Ladder Operators and the Angular Momentum Algebra

The ladder-operator algebra is not a historical curiosity — it is the computational backbone of every modern treatment of angular momentum in chemistry. In atomic and molecular spectroscopy, term symbols and selection rules for electronic transitions are worked out using exactly the raising/lowering formulas derived here, applied to coupled orbital and spin angular momenta to build total $J$ states. In magnetic resonance — both NMR and EPR/ESR — the same $\hat{L}_\pm$-style operators (there called $\hat{I}_\pm$ for nuclear spin or $\hat{S}_\pm$ for electron spin) generate the transition operators that appear directly in the spin Hamiltonians used to predict coupling patterns, relaxation pathways, and pulse-sequence dynamics; the raising and lowering operators are literally what create observable coherences in a two-dimensional NMR experiment. In transition-metal and lanthanide chemistry, crystal-field and ligand-field theory diagonalize $d$- and $f$-orbital manifolds using operator-equivalent techniques built from exactly this $\hat{L}_+,\hat{L}_-,\hat{L}_z$ algebra, predicting magnetic anisotropy and single-molecule-magnet behavior in materials used for high-density data storage research. In quantum computing with spin qubits and in electron paramagnetic resonance imaging of radicals and metal centers in biological systems, the same ladder formalism describes how microwave or radiofrequency pulses coherently move population between $m$-sublevels. Even in computational quantum chemistry software, angular momentum coupling coefficients (Clebsch–Gordan and Wigner 3j/6j symbols used to combine atomic orbitals into symmetry-adapted linear combinations) are built from the identical raising and lowering machinery introduced here, making this one of the most reused algebraic tools in the entire discipline.

:::quiz[What is [Lz, L₊] equal to?]

• 0
• −ħL₊
• iħL₊
• +ħL₊ ✓

:::quiz[Using L±|l,m⟩ = ħ√[l(l+1) − m(m±1)]|l,m±1⟩, what is the coefficient for L₋ acting on |3,2⟩?]

• ħ√(12−6) = ħ√6 ✓
• ħ√(12−4) = ħ√8
• ħ√(9−4) = ħ√5
• ħ√(6−6) = 0

:::faq[How is the exact normalization constant ħ√[l(l+1) − m(m±1)] actually derived, and why isn't it just ħ√(l(l+1)−m²)?] It comes from computing the squared norm $\langle l,m|\hat{L}_\mp\hat{L}_\pm|l,m\rangle$ using the operator identities $\hat{L}_+\hat{L}_-=\hat{L}^2-\hat{L}_z^2+\hbar\hat{L}_z$ and $\hat{L}_-\hat{L}_+=\hat{L}^2-\hat{L}_z^2-\hbar\hat{L}_z$. Evaluating these on an eigenstate replaces $\hat{L}^2\to l(l+1)\hbar^2$ and $\hat{L}_z\to m\hbar$, producing the extra linear-in-$m$ term $\mp\hbar\hat{L}_z\to\mp m\hbar^2$ alongside $-m^2\hbar^2$. Combining them gives exactly $l(l+1)-m(m\pm1)$, not the simpler-looking $l(l+1)-m^2$; dropping the cross term breaks the exact zero at $m=\pm l$ and is a very common algebra slip.

:::faq[Why does [L+, L−] equal 2ħLz rather than zero?] Because $\hat{L}_+$ and $\hat{L}_-$ are built from $\hat{L}_x$ and $\hat{L}_y$, which themselves do not commute. Expanding directly, $\hat{L}_+\hat{L}_- - \hat{L}_-\hat{L}_+ = (\hat{L}_x+i\hat{L}_y)(\hat{L}_x-i\hat{L}_y) - (\hat{L}_x-i\hat{L}_y)(\hat{L}_x+i\hat{L}_y)$, and multiplying out both products, all the $\hat{L}_x^2$ and $\hat{L}_y^2$ terms cancel, leaving only cross terms proportional to $[\hat{L}_x,\hat{L}_y]=i\hbar\hat{L}_z$, doubled by the two ways the cross terms appear, giving exactly $2\hbar\hat{L}_z$. This nonzero commutator is what ultimately supplies the $m$-dependent linear term in the normalization constant.

Ask the tutor: How would the ladder-operator method change if I wanted to derive the raising/lowering rules for two coupled angular momenta, like electron spin coupled to orbital angular momentum in an atomic term symbol?

