What is spin?

This is the first in a series of posts explaining fundamental physics concepts in simple terms. I will try to explain as deeply as possible from first principles, but without assuming any math beyond high-school level. However, the footnotes contain details for more advanced readers.


The first topic is spin. Spin is a measure of the internal rotational degrees of freedom of a particle. Consider a particle at rest at the origin. We will assume the particle has nonzero mass for now and discuss the massless case later. What transformations can we make to it that leave it looking externally the same? There are just the three rotations, one around each axis1.

Spin 0

Now consider describing the particle with a sequence of numbers (degrees of freedom, or DOFs) that change in a defined way under rotations. The simplest case is just to give it a single number that doesn’t change under rotations. This is spin 0. The Higgs boson is the only known fundamental spin 0 particle.
 

What if we try to make this single number change? Let’s say a 180° rotation around the x, y, or z axis multiplies it by 2. This doesn’t work, because a 180° rotation around x followed by a 180° rotation around y is the same as a 180° rotation around z, which you can check. But the former results in a factor of 4, while the latter gives a factor of 2. So not every possible choice of transformation works: it must be compatible with the behavior of rotations.

The little group

Finding all the possible ways that a set of numbers can transform under some symmetries (called a group) is known as representation theory. In a landmark paper, Wigner first classified particles as representations of the Poincaré group, the group of symmetries of special relativity. In addition to rotations, this group includes boosts and translations in space and time. He showed that internal DOFs are described by the little group, the group that leaves the particle externally the same. In this case, it is the group of rotations in three dimensions, called SO(3). Spin 0 is a 1-dimensional representation of the little group, since it is just a single number.

Spin 1

Anyway, back to our particle. Another obvious choice is to describe the particle with a 3D vector. Under 3D rotations, this just rotates in the usual way. This is called spin 1. It is a 3-dimensional representation of the little group. Spin-1 particles include the W and Z vector bosons, which mediate the weak force2.

Spin 1/2

We have been working with representations where all the numbers are real. For example, a real 3D vector stays real under rotations, since rotation matrices are real. But quantum mechanics says the universe uses complex numbers. It turns out that there is a complex representation in between spin 0 and spin 1, called spin 1/2. It has two complex DOFs. The majority of particles in our universe are spin-1/2: electrons, muons, quarks, etc.

Higher spins

We can continue upwards, constructing larger and larger representations with spin > 1. In general, a spin-s representation has 2s + 1 DOFs. There is a representation for every integer and half-integer s \geq 0. Integer spin particles are called bosons, and half-integer particles are called fermions.
 
Composite particles form so-called product representations that decompose into independent spin representations. For example, two spin-1/2 particles have 4 DOFs. These split into a spin-1 representation (the “triplet”) and a spin-0 (the “singlet”). In group theory notation this is sometimes written as
 
2\times 2 = 3 + 1
 
(Who said group theory was hard?) This means that under rotations, the spin-1 DOFs transform as a 3D vector while the spin-0 part doesn’t change. Specifically, there is a linear combination of these DOFs that transform as spin-1 and spin-0. An arbitrary linear combination of DOFs will all mix into each other under rotations.
 
While composite particles can have high spin, no fundamental massive particles with spin \geq 3/2 are known to exist.

Spin and statistics

Fermion representations have the peculiar property3 that a full rotation around 360° multiplies the state by -1 instead of 1. In fact, this implies that identical fermions cannot occupy the same state, known as the Pauli exclusion principle. This is responsible for the diverse matter in our universe such as atoms and molecules. Otherwise, fermions in a system would all collapse near the state of lowest energy, the ground state. On the other hand, identical bosons acquire a +1 under rotation, so can occupy the same state. At low temperatures, they almost all occupy the ground state, forming a Bose-Einstein condensate.
 
This connection between the behavior of large numbers of particles and their spin is also called the spin-statistics theorem. Proving this theorem requires quantum field theory and relativity, which are beyond the scope of this article.

Spin and angular momentum

DOFs associated with rotations are known as angular momentum. We have only discussed the internal DOFs (spin), but particles also carry external rotational DOFs called orbital angular momentum. The total angular momentum is the sum of spin and orbital contributions. Since spin representations are finite-dimensional, spin angular momentum is quantized4. A measurement in a particular direction will give a result from \{-s\hbar, (-s+1)\hbar, \cdots, (s-1)\hbar, s\hbar\} for a particle of spin s, where \hbar is Planck’s constant. You can see there are 2s+1 different values. Because rotations around different axes don’t commute, angular momentum in different directions cannot be measured simultaneously: once angular momentum in one direction is known exactly, the other directions become uncertain.

For spin-1/2, the two values are -\hbar/2 and \hbar/2, corresponding to states called “spin down” and “spin up” with respect to a particular direction. These states can be visualized as little arrows in the direction of angular momentum:

Screen Shot 2020-05-19 at 10.12.35 PM

Spin and magnetism

When electromagnetism is included into the theory, it turns out that spin couples to the magnetic field5. Classically, you can visualize spin angular momentum as arising from a particle literally spinning around:Screen Shot 2020-05-28 at 5.00.14 PM

The potential energy of a current loop in a magnetic field is6:

U = -IA\vec{B}\cdot\hat{n}

where I is the current, A is the area, and \hat{n} is the normal vector in the direction given by the right-hand rule on the current. Using A=\pi r^2, I = qv/2\pi r, and angular momentum \vec{L} = rmv\hat{n}, this becomes

U = -\frac{q}{2m} \vec{B}\cdot\vec{L}

Thus, spins tend to “align” with an external magnetic field, since the states with spin in the same direction as the field have lower energy than the states in the opposite direction (for a positive charge q). q/2m is known as the gyromagnetic ratio. It turns out in the quantum theory that this ratio is actually twice the value of the classical theory: q/m. So a particle with spin cannot really be thought of as a classical current loop.

