← quantum · contents

§ Phase iii · The Hydrogen Atom · Chapter 01

Spherical harmonics

Before atoms, before quantum mechanics, before anyone knew what an electron was, the shapes were already there. Pierre-Simon Laplace wrote them down in 1782 while trying to compute the gravitational pull of a lumpy planet. A century and a half later they were rediscovered, unchanged, as the patterns of electron clouds in the hydrogen atom. This chapter is about why the same handful of shapes keeps reappearing every time geometry meets symmetry.

In the spring of 1782, a thirty-three-year-old Frenchman sat down to compute the gravitational potential of a non-spherical Earth. His name was Pierre-Simon Laplace. He was already on his way to becoming the most formidable mathematical physicist in Europe, the man who would later boast to Napoleon that his five-volume celestial mechanics had no need of the hypothesis of God. The problem in front of him was, on its face, a problem in eighteenth-century geodesy. The Earth is not exactly a sphere. It bulges slightly at the equator because it spins. That tiny bulge perturbs the orbit of the Moon. To predict the lunar position one year in advance, to within the arc-second that the new Paris Observatory could measure, Laplace needed to expand the gravitational potential of a lumpy planet in some clean, computable way.

The trick he reached for was an old one with a new twist. Adrien-Marie Legendre, his slightly older rival at the Académie des Sciences, had already shown that the gravitational potential of a point mass, written as a function of distance and angle, expands into a tidy series of polynomials in the cosine of the angle. Those Legendre polynomials handle one direction, the polar angle, but they say nothing about the azimuth (the longitude). Laplace’s contribution was to do the whole job. He wrote the potential as a double sum over two integer indices and showed that the angular factors, the things that depend on latitude and longitude but not on distance, are an infinite family of functions on the surface of the sphere. They were not yet called spherical harmonics. The English physicist William Thomson, later Lord Kelvin, would give them that name in 1867 together with Peter Tait. But by 1785 every working astronomer in France had Laplace’s series at the back of their notebook.

The motivation was almost embarrassingly practical. Laplace was not chasing an abstract idea. He was paid by the French Bureau des Longitudes to produce ephemerides, tables of where the planets would be on every night of every year. Errors in those tables stranded ships at sea and embarrassed naval captains. The new functions were a tool for predicting tides, for computing the shape of the Earth from pendulum measurements, for understanding why Saturn’s rings did not collapse. They were the right basis set for any quantity defined on a sphere, the spherical analogue of the trigonometric sines and cosines Fourier would soon use on a line. None of this had anything to do with atoms. Atoms, in 1782, were a philosophical conjecture.

Now skip a hundred and forty years. The year is 1926. Erwin Schrödinger, working in a chalet in the Swiss Alps over the Christmas of 1925 and through the early months of 1926, is trying to write down a wave equation for the electron in a hydrogen atom. The problem he is solving is, geometrically, very similar to Laplace’s. A single point of negative charge (the electron) lives in the spherically symmetric Coulomb field of a single point of positive charge (the proton). The wave equation will involve the Laplacian operator (the same one Laplace defined in 1782) acting on a complex-valued wavefunction. The whole setup has the rotational symmetry of a sphere. Anything Schrödinger writes down for ψ(r, θ, φ) must respect that symmetry.

The first move is the obvious one. Try to separate the wavefunction into a radial part and an angular part. Write ψ(r, θ, φ) = R(r) Y(θ, φ) and ask what each piece must satisfy. The radial part R(r) will encode the energy and the distance scale, both of which