Coupling of Angular Momenta and Total J

Angular momentum coupling is not a historical curiosity confined to term-symbol bookkeeping — it is the operating grammar behind entire branches of modern spectroscopy and technology. In atomic and molecular spectroscopy, term symbols $^{2S+1}L_J$ generated by LS coupling remain the standard language for cataloging electronic states of atoms and small molecules, and the selection rules on $\Delta J = 0,\pm1$ for electric dipole transitions follow directly from the coupled-basis structure and the Wigner-Eckart theorem. Fine structure, which splits terms like $^3P$ of carbon or the sodium D-lines ($^2P_{1/2}$ and $^2P_{3/2}$) into distinct $J$ levels, is the direct successor topic — Spin-Orbit Coupling and Fine Structure is precisely the physical mechanism (relativistic coupling of orbital and spin magnetic moments) that makes the coupled $J$ basis the energetically preferred one, splitting what coupling theory shows to be otherwise degenerate levels. In nuclear magnetic resonance and EPR spectroscopy, the coupling of nuclear spins to each other (scalar $J$-coupling in NMR, hyperfine coupling in EPR) produces the characteristic multiplet splitting patterns — doublets, triplets, quartets — used every day to determine molecular structure; the intensity ratios within these multiplets are Clebsch-Gordan-derived binomial coefficients. In lanthanide and actinide chemistry, where spin-orbit coupling is comparable to or exceeds inter-electron repulsion, the same coupling algebra (now in $jj$ or intermediate coupling schemes) determines the magnetic and optical properties exploited in MRI contrast agents (gadolinium complexes), lanthanide-based luminescent probes, and single-molecule magnets. In quantum information science, coupling of two qubits' angular-momentum-like degrees of freedom (e.g., coupling nuclear and electron spins in nitrogen-vacancy centers in diamond) uses precisely this formalism to design controlled quantum gates. And in nuclear and particle physics, the same triangle rule governs the coupling of nucleon spins and isospins, angular momentum conservation in nuclear reactions, and the classification of hadron multiplets in the quark model — all descendants of the identical algebraic structure first worked out for two orbiting electrons.

Ask the tutor: How does adding spin-orbit coupling as a perturbation actually split the $^3P$ term of carbon into its $J=0,1,2$ fine-structure levels, and why is $J=0$ lowest in energy?

Classically, adding two angular momentum vectors is trivial: you add the vectors component by component, and the magnitude of the sum can be anything between $|j_1 - j_2|$ and $j_1+j_2$ depending on the relative orientation. Quantum mechanically the situation is both similar and strikingly different. The allowed magnitudes of the total do fall in that same classical range — but instead of a continuum of possible resultant lengths, only a discrete ladder of values is allowed, each differing from its neighbor by exactly one unit of $\hbar$. The reason is representation theory in disguise. Each individual angular momentum $\mathbf{J}_1$ and $\mathbf{J}_2$ furnishes an irreducible representation of the rotation group, of dimension $2j_1+1$ and $2j_2+1$ respectively. The combined system's Hilbert space, of dimension $(2j_1+1)(2j_2+1)$, carries a reducible representation of the rotation group (rotations act on both particles simultaneously). Decomposing that reducible representation into irreducibles — each irreducible piece being a definite total angular momentum $J$ — is exactly the Clebsch-Gordan decomposition. This is the same logic that appears whenever Ladder Operators and the Angular Momentum Algebra are used to build the $|j,m\rangle$ ladder for a single angular momentum; coupling simply asks how two such ladders combine. Physically, the uncoupled states $|j_1 m_1\rangle|j_2 m_2\rangle$ are eigenstates of $J_{1z}$ and $J_{2z}$ separately — they describe a situation where each angular momentum vector has a definite projection on the $z$-axis, but the relative orientation of $\mathbf{J}_1$ and $\mathbf{J}_2$ is completely undetermined (equivalently, $\mathbf{J}_1\cdot\mathbf{J}_2$ has no definite value). The coupled states $|JM\rangle$ instead have a definite total angular momentum magnitude $J(J+1)\hbar^2$ and a definite $z$-projection $M\hbar$ of the sum, but the individual $z$-projections $m_1, m_2$ are now indefinite — the two vectors are precessing around their fixed resultant. Neither description is more "real" than the other; they are two orthonormal bases for the identical physical Hilbert space, and the correct choice is whichever one diagonalizes the actual Hamiltonian governing the system, exactly as prefigured by the Spin Angular Momentum and the Stern-Gerlach Experiment discussion of measurement bases.