Magnetic fields allow us to manipulate and measure spin, as in the Stern-Gerlach experiment.

Massless particles and helicity

Massless particles such as the photon do not have spin, because their little group is different. Special relativity tells us that massless particles must travel at the speed of light. Therefore, we cannot imagine them “at rest”: they are always going in a particular direction. However, we can still make rotations around this direction and leave the particle the same. The little group in this case is SO(2), the rotation group in 2 dimensions. This is a very simple group that only has 1-dimensional representations7. In fact, we can find them all here. A helicity-s representation acts on the number by multiplying it by

R(\theta) = e^{i s\theta}

under a rotation by angle \theta. There is a representation for every integer and half-integer s, where s can be less than 0 now. Note that half-integer representations are fermions again, since a rotation by 2\pi gives -1.

It would then seem that all massless particles have only one degree of freedom. In fact, another symmetry principle requires us to stick two of these representations together: parity, or symmetry under spatial reflections (x\rightarrow -x, y\rightarrow -y, z\rightarrow -z). Under parity, a rotation in one direction around the particle’s velocity goes in the opposite direction:

Screen Shot 2020-05-29 at 9.58.49 PM

We have reflected both the black rotation arrow and the velocity vector v. Particle representations must be invariant under parity. If we include the +s representation, we also need to include the representation that transforms in the opposite way under rotation. Therefore, massless particles have two DOFs: both the +s and -s representations. Each degree of freedom transforms independently under this rotation (see footnote 7). Note that massive particles have no preferred direction when at rest, so representations are automatically invariant under parity. Imagine removing the blue arrows in the above figure; you will see that a rotation reflected is the same rotation.

The photon has helicity 1, and the graviton (the force carrier of gravity) has helicity 2. One fascinating result of quantum field theory is that it is impossible to have a locally interacting theory of massless particles greater than helicity 2.

Further resources

So much for this whirlwind tour of spin and related topics. For more info, see any standard textbooks on quantum mechanics or quantum field theory. Some I recommend:

  • Griffiths, D.J. Introduction to Quantum Mechanics, Ch. 4.
  • Schwartz, Matthew. Quantum Field Theory and the Standard Model, Chs. 8, 10, 12.
  • Weinberg, Steven. The Quantum Theory of Fields, Ch. 2.4, 2.5.

1 In special relativity, there are also transformations called “boosts”, but these give the particle a constant velocity, so it is no longer at rest.

2 You may have heard that the photon is spin 1. But the photon is massless, so has helicity instead of spin. More on this later.

3 This may seem impossible, since a rotation by 360° is the same as no rotation at all. And you are right! I slightly lied earlier. Technically, we are finding representations of SU(2), the double cover of the rotation group SO(3). This is because the Lie algebras are identical, su(2) \sim so(3), but exponentiating the spinor representation of the algebra produces SU(2) instead of SO(3).

4 Orbital angular momentum is also quantized, but only integer representations exist. This is because the DOFs here are actually fields: functions of space. A rotation by 360° must take the field to the same field, because the position vector itself is a vector representation of rotations.

5 This can be derived by starting with the Lagrangian of quantum electrodynamics and taking the non-relativistic limit, where all energies are smaller than the rest energy of the electron E=mc^2. However, in the spirit of effective field theory, we can also consider writing down all terms consistent with the non-relativistic symmetries: SO(3), parity, and gauge invariance. The spin operator \vec{S} is a vector, so it must be dotted with another vector to create an SO(3) invariant. Actually, it is a pseudovector: \vec{L}=\vec{r}\times \vec{p} is invariant under parity \vec{r}\rightarrow -\vec{r}, \vec{p}\rightarrow -\vec{p}. It must be dotted with another pseudovector to be invariant under parity. This must be the magnetic field \vec{B}, since the electric field \vec{E} is a vector. It cannot be any other function of the vector potential A_\mu due to gauge invariance. Thus, the lowest order interaction is

c \vec{S}\cdot \vec{B}

for some constant c.

6 A nice derivation is as follows. Start with the integral form of Faraday’s law:

\frac{d\phi}{dt}=-\oint \vec{E}\cdot \vec{dl}

where \phi is the magnetic flux through the loop. Multiply by current and integrate over time:

I\Delta \phi=-\int dt IV=-\int dt P_{diss}

where V=\oint \vec{E}\cdot \vec{dl}, I is the current, and P_{diss}=IV is the dissipated power. Thus, it takes an energy

E=\int dt P_{diss}=-I\Delta \phi

to change the magnetic flux by an amount \Delta \phi. It is easiest to draw a picture to get the sign right. This derivation shows that the energy is independent of the shape of the loop.

7 A rotation matrix in 2D is, of course, two-dimensional. However, this is actually a reducible representation made of two irreducible, one-dimensional (complex) representations: helicity +1 and -1. You can see this by noting that (1, i) and (1,-i) are both eigenvectors under rotation, so do not transform into each other. We are only classifying the irreducible representations here.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s