Physics 449/451 - Statistical Mechanics - Course Notes
David L. Feder
September 13, 2011 Contents
1 Energy in Thermal Physics (First Law of Thermodynamics) 3 1.1 Thermal Equilibrium ...... 3 1.2 TheIdealGas...... 3 1.2.1 Thermodynamic Derivation ...... 3 1.2.2 Mechanical Derivation ...... 4 1.3 EquipartitionofEnergy ...... 6 1.4 HeatandWork...... 7 1.5 CompressionWork:theAdiabat ...... 7 1.6 HeatCapacity...... 9
2 The Second Law of Thermodynamics (aka The Microcanonical Ensemble) 13 2.1 Two State Systems (aka Flipping Coins) ...... 13 2.1.1 Lots and lots of trials ...... 15 2.1.2 Digression: Statistics ...... 16 2.2 Flow toward equilibrium ...... 17 2.3 LargeSystems ...... 18 2.3.1 DiscreteRandomWalks ...... 18 2.3.2 ContinuousRandomWalks ...... 20 2.3.3 QuantumWalksandQuantumComputation ...... 22 2.4 Entropy ...... 25 2.4.1 Boltzmann ...... 25 2.4.2 ShannonEntropy...... 26 2.4.3 vonNeumannEntropy...... 28
3 Equilibrium 32 3.1 Temperature ...... 32 3.2 Entropy,Heat,andWork ...... 34 3.2.1 ThermodynamicApproach ...... 34 3.2.2 StatisticalApproach ...... 35 3.3 Paramagnetism...... 36 3.4 Mechanical Equilibrium and Pressure ...... 38 3.5 Diffusive Equilibrium and Chemical Potential ...... 39
4 Engines and Refrigerators 41 4.1 HeatEngines ...... 41 4.2 Refrigerators ...... 44 4.3 RealHeatEngines ...... 44
1 PHYS 449 Course Notes 2009 2
4.3.1 Stirling Engine ...... 44 4.3.2 SteamEngine...... 46 4.3.3 InternalCombustionEngine...... 47 4.4 RealRefrigerators ...... 49 4.4.1 HomeFridges...... 49 4.4.2 Liquefaction of Gases and Going to Absolute Zero ...... 50
5 Free Energy and Chemical Thermodynamics 51 5.1 FreeEnergyasWork...... 51 5.1.1 Independent variables S and V ...... 51 5.1.2 Independent variables S and P ...... 52 5.1.3 Independent variables T and V ...... 52 5.1.4 Independent variables T and P ...... 53 5.1.5 ConnectiontoWork ...... 53 5.1.6 Varying particle number ...... 54 5.2 Free Energy as Force toward Equilibrium ...... 54
6 Boltzmann Statistics (aka The Canonical Ensemble) 56 6.1 TheBoltzmannFactor...... 56 6.2 Z andtheCalculationofAnything ...... 58 6.2.1 Example: Pauli Paramagnet Again! ...... 60 6.2.2 Example: Particle in a Box (1D) ...... 62 6.2.3 Example: Particle in a Box (3D) ...... 63 6.2.4 Example: Harmonic Oscillator (1D) ...... 64 6.2.5 Example: Harmonic Oscillator (3D) ...... 65 6.2.6 Example:Therotor ...... 66 6.3 The Equipartition Theorem (reprise) ...... 68 6.3.1 DensityofStates ...... 69 6.4 The Maxwell Speed Distribution ...... 71 6.4.1 InterludeonAverages ...... 73 6.4.2 MolecularBeams...... 73 6.5 (AlreadycoveredinSec.6.2) ...... 75 6.6 Gibbs’Paradox...... 75
7 Grand Canonical Ensemble 77 7.1 ChemicalPotentialAgain ...... 77 7.2 GrandPartitionFunction ...... 78 7.3 GrandPotential...... 80
8 VirialTheoremandtheGrandCanonicalEnsemble 81 8.1 VirialTheorem ...... 81 8.1.1 Example:idealgas...... 82 8.1.2 Example: Averagetemperatureofthesun ...... 82 8.2 ChemicalPotential...... 83 8.2.1 Free energies revisited ...... 84 8.2.2 Example:PauliParamagnet...... 85 8.3 GrandPartitionFunction ...... 85 8.3.1 Examples ...... 87 8.4 GrandPotential...... 87 PHYS 449 Course Notes 2009 3
9 Quantum Counting 89 9.1 Gibbs’Paradox...... 89 9.2 ChemicalPotentialAgain ...... 91 9.3 Arranging Indistinguishable Particles ...... 92 9.3.1 Bosons...... 92 9.3.2 Fermions ...... 93 9.3.3 Anyons!...... 95 9.4 Emergence of Classical Statistics ...... 96
10 Quantum Statistics 99 10.1 BoseandFermiDistributions ...... 99 10.1.1 Fermions ...... 100 10.1.2 Bosons...... 102 10.1.3 Entropy...... 104 10.2 Quantum Classical Transition ...... 106 10.3 EntropyandEquationsofState...... 107
11 Fermions 110 11.1 3DBoxatzerotemperature...... 110 11.23DBoxatlowtemperature ...... 111 11.3 3Disotropicharmonictrap ...... 113 11.3.1 DensityofStates ...... 113 11.3.2 LowTemperatures ...... 114 11.3.3 SpatialProfile ...... 115 11.4AFewExamples ...... 117 11.4.1 Electrons in Metals ...... 117 11.4.2 ElectronsintheSun ...... 117 11.4.3 Ultracold Fermionic Atoms in a Harmonic Trap ...... 118
