47774 95270 79953 67897 54387 97344 08721 73208 89837 75246 33824 64051 88159

34673 54876 24805 24037 23209 02560 38311 31165 64032 36653 36697 36170 35303 42614 68665 74818 90553 57548 35808 34282 22109 40558 50725 68248 13746 70078 36766 67951 91826 08928 58047 76974 45318 22374 43236 00210 36936 87203 46427 56788 46162 83554 70297 34135 32979 26575 12860 74697 40219 52563 51924 33729 59367 83848 54622 44431 16868 48767 68935 91416 45862 51025 96119 63896

80959 09117 20636 10402 15953 34764 88676 74397 98951 16877 65813 39885 86799 07439 73053 85247 28468 28709 60935 20344 60970 93433 29405 24201 18475 40610 90364 76493 93785 61368 73039 57186 21115 78253 45521 64237 76621 13990 96297 78822 94750 89923 53140 33340 57600 40881 96644 89439 43651 77082 65394 59593 82396 10118 91190 42592 03071 12059 26252 29663 61962 79335 54692 82391

39292 74945 00822 91665 35080 33606 04436 27659 12171 76833 11199 29170 23403 09732 18623 88579 83491 25624 35273 88435 50500 73998 52775 67851 68711 77817 29609 11062 23478 34113 40218 16544 14385 53763 96286 02655 94400 56418 54382 14598 37089 20048 42050 82341 22222 06413 28707 25815 07207 31790 42582 60527 33211 59466 92927 45973 25701 46670 05522 82562 65337 12472 23287 29529

Here is the game : we average N consecutive numbers in this table. We do that until we exhaust the table. What is the statistics of the sum ?

N=1 (10^6 numbers) 0 1 2 3

4
5
6
7
8
9
0.099851
0.100043
0.100634
0.10031


0.100086

0.100208
0.099937

0.099555

0.100099

0.099277

N=10 (10^5 averages of 10 numbers)

0 1 2 3 4 5 6 7 8 9

5e-05 0.00215 0.04351 0.23734 0.4115 0.25202 0.05075 0.00265 3e-05 0.0

(with 20 bins)

N=100 (10^4 averages of 100 numbers)

0 1 2 3 4 5 6 7 8 9

0.0 0.0 0.0001 0.0419 0.9196 0.0384 0.0 0.0 0.0 0.0

(with 100 bins)

N=1000 (1000 averages of1000 consecutive numbers)

(with 250 bins)

Distribution of the average converges toward a Gaussian Same average as the original random process (center of the gaussian)
N = N

(width of the gaussian)

Here : Average : Variance :

= 4.5 = E((X ) ) = 2.9
2 2

N=10 (10^5 sums of 10 numbers) Half-with

2 2 ln 2 2.355

0 1 2 3 4 5 6 7 8 9

5e-05 0.00215 0.04351 0.23734 0.4115 0.25202 0.05075 0.00265 3e-05 0.0

(with 20 bins)

10 0.9

N=1000 (1000 sums of1000 consecutive numbers)

2 2 ln 2 2.355

1000 0.09

For one random process governed by a gaussian, you have (statistically): - about 68% to be within one sigma of the average - about 95% to be within 2 sigma of the average - about 99.7 % to be within 3 sigma of the average - about 99.99 % to be within 4 sigma of the average

This is important for all measurements : e.g. Higgs Boson (beware the lookelsewhere effect)

Molecular model of Ideal gas It is possible to derive Ideal gas law from a microscopic description. Pressure P Temperature T Volume V N molecules

here there is no energy/particle/volume gain.

One gas molecule moving towards the container Momentum

m v

Elastic shock : transfers to the container 2mvx How many of these during t ?

vx t

Gas molecules at distance lower than vx t contribute Cylinder vx St

S

How many molecules ?

Pressure P Temperature T Volume V N molecules

Density is N/V

vx t

N/V vx tS molecules in

that volume Density Volume

But molecules move in 3D, 6 S directions

=> only the vx > 0 molecules contribute

vx t

1N vx tS molecules hit surface 2V S within t

Each gives momentum 2mvx

S

Total momentum received :

N 2 p = mvx tS V

In reality, each molecule has a different speed, we can actually safely average to get :
N 2 p = mvx tS V

vx
2 vx
2

: average speed along x : average speed squared along x

+ v2 + v2 2 v = vx y z
Isotropy :

= v2 = v2 2 vx y z
1 = v2 2 = vx 3

vx t

Total momentum received :

1N 2 tS p = mv 3V
S

p 1N 2 S = mv t 3V
This is a force By unit of surface

... i.e. the Pressure !

p 1N 2 S = mv t 3V

1N 2 P = mv 3V
1 P V = N mv 2 3

kinetic pressure

1 2 P V = N mv 3

P V = N kT

3kT = 2 v m
Speed squared is proportional to temperature Thermal agitation

Kinetic translational energy :

1 3 2 = kT mv 2 2

1 3 + v 2 + v 2 ) = kT 2 m(vx y z 2 2
Isotropy :

1 1 1 1 2 2 2 mvx = mvy = mvz = kT 2 2 2 2

Consequence : temperature is a measure of energy.

Equipartition theorem Consider all form of mechanical energy : kinetic energy, rotational energy, vibrational motion, elastic potential energy (stored in a spring). We call it degree of freedom

1 2 1 2 1 2 1 2 1 2 mvx , mvy , mvz , Ix , kx 2 2 2 2 2

Theorem : the average energy of any quadratic degree

1 of freedom is kT 2

Examples

Monoatomic gas 3 Translational DOF No rotational DOF (indistinguishable from QM)

3 kT 2

Examples

Diatomic gas at low T 3 Translational DOF QM mechanics tells us there is 2 rotational DOF

5 kT 2

Examples
1 2 mv 2 1 2 kx 2

Diatomic gas at higher T 3 Translational DOF QM mechanics tells us there is 2 rotational DOF Vibrational modes at high T : 2 extra DOF (recall this later for Einstein solid)

7 kT 2

Examples : Solid can be seen as a lattice of little masses connected by springs. Per atom, on average: 3 translation 3 vibration

3kT