The dimension count in that figure is not a coincidence — it is the master consistency check for any coupling problem, however large $j_1$ and $j_2$ are. Every valid Clebsch-Gordan decomposition must satisfy

where:

• $2j_1+1$, $2j_2+1$ are the dimensions of the individual irreducible spaces,
• the sum runs over the triangle-rule ladder of allowed total $J$ values in integer steps,
• $2J+1$ is the dimension (number of $M$ states) of each total-$J$ multiplet.

Reading: the left side counts uncoupled basis states $|m_1\rangle|m_2\rangle$; the right side counts coupled basis states $|JM\rangle$ grouped into multiplets. Since both bases span the same space, the totals must match — and they always do, because the sum of consecutive odd (or the appropriate arithmetic) integers telescopes exactly. A second essential picture is how the maximum $M$ value pins down where the ladder starts. Since $m_1$ ranges over $2j_1+1$ values and $m_2$ over $2j_2+1$ values, the largest possible sum is $M_{max} = j_1+j_2$, achieved in only one way: $m_1 = j_1, m_2 = j_2$. That unique state must belong to a multiplet with $J = j_1+j_2$ (since no larger $J$ can exist — nothing produces a bigger $M$). This is the top rung of the tallest ladder. Descending one unit at a time and comparing degeneracies at each $M$ level to how many are already "used up" by higher $J$ multiplets, reveals successively that $J = j_1+j_2-1, j_1+j_2-2,\ldots$ must each appear exactly once, terminating at $J=|j_1-j_2|$ — precisely the triangle rule.

The starting point is the definition of the total angular momentum operator as a vector sum acting on the tensor product Hilbert space:

where:

• $\mathbf{J}_1$ acts only on the first particle's/subsystem's degrees of freedom, $\mathbf{J}_2$ only on the second's,
• the tensor product structure guarantees $[\mathbf{J}_1, \mathbf{J}_2] = 0$ (they act on different factors),
• $\mathbf{J}$ is then shown to satisfy the standard angular momentum commutation relations $[J_x,J_y]=i\hbar J_z$ (cyclic) purely as a consequence of $\mathbf{J}_1$ and $\mathbf{J}_2$ separately satisfying them.

Reading: because $J_1$ and $J_2$ commute with each other (different subsystems), their sum automatically obeys the same algebra as either one alone — this is why $J^2$ and $J_z$ can be simultaneously diagonalized just like for a single angular momentum, giving a legitimate $|JM\rangle$ ladder. The triangle rule (also called the Clebsch-Gordan series) states the allowed total quantum numbers:

where:

• $j_1, j_2$ are the (fixed) individual angular momentum quantum numbers, integer or half-integer,
• $J$ steps down by exactly 1 from the maximum to the minimum — never skipping,
• each allowed $J$ value occurs exactly once in the decomposition (multiplicity 1), not more.

Reading: this is the master selection rule for what totals are even possible — before asking about probabilities or coefficients, this tells you the complete list of total-$J$ multiplets present. The four commuting operators of the coupled scheme, and their simultaneous eigenstates, are:

where:

• the first pair fixes the individual magnitudes $j_1, j_2$ (these don't change under coupling — $J_1^2$ commutes with everything since $[\mathbf J_1^2,\mathbf J_2]=0$),
• $J^2 = J_1^2+J_2^2+2\mathbf J_1\cdot\mathbf J_2$ is the total-squared-magnitude operator,
• $M = m_1+m_2$ always, since $J_z = J_{1z}+J_{2z}$ commutes with everything in both bases.

Reading: $\{J_1^2,J_2^2,J^2,J_z\}$ is a complete set of commuting observables — an equally valid alternative to $\{J_1^2,J_2^2,J_{1z},J_{2z}\}$ — and the coupled kets $|j_1j_2;JM\rangle$ are simultaneous eigenkets of the first set. The connection between the two complete bases is the Clebsch-Gordan expansion:

where:

• $\langle j_1 m_1\, j_2 m_2|JM\rangle$ is the Clebsch-Gordan coefficient — a real number (by convention) fixed by the algebra, tabulated in standard references,
• the sum is constrained by $m_1+m_2 = M$ (all other terms vanish identically),
• the coefficients form a real orthogonal matrix, so the inverse transformation (uncoupled in terms of coupled) uses the same coefficients.

Reading: this equation is the precise statement of "coupled states are superpositions of uncoupled states with definite $m_1+m_2=M$" — the coefficients are what carries all the quantitative information about how the two angular momenta correlate in a state of definite total $J,M$.