Angular momentum eigenstates are simultaneous eigenfunctions of $\hat{L}^2$ and $\hat{L}_z$, written $|l,m\rangle$, satisfying

For a fixed $l$ there are $2l+1$ allowed values of $m$. Rather than solving the associated Legendre equation afresh for each $m$, chemists exploit an algebraic shortcut: define two new non-Hermitian combinations of $\hat{L}_x$ and $\hat{L}_y$,

These are called raising and lowering (or shift, or step) operators. They are adjoints of one another, $\hat{L}_+^\dagger=\hat{L}_-$, which is why neither is separately Hermitian nor separately an observable — but every physical quantity built from them (such as $\hat{L}_+\hat{L}_-$) is Hermitian and measurable. Physically, $\hat{L}_+$ and $\hat{L}_-$ do something no rotation of a classical vector could do cleanly: they take one stationary quantum state and return another stationary state, differing only in the $z$-projection quantum number $m$, while never changing the total angular momentum magnitude quantum number $l$. This is the operator-language version of a $p$-orbital's three lobes ($m=-1,0,+1$) or a $d$-orbital's five lobes ($m=-2,\dots,+2$) being connected members of one family, not independent problems. The diagram below shows the ladder structure for $l=2$ (a $d$-manifold): five rungs at $m=-2,-1,0,1,2$, connected by $\hat{L}_+$ moving up and $\hat{L}_-$ moving down, with the ladder capped at both ends.

The stopping rule shown at the bottom — $\hat{L}_+$ annihilates the top rung, $\hat{L}_-$ annihilates the bottom rung — is not an assumption; it falls directly out of the requirement that the norm $\langle l,m|\hat{L}_-\hat{L}_+|l,m\rangle$ can never be negative, since $\hat{L}_-\hat{L}_+$ is a positive semi-definite Hermitian operator. This single positivity argument is what pins down the allowed range of $m$ and, ultimately, the eigenvalue $l(l+1)\hbar^2$ of $\hat{L}^2$ — the subject taken up in full when computing the Eigenvalues of L-squared and Lz. It is worth visualizing what $\hat{L}_+$ and $\hat{L}_-$ do to the classical picture of angular momentum precession. Classically, $L_x$ and $L_y$ oscillate as the vector precesses about the $z$-axis at some fixed $L_z$ and fixed $|\mathbf{L}|$; the complex combination $L_x+iL_y$ is exactly the term that rotates at a single fixed frequency in the complex plane. Quantum mechanically, that same combination becomes an operator that shifts the discrete "phase" quantum number $m$ rather than continuously rotating a vector — the ladder replaces the classical precession cone with a stack of discrete allowed cones, shown schematically below.

The entire ladder-operator formalism rests on three commutators, all derivable from the fundamental relations established for $\hat{L}_x$, $\hat{L}_y$, $\hat{L}_z$:

where: $[\hat{A},\hat{B}]\equiv\hat{A}\hat{B}-\hat{B}\hat{A}$, and the three relations are cyclic permutations of $x\to y\to z\to x$. Reading: these say the three Cartesian components of angular momentum cannot be measured simultaneously with arbitrary precision — measuring one component disturbs the others, which is why chemists work instead with $\hat{L}^2$ and a single component (conventionally $\hat{L}_z$) as the compatible pair of observables. From these, the ladder operators satisfy

where: the $+$ sign belongs to $\hat{L}_+=\hat{L}_x+i\hat{L}_y$ and the $-$ sign to $\hat{L}_-=\hat{L}_x-i\hat{L}_y$. Reading: this single equation is the algebraic definition of "raising" and "lowering" — it says that $\hat{L}_\pm$ shifts the $\hat{L}_z$-eigenvalue of any state it acts on by exactly $\pm\hbar$. Also essential:

where: $\hat{L}^2=\hat{L}_x^2+\hat{L}_y^2+\hat{L}_z^2$ commutes with every component of $\hat{\mathbf{L}}$, hence with $\hat{L}_\pm$ too. Reading: $\hat{L}_\pm$ never changes the total-angular-momentum quantum number $l$ — it only moves a state up or down within the same $2l+1$-member family. Finally, the two ladder operators do not commute with each other:

where: this follows from expanding $\hat{L}_+\hat{L}_- - \hat{L}_-\hat{L}_+$ in terms of $\hat{L}_x,\hat{L}_y$ and using $[\hat{L}_x,\hat{L}_y]=i\hbar\hat{L}_z$. Reading: this commutator is what generates the exact normalization constant in $\hat{L}_\pm|l,m\rangle=\hbar\sqrt{l(l+1)-m(m\pm1)}\,|l,m\pm1\rangle$, since it lets you rewrite $\hat{L}_+\hat{L}_-$ and $\hat{L}_-\hat{L}_+$ in terms of $\hat{L}^2$ and $\hat{L}_z$. Two more useful identities, obtained by direct expansion, complete the toolkit:

where: both are Hermitian, positive semi-definite operators (each is of the form $\hat{O}^\dagger\hat{O}$), which is the algebraic seed of the ladder-termination argument. Reading: sandwiching these between $\langle l,m|$ and $|l,m\rangle$ converts an abstract operator identity into a concrete, checkable number — the squared norm of $\hat{L}_\pm|l,m\rangle$.