12 Bosons 119 12.1QuantumOscillators ...... 119 12.2Phonons...... 120 12.3BlackbodyRadiation...... 123 12.4 Bose EinsteinCondensation ...... 126 12.4.1 BECin3D ...... 126 12.4.2 BEC in Lower Dimensions ...... 127 12.4.3 BECinHarmonicTraps...... 129 PHYS 449 Course Notes 2009 2
Introduction
The purpose of these course notes is mainly to give your writing hand a break. I tend to write lots of equations on the board, because I want to be rigorous with the material. But I write very quickly and my handwriting isn’t pretty (this is probably a huge understatement). So these course notes contain (hopefully) all the equations that I will be writing on the board, so that when you take notes during class you can focus on the concepts and my mistakes, rather than furiously trying to scribble down everything I am writing, which will probably contain mistakes anyhow. Not to say that these notes don’t contain mistakes! These notes also have occasional non mathematical expressions (i.e. sentences). Chapter 1
Energy in Thermal Physics (First Law of Thermodynamics)
This chapter deals with very fundamental concepts in thermodynamics, many of which you can intuit from your experience.
1.1 Thermal Equilibrium
Some questions to ponder:
What are the ways that you measure room temperature? • What are the ways to measure temperatures that are much hotter? Colder? • What exactly is temperature? • What is absolute zero? • How do systems reach a given temperature? • What does it means to say a system is in equilibrium? • What is thermal equilibrium? • 1.2 The Ideal Gas 1.2.1 Thermodynamic Derivation Robert Boyle (1627 1691) was an Irish alchemist (!) who helped to establish chemistry as a legitimate field. After much observing, he found in 1662 that gases tended to obey the follow equation:
P V = k, (1.1) where P is the pressure, V is the volume, and k is some constant that depends on the specific gas. This equation was known as Boyle’s Law. In 1738 Daniel Bernoulli derived it using Newton’s
3 PHYS 449 Course Notes 2009 4 equations of motion (see more about this in the next section), under the assumption that the gas was made up of particles too tiny to see, but no one paid any attention because these particles were not believed to actually exist. Later, Joseph Louis Gay Lussac (1778 1850) observed that at constant pressure, one always has V T , or V1T2 = V2T1. This is known as Charles’ Law after some guy named Charles. BenoˆıtPaul∝Emile´ Clapeyron (1799 1864) put the two laws together to obtain
P V = P0V0 (267 + T ) , (1.2) in which the temperature is measured in degrees Celcius. The number 267 came from observations of Gay Lussac. This was pretty impressive, since absolute zero is known today to be 273.15◦C. Lorenzo Romano Amedeo Carlo Avogadro di Quaregna (Quaregga) e di Cerreto (1776 1856), otherwise known as Avogadro, showed in 1811 that the P0V0 out front of Clapeyron’s equation was related to the ‘amount of substance’ of the gas, and wrote:
P V = nR (267 + T ) , (1.3) where n is the number of moles of the gas, and R =8.31 J/mol/K is a universal constant (indepen dent of the type of gas). It’s easier to think of the number of particles N (atoms or molecules) rather 23 than the number of moles, so one can write N = nNA, where NA =6.02214179(30) 10 is known as Avogadro’s number and corresponds to the number of atoms in a mole of gas.× Finally, if we measure temperature in units of Kelvin (K = 273.15+◦ C) and make the substitution R = NAkB, 23 where k =1.381 10− J/K is Boltzmann’s constant, we finally obtain the ideal gas law: B ×
P V = NkBT. (1.4)
This is nice because we don’t have to worry about moles. What’s a mole anyhow??