Example 1 — Foundational: coupling two spin-1/2 electrons. Two electrons each have $j_1=j_2=\tfrac12$. The triangle rule gives

so total spin $S=1$ (triplet, 3 states) or $S=0$ (singlet, 1 state), totaling $3+1=4=2\times2$ states — matching $(2\cdot\tfrac12+1)^2=4$. Explicitly, using the ladder-operator method above (or simply pattern-matching the $j_1=j_2=1/2$ Clebsch-Gordan table):

As a limiting-case check: setting one spin to "off" conceptually (i.e., taking $j_2\to0$ formally) would collapse the coupling to $J=j_1$ only, a single multiplet with the uncoupled and coupled bases coinciding — consistent with the general rule that coupling to $j_2=0$ never splits anything, since the triangle rule with $j_2=0$ forces $J=j_1$ exclusively. This singlet/triplet decomposition is precisely what underlies the Pauli-allowed spatial symmetries in the helium atom and in covalent bond formation between two electrons.

Example 2 — Applied: coupling $L$ and $S$ for a $^3P$ carbon-like term. For the ground configuration of carbon, $2p^2$, Hund's rules and the Pauli principle restrict the allowed terms to $^1S, \,^1D,\,^3P$ (a standard exercise in term-symbol construction). Focus on the $^3P$ term: here $L=1$ (P) and $S=1$ (triplet, spin multiplicity $2S+1=3$). Coupling $L$ and $S$ via the same triangle rule (now applied to orbital and spin angular momentum rather than two orbital or two spin momenta) gives total $J$:

So the $^3P$ term splits into three fine-structure levels: $^3P_0, ^3P_1, ^3P_2$, with degeneracies $2J+1 = 1,3,5$. As a numerical check on total degeneracy: the full $^3P$ term (before spin-orbit splitting) has degeneracy $(2L+1)(2S+1)=3\times3=9$, and indeed $1+3+5=9$ — the $L,S$ multiplet redistributes exactly, level by level, into the coupled $J$ basis, with no states lost or gained. (In reality, carbon's $^3P_0$ ground level lies about $16.4\ \text{cm}^{-1}$ below $^3P_1$ and $43.4\ \text{cm}^{-1}$ below $^3P_2$, an ordering and spacing set by the sign and magnitude of the spin-orbit constant — the subject that this coupling machinery feeds directly into.)

Example 3 — Exam/competition level: three coupled angular momenta and a nontrivial branching count. Consider coupling three angular momenta with $j_1 = \tfrac32,\ j_2=1,\ j_3=\tfrac12$ (a stylized model, e.g., relevant to hyperfine coupling of two electron-like momenta plus a nuclear spin). Coupling is done pairwise: first combine $j_1$ and $j_2$ into an intermediate $j_{12}$, then couple $j_{12}$ with $j_3$ to get the final $J$. Step A. $j_1=\tfrac32, j_2=1$: triangle rule gives $j_{12} = \tfrac32+1,\ldots,|\tfrac32-1| = \tfrac52,\,\tfrac32,\,\tfrac12$. Step B. For each $j_{12}$, couple with $j_3=\tfrac12$:

• $j_{12}=\tfrac52$: $J = 3,\,2$
• $j_{12}=\tfrac32$: $J = 2,\,1$
• $j_{12}=\tfrac12$: $J = 1,\,0$

So the final list of total $J$ values (with multiplicities, since $J=2$ and $J=1$ each arise from two different $j_{12}$ "routes") is:

Dimension check. Total uncoupled dimension: $(2\cdot\tfrac32+1)(2\cdot1+1)(2\cdot\tfrac12+1) = 4\times3\times2=24$. Coupled dimension:

which matches exactly, confirming the branching diagram is complete and correctly enumerated — and illustrating the important general point that when $J$ is reachable by more than one intermediate-coupling route, its overall multiplicity in the final decomposition is genuinely greater than one, unlike the two-body case where every $J$ appears exactly once.

The mathematics of angular momentum coupling was worked out by physicists rather than chemists, but its constant use in atomic spectroscopy makes it as much a tool of the physical chemist as of the physicist. The two names permanently attached to the coefficients are Alfred Clebsch and Paul Gordan, nineteenth-century mathematicians who developed the underlying theory of invariants of binary forms — group representation theory did not yet exist as such, and it was Eugene Wigner decades later who recognized that their algebraic formulas were exactly the coefficients needed to decompose products of rotation-group representations, cementing the "Clebsch-Gordan coefficient" as the standard tool of atomic and nuclear spectroscopy.