Example 1 — Foundational: stepping through a $p$-orbital ladder ($l=1$). Take $l=1$, so $m\in\{-1,0,1\}$, the three components of a $p$-subshell. Start at $|1,0\rangle$ (a $p_z$-type state in the $|l,m\rangle$ basis) and apply $\hat{L}_+$:

Now apply $\hat{L}_+$ again to $|1,1\rangle$, the top rung:

confirming the ladder stops exactly at $m=1$ (there is no $m=2$ state for $l=1$; the coefficient vanishes identically rather than being merely small). Going the other way from $|1,0\rangle$,

and applying $\hat{L}_-$ once more to $|1,-1\rangle$ gives $\hbar\sqrt{2-(-1)(-2)}=\hbar\sqrt{2-2}=0$, again terminating cleanly. As a limiting-case check: setting $l=0$ (an $s$-orbital) the only allowed $m$ is $0$, and indeed $\hat{L}_+|0,0\rangle=\hbar\sqrt{0-0}=0$ and $\hat{L}_-|0,0\rangle=0$ — the ladder algebra correctly predicts that an $s$-state has no partners to raise or lower into, consistent with its single, non-degenerate character.

Example 2 — Applied: a non-round matrix element on the $d$-manifold ($l=2$). Consider a $d$-orbital ladder, $l=2$, and evaluate the raising step from $m=-1$ to $m=0$ (chosen deliberately off-center and asymmetric rather than the more commonly quoted $m=0\to1$ step):

Now check the corresponding lowering step from $m=0$ back down using $\hat{L}_-$, which should return to $m=-1$ with a different, independently computed prefactor:

The two prefactors agree ($\hbar\sqrt{6}$ both ways) because $\hat{L}_-$ is the adjoint of $\hat{L}_+$ and the states are real-normalized — this cross-check is a built-in consistency test whenever you apply the ladder in both directions. As a further non-symmetric check, compute the step from $m=1$ to $m=2$ (the very top of the $l=2$ ladder):

Applying $\hat{L}_+$ once more from the top, $m=2\to3$, must vanish: $\hbar\sqrt{6-2(3)}=\hbar\sqrt{0}=0$, confirming that $l=2$ truly has no sixth rung.

Example 3 — Exam/competition level: expectation value of $\hat{L}_x^2$ using ladder operators, no integrals. A common competition-style question: for a state $|l,m\rangle$, find $\langle \hat{L}_x^2\rangle$ without ever writing an explicit spherical harmonic, and evaluate it numerically for $l=3$, $m=1$ (an $f$-orbital state, again a deliberately asymmetric, non-round choice). Since $\hat{L}_x = \tfrac{1}{2}(\hat{L}_++\hat{L}_-)$, we have

The cross terms $\hat{L}_+^2$ and $\hat{L}_-^2$ shift $m$ by $\pm2$, so their expectation value in $|l,m\rangle$ vanishes by orthogonality of the ladder states. What survives is

Using the two operator identities from the core-ideas section,

so

Plugging in $l=3$, $m=1$:

As a check by symmetry, the same manipulation gives an identical formula for $\langle\hat{L}_y^2\rangle$ (since $\hat{L}_y=\tfrac{1}{2i}(\hat{L}_+-\hat{L}_-)$ produces the same surviving cross terms with signs that cancel identically), so $\langle\hat{L}_x^2\rangle=\langle\hat{L}_y^2\rangle=5.5\,\hbar^2$ here — consistent with $\langle\hat{L}_x^2\rangle+\langle\hat{L}_y^2\rangle+\langle\hat{L}_z^2\rangle = \langle\hat{L}^2\rangle = l(l+1)\hbar^2 = 12\hbar^2$, since $5.5+5.5+1^2=12$ exactly ($\langle\hat{L}_z^2\rangle=m^2\hbar^2=\hbar^2$ here). This internal consistency check — that the three mean-square components sum correctly to $\langle\hat{L}^2\rangle$ — is exactly the kind of cross-verification the ladder algebra makes almost automatic.

The ladder-operator technique for angular momentum is usually credited to Paul Adrian Maurice Dirac, who introduced the general algebraic method of raising and lowering operators in his 1930 treatment of quantum mechanics, generalizing an approach he had already used for the quantum harmonic oscillator. The broader angular-momentum commutation algebra itself traces to the early matrix-mechanics work of Max Born, Werner Heisenberg, and Pascual Jordan, and was placed on its modern operator footing through the wave-mechanical formalism of Erwin Schrödinger and the transformation theory of Dirac. Dirac's contribution was to show that the entire quantization of angular momentum — the discreteness of $l$ and $m$, the $2l+1$ degeneracy, even the eventual admission of half-integer spin — could be extracted purely from commutator algebra, without solving any differential equation, a method that became indispensable once electron and nuclear spin (which have no coordinate-space wavefunction at all) needed to be quantized by the same rules.