That’s about all the history you’re going to get!
The ideal gas law (1.4) is called the Equation of State for the ideal gas, because it relates the three state variables (thermodynamic coordinates) P , V , and T at equilibrium. These are macroscopic properties of the system, and their relationship doesn’t depend on how the system was changed to get to where it is. I could write it instead as P = P (V,T ), i.e. the pressure is given by some function that depends on two variables V and T . All systems at equilibrium have such constitutive equations, relating parameters of the system over which one has external control. The van der Waals (non ideal) gas, for example, was found to have the equation of state a Nk T = P + (V b) , (1.5) B V 2 − where a and b are molecule dependent constants. Magnets have a similar equation that relates the temperature, magnetic field B, and the amount of magnetization M. See Section 3.3 for more details.
1.2.2 Mechanical Derivation The ideal gas law can also be derived from first principles, which is what both Bernoulli (approx imately) and Boltzmann did (thus the konstant with his initial on it) by assuming an atomistic theory. We’ll do a good job of this in Chapter 3, but a rough job of it can be done right now. Let’s assume that there is a single molecule or atom, bouncing around elastically in a volume V = AL, PHYS 449 Course Notes 2009 5
Figure 1.1: One molecule bouncing into a piston. where the length along the x direction is L, as shown in Fig. 1.1. After many bounces against the piston, the average pressure directly on it is
∆vx F , on piston F , on molecule m ∆t P = x = x = . (1.6) A − A − A
Let’s set t = 2L/vx, the time it takes to make a full round trip. When it undergoes one elastic collision, its change in velocity is v = 2v . Putting these together gives x − x
∆vx m ∆t 2mv mv2 P = = x = x . (1.7) − A 2AL/vx V If there were actually lots of molecules, each colliding with each other and the walls totally elastically, then we can forget the averageness of the pressure to obtain
N 2 2 2 2 P V = m vx 1 + vx 2 + ... = m vx i = mNvx = NkBT (1.8) i=1 X using the ideal gas law at the end. So we have the mean kinetic energy per molecule in the x direction is equivalent to temperature: K 1 1 x = mv2 = k T. (1.9) N 2 x 2 B This means that the mean kinetic energy is 1 1 1 1 K = mv2 + mv2 + mv2 = mv2 = (3/2)Nk T. (1.10) 2 x 2 y 2 z 2 B
The Root mean square (RMS) speed (which is usually close to the average value) of each atom/molecule 2 is defined as vRMS = v and is equal to 3kBT/m in this case. Suppose we have oxygen at room p p PHYS 449 Course Notes 2009 6
26 temperature. Diatomic oxygen has atomic mass 32, which is 5.297 10− kg. Assuming room ∗ temperature is 300 K, one obtains vRMS = 484 m/s. This is really fast! How fast? Consider that the speed of sound in air at standard temperature and pressure is only 343 m/s. Is it reasonable that the mean speed of the atoms is higher than the sound speed?
1.3 Equipartition of Energy
1 The equipartition theorem states that ‘every degree of freedom contributes an energy of 2 NkB T to the total energy of the system.’
To make sense of this, we need to talk about what a degree of freedom actually is. We’ll be much more rigorous about it in Section 1.13, but because this will be months from now it makes sense to talk about it now. A degree of freedom tells you about what the atom or molecule can actually do. Maybe it can move around (translate), or jiggle (vibrate) or turn around (rotate), or scream, 1 or shine, or a host of other things. Each one of these capabilities contributes an energy of 2 kB T to its energy. Likewise, suppose all your particles were constrained in some way to move only in one dimension; then the total translational energy would be a kB T factor smaller than molecules moving in free space. The real value of the equipartition theorem is that it allows you to calculate the total energy without having to do any work. So lazy people use it all the time in the real world!
It helps to remember your classical mechanics, though. For example, the kinetic energy of a rotating body is given by something that looks like 1 K = I ω2 + I ω2 + I ω2 , (1.11) rot 2 xx x yy y zz z where Iii are the moments of inertia in the diagonalized inertial basis, and the ωi are the angular frequencies in radians. Each of these terms is quadratic, so each one corresponds to a degree of freedom. But recall that the energy of an oscillator is made up of both a kinetic energy and potential energy: 1 m E = p2 + p2 + p2 + ω2 + ω2 + ω2 , (1.12) osc 2m x y z 2 x y z where I have already used that ωi = ki/m. Because there are two kinds of quadratic terms comprising the energy in each dimension, the three dimensional oscillator contributes 3k T per p B particle!
So the total energy of the system (in some limit that will be explained more clearly later on) is f U = Nk T, (1.13) 2 B where f is the number of degrees of freedom.
Many examples of equipartition will be covered in class.
It is interesting to note that, even though these results were derived under the assumption of an ideal gas, the equipartition theorem remains correct even when the gases are interacting. In fact, it can be applied to a huge variety of physical systems, many of which would surprise you. PHYS 449 Course Notes 2009 7
1.4 Heat and Work
So far we have seen a connection between temperature and energy through the equation of state of the ideal gas and the idea of molecules bouncing around in a box. Using the equipartition theorem, it looks like the temperature is closely related to the heat of a system, if one considers the heat as being somehow related to the amount of energy a system has. In fact, the word ‘heat’ in thermodynamics is not the same as what we think of as the amount of ‘hotness,’ i.e. is not equivalent to the temperature.
In fact, ‘heat’ is the amount of energy that spontaneously flows between two systems: heat moves from the system at higher temperature to the system at lower temperature. You might worry that this definition is circular, though, because temperature is itself a measure of a system’s tendency to either absorb or release energy. But now we can define temperature slightly better through the equipartition theorem: it is a measure of the total energy of the system U i.e. T 2U/fkB, where f is the number of degrees of freedom. ≈
D∗: Heat: The heat Q is the energy added to the system by a spontaneous (i.e. not forced) process.
Please take careful note: the magnitude of Q is positive if heat is added to the system, and is negative if it is removed. Of course, we could also change a system’s energy by doing something to it. This is called ‘work,’ and it roughly corresponds to what you naturally think of as work: it takes work to do something to something! But it corresponds to anything where forces are present, it doesn’t have to be you doing the work.
D: Work: The work W done on a system is the energy added to the system by a forced process.
Again, the work is positive if energy is added to the system; one says that work is done on the system. If energy is expended (released) by the system by the work, then one says that work is done by the system. But probably these concepts are familiar to you, because they are the same ones used in classical mechanics. Work in that case was defined as
W = F dr, (1.14) Z and it is the same now. Of course, in situations where the applied force doesn’t change with position, then it reduces to W = F r. Anyhow, heat and work together yield the First Law of Thermodynamics: U = Q + W. (1.15) This is really just a statement of the conservation of energy, so it is weird that it is called a Law, but there you go. Processes of heat transfer include conduction, convection, and radiation.
1.5 Compression Work: the Adiabat
There are many ways to do work on a system (or for a system to do work). One of the simplest is to squeeze it, such as pushing or pulling the piston in Fig. 1.1. In the case where we push on the piston (force to the left), the work is W = F x. Now, F P A as long as the pressure is always − ≈ ∗The D stands for a definition. PHYS 449 Course Notes 2009 8 uniform and at equilibrium during the operation. So we have W = P A x. But A x = V is the change in volume of the system, so −
W = P V. (1.16) − In our case V < 0 so work W > 0 is done on the system, as expected, because we are the ones doing the work!
If the pressure changes during the course of the process, then the above result will no longer be correct. This is entirely possible because the volume is changing and generally this will change the pressure. In this case one needs to know the equation of state P = P (V,T ) and determine
Vf W = P (V,T )dV. (1.17) − ZVi
Let’s consider an ideal gas. Obviously, squeezing the container will pump energy into the system, because the work (1.17) is positive when dV < 0. But let’s suppose that somehow we manage to keep the temperature completely constant. Then the work done is
Vf dV V W = Nk T = Nk T ln f > 0. (1.18) − B V − B V ZVi i Because the temperature is closely tied to the total internal energy of the system, this is equivalent to stating that the total energy must remain constant. In this case, we must have as much heat flowing out of the system as there is work being done on the system, i.e.
V Q = Nk T ln f < 0 (1.19) B V i which is negative because heat is flowing out. This process is called an isotherm.
Suppose now that we don’t let any heat escape at all. Then the internal energy and temperature must both go up, U = W . Using the equipartition theorem (1.13), we have
f U = Nk T = P V. (1.20) 2 B − Because things can vary during the course of the changes, it is better to write this as a differential equation, (f/2)Nk dT = P dV . Using the ideal gas formula again gives B − f dT dV = (1.21) 2 T − V which can be integrated to yield (f/2) ln(T /T )= ln(V /V ) = ln(V /V ). This yields f i − f i i f T f/2 V f = i VT f/2 = constant V γ P = constant, (1.22) T V ⇒ ⇒ i f where γ = (f + 2)/f is the adiabatic exponent, and the process is called an adiabat. PHYS 449 Course Notes 2009 9
A few comments here are in order. Since no heat is flowing in or out of the system in an adiabatic process, then in the absence of any friction we could completely recover the original state of the system by simply reversing the procedure, i.e. by pulling the piston back to where it was before. This will allow the system to do the work, and the total energy and temperature will go back to where they were at the beginning. Though we haven’t talked about entropy yet, it turns out that such a process conserves the entropy, because nothing has changed when get back to where we started. The process is then called isentropic. I mention this because often when people are talking about adiabatic processes, they sometimes actually mean isentropic, and vice versa.
The adiabatic heating of a gas on compression has many practical consequences. When gases are compressed (eventually into liquids) for use in scuba gear or industrial processes, they rapidly heat. So this needs to be performed under cold conditions where the compressing gases can quickly cool, with metal (i.e. thermally conducting) tanks under cold water in the case of scuba gear.
You might be interested to know that this compression is precisely the principle behind diesel engines: when the diesel gas is hot enough it spontaneously explodes. This is why diesel engines don’t need spark plugs and why it is better to have one of those cars if you need to ford a deep river on your way to the mountains! Here’s another application: the Chinooks. When the air descends from the tops of the mountains into the foothills, the air pressure increases because of the lower altitude, which heats it as an adiabatic process. It is largely for this reason that Chinook winds are warm. A similar thing happens in Antarctica with the katabatic winds, which can blow continuously for months. Unfortunately for researchers down there, in that case the air loses its adiabatic heat due to equilibration with the glaciers.
1.6 Heat Capacity
The heat capacity of an object not exactly what it sounds like, i.e. how much heat can it hold. But it’s close: it is how much heat is needed to raise its temperature by one degree. Of course, the properties of the object can change as it heats or cools (like it might freeze, catch fire, etc.), so to the heat capacity needs to be carefully defined. Here’s the way the book does it: the heat capacity is C Q/ T . In books you often hear about the ‘specific heat’ instead, which is the heat capacity per unit≡ mass c C/m. The problem with this definition is that Q = U W , so that the heat capacity depends≡ on how the heat is added because of the work term. To make− things simpler we assume that the applied work is zero. Assuming the work is of the form W = P V then this means the volume is kept constant. So we finally have − U ∂U C = = . (1.23) V T ∂T V V
Most real objects actually prefer to expand when heated, keeping their pressure constant. This effect is pretty small in solids and liquids (see more on this below) but is important in gases. So it is convenient to include the work done by the object during expansion: U ( P V ) ∂U ∂V C = − − = + P . (1.24) P T ∂T ∂T P P P In solid state physics people usually just say ‘heat capacity’ or ‘specific heat’ and you should always assume they are talking about CV . You should convince yourself that for an ideal gas CP = PHYS 449 Course Notes 2009 10
CV + NkB = CV + R if N = NA. Often in circumstances where the system is allowed to do volume changing work it is useful to define the enthalpy:
H U + PV. (1.25) ≡
Then it’s obvious that CP = (∂H/∂T )P . The enthalpy doesn’t really mean anything particularly physical to me; it seems more like a mathematical device to me. Some textbooks say that it can be thought of as the total energy that you need to create something out of nothing, i.e. it includes the energy you need to displace the air to put something there. The enthalpy is hugely popular in chemistry, so because this is a physics class I’ll try to avoid it as much as possible!
Let’s calculate the heat capacity for an ideal gas. Using the equipartition theorem (1.13) we have
∂ f f f C = Nk T = Nk = R, (1.26) V ∂T 2 B 2 B 2 where the last equality follows if N = NA (see Sec. 1.2). In fact, for many people this equation is actually the equipartition theorem, i.e. that every degree of freedom contributes NkB/2 to the heat capacity. The chemists (who have moles on the brain) state that every degree of freedom contributes R/2. With this fact in mind, let’s consider the heat capacity of a gas of hydrogen (recall that hydrogen is diatomic, as are all the HOBFINC gases: hydrogen, oxygen, bromine, fluorine, iodine, nitrogen, and chlorine): It is clear from this figure that the heat capacity is strongly dependent on
Figure 1.2: The heat capacity of hydrogen gas. the temperature. Is there an error in the ideal gas result above? Or perhaps is there something wrong with what we call a degree of freedom? I will justify the shape of this curve properly in Sec. 6.3.
At the risk of getting ahead of ourselves, it turns out that at a phase transition one can add heat to a system without having it increase its temperature at all (phase transitions will be considered quite a bit in Statistical Mechanics II). Think about a pot of water just at the boiling point. You keep the PHYS 449 Course Notes 2009 11
burner on, but it stay at 100◦C. It’s heat capacity is infinite! A similar thing happens when you cool water down to the freezing point: a huge amount of heat leaves the system without the temperature dropping below 0◦C. In fact, a sudden spike in the energy as a function of temperature is used as a signature of a phase transition. In 1908 (almost exactly 100 years ago!), Kamerlingh Onnes was cooling helium and found that it turned into a liquid at 4.2 K. Cooled even further, it started going bananas at Tλ =2.17 K, with a huge release of energy, but below this it stayed a liquid. Only later was it realized that it had turned into a superfluid! These days, the fluid superfluid transition is called the ‘lambda point’ because of the shape of the heat capacity curve:
Figure 1.3: Lambda point of liquid helium, showing the energy released at the superfluid transition.
Most materials have a so called ‘cooling curve’ that shows how the cooling occurs as a function of time. Fig. 1.4 shows a curve for a generic metal as it cools from a liquid to solid. Notice the non linear shape of the curve even far from the phase transition. For the hottest cup of coffee (or tea or hot chocolate as per your preference!), is it better to add the milk right away when you pour it, or just before you drink it? Do you need to use the graph to figure this out?
The heat capacity is not a very useful quantity at phase transitions because of this divergence. But it would still be able to make quantitative statements at this point. Instead, one uses the ‘latent heat of formation,’ or the ‘latent heat’ for short: Q L , (1.27) ≡ m which measures the amount of heat needed (or released) to fully transform the substance from one phase to the other. One assumes that the pressure is constant and that no other work is done either. The latent heat for water to boil is L = 2260 J/g, or 540 cal/g (1 J 0.239 cal). Contrast this with ≈ the 418.4 J (100 cal) needed to bring water from 0◦ to 100◦!
While we’re on the topic of phase transitions, you might be interested in looking at the generic phase diagram of many materials. What important biological consequences are there of water’s ‘anomalous’ behaviour? PHYS 449 Course Notes 2009 12
Figure 1.4: Cooling curve for a generic metal as it goes from liquid to solid.
Figure 1.5: Generic phase diagram. The dotted line corresponds to the ‘anomalous’ behaviour of water. Chapter 2
The Second Law of Thermodynamics (aka The Microcanonical Ensemble)
2.1 Two-State Systems (aka Flipping Coins)
D Classical Probability: All events are equally likely.
D Statistical Probability: Probability that an event occurs is the measured relative frequency of occurrence. Q∗ Is this definition circular?
D Trials: Experiments or tests.
D Events: Results of above, designated ei.
The collection of events forms an abstract space called event space. Each point i in space is assigned a probability pi = p(ei) which is the probability of event i, so that
pi =1. i X
The classical probability of event i is p =1/ , i when there are points in event space. For a i ∀ single coin, c =2; for a die, d = 6. 4 4 Suppose we toss our coin four times. How many events are there? = c =2 = 16. Each event is listed in the table on the next page. The probability of obtaining the result HHHH is just as likely as obtaining HTTH or HTHT. Does this seem right? ∗The Q stands for a question.
13 PHYS 449 Course Notes 2009 14
Combination (#H, #T) HHHH (4,0) HHHT (3,1) HHTH (3,1) HHTT (2,2) HTHH (3,1) HTHT (2,2) HTTH (2,2) HTTT (1,3) THHH (3,1) THHT (2,2) THTH (2,2) THTT (1,3) TTHH (2,2) TTHT (1,3) TTTH (1,3) TTTT (0,4)
Table 2.1: All the possible outcomes of tossing a coin four times.
Because the result of each coin flip is completely independent of the results of any previous coin 1 flips, the probability of obtaining H or T is always 2 every time. So the probability of getting any of the 4 coin flip events is p = 1 1 1 1 = 1 . i 2 × 2 × 2 × 2 16 D Microstate: Each compound event. For this example, there are 16 microstates. All microstates are formed from single events that are themselves equally likely.
Theorem: For a system in equilibrium, all microstates are equally possible.
Ergodic Hypothesis (Boltzmann 1871, Ehrenfest 1911): Given a sufficiently long time, all mi crostates will be observed and all points in event space will be accessed. THIS IS THE UNPROVEN FOUNDATION OF STATISTICAL MECHANICS!
Now, how many microstates have 4 heads and no tails, or (4, 0)? Or (3, 1), (2, 2), (1, 3), or (0, 4)? If we don’t care what order the faces come up, then we have #(4, 0) = #(0, 4) = 1; #(3, 1) = #(1, 3) = 4; #(2, 2) = 6. Recognize the pattern?
D Macrostate: Collection of microstates with some common property. In the coin case above, the macrostates correspond to all events that share the same number of heads and tails.
Theorem: The state of the system (the macrostate that is actually observed) is the macrostate with the largest number of microstates. PHYS 449 Course Notes 2009 15
2.1.1 Lots and lots of trials
How to assign values to the macrostate when the number of trials (N) gets really huge? Suppose that we have a weighted coin, i.e. where the probability that it comes up heads is p(H) p and probability of seeing tails is p(T ) q =1 p so that p + q = 1. Then the microstate probability≡ ≡ − Pmicro is n1 n2 n1 N n1 P = P (n )= p q = p (1 p) − micro micro 1 − where n1 and n2 are the number of p and q events, respectively, for a given trial. The total number of coin tosses is N = n1 + n2.
n1 N n1 Then the macrostate probability Pmacro(n1)= (n1)p q − , where (n1) is the number of ways of arranging n events with p and N n events with q. We also know that 1 − 1 n1 N n1 Pmacro(n1)= (n1)p q − =1. (2.1) n1 n1 X X So now let’s use the binomial theorem:
N N N! n N n (p + q) =1= p q − , (2.2) n!(N n)! n=0 X − Comparison of Eqs. (2.1) and (2.2) shows that the number of ways of distributing microstates in a macrostate is given by: N! N (N,n)= . (2.3) n!(N n)! ≡ n − These are the binomial coefficients, which is why those numbers (1,4,6,4,1) appeared when count ing microstates in the 4 coin macrostates above. By the way, the right hand side of the above equation is read ‘N choose n.’ Note that if p = q =1/2 then
N N! =2N . (2.4) n!(N n)! n=0 X − The binomial coefficients for increasing values of N are shown in Fig. 2.1.
What is the intuitive reason for this expression for ? Suppose we have 4 objects, labeled A, B, C, and D. How many ways can these be arranged? Let’s see: ABCD, ABDC, ACBD, ACDB, ADBC, ADCB, BACD,BADC, BCAD, BCDA, BDAC, BDCA, CABD, CADB, CBAD, CBDA, CDAB, CDBA, DABC, DACB, DBAC, DBCA, DCAB, DCBA. So there’s 4! = 24 ways. In general, N objects can be arranged N! ways. Why? Well, for the example above there’s 4 ways of arranging the first letter, 3 ways of arranging the second, 2 ways for the third, and only one way for the last: i.e., 4 3 2 1 = 4!. × × × But A, B, C, and D might not be completely distinguishable! Suppose all of these letters actually represented heads (H). Then obviously there would be only one way of arranging them. If n of the letters stand for heads, then one would need to divide by n!, which is just the number of ways of arranging the previously distinguishable letters, which are now all heads. But we also need to divide PHYS 449 Course Notes 2009 16
Figure 2.1: Normalized binomial coefficients for increasing N. These are the binomial coefficients divided by the total, 2N . by (N n)!, which is the number of ways of arranging the previously distinguishable letters that now stand− for tails. And so the total number of arrangements is N!/n!/(N n)!. − From the above discussion, it should now be obvious how to extend this analysis to ‘coins’ of more than two sides (whatever they would look like!). A three sided coin would have
N! (N,n ,n )= , 1 2 n !n !(N n n )! 1 2 − 1 − 2 while an m sided coin would have N! N! (N,n1,n2,...,nm 1)= m = . (2.5) − m 1 m 1 i=1 ni! − n ! N − n ! i=1 i − i=1 i Q This equation is going to be important later in the term.Q P
I’ll cover some examples of calculating probabilities and ’s in class.
2.1.2 Digression: Statistics
Suppose that there is some variable u that have values u1,u2,...,um appearing with probabilities p(u ),p(u ),...,p(u ) = p ,p ,...,p . Then { } { 1 2 m } { 1 2 m} PHYS 449 Course Notes 2009 17
D Mean Value (Average): of u is designated u and is defined as
m m m u p(u )+ u p(u )+ + u p(u ) uip(ui) uipi u 1 1 2 2 m m = i=1 = i=1 = u p (2.6) ≡ p(u )+ p(u )+ + p(u ) m p(u ) m p i i 1 2 m i=1 i i=1 i i=1 P P X m P P because i=1 pi = 1 by the definition of probability.
P m m In general, f(u)= i pif(ui)= i pifi. P P D Standard Deviation about the mean: of u is designated ( u)2 and is defined as
m ( u)2 p (u u)2 (u u)2 ≡ i i − ≡ − i=1 X = u2 2uu + u2 = u2 u2. (2.7) − − Note that u2 is just u2. The standard deviation is also called the fluctuations about the mean, the second moment of u about the mean, and the dispersion of u.
D Root-Mean Square (RMS) value: This is just what it sounds like. It is the square root of the mean of the square of the function u: RMS(u) u2. ≡ p 2.2 Flow toward equilibrium
The discussion of flipping coins in the previous section must seem a bit out of place after all the discussion of gases in Chapter 1. In fact, it relates closely with the behaviour of a one dimensional gas (which might also seem a bit artificial!), as will be seen in the following section. A simple extension relates it to a real 3D gas. But before we get to that, it turns out that the simple model of a two sided coin corresponds closely to something called the Pauli paramagnet. This will be studied completely later on in Sec. 3.3, but it is useful to introduce it here, to illustrate how equilibrium works.
The Pauli paramagnet is a collection of tiny atomic magnetic spins that can either point up or down. One could imagine iron atoms this way. If you associate with heads and with tails, for example, then the correspondence with flipping coins is explicit. If↑ you apply a magnetic↓ field, all the spins will want to align themselves with the field, for example all pointing up. If you do this at high temperature and then cool the metal down, the spins will be frozen in the up position, and you get a bar magnet, something where all the little spins add up to a macroscopic magnet. For many metals the spin directions aren’t frozen even at low temperature, and quickly become disordered, which is why permanent magnets tend to be iron. If you heat up the metal, though, then the spins will again become disordered and the total magnetization will go to zero, that is on average there will be as many as atoms. ↑ ↓ With our statistical mechanics hats on, we should ask: why is the configuration with equal numbers of as atoms the most likely? Consider again Fig. 2.1, which shows the binomial coefficients ↑ ↓ PHYS 449 Course Notes 2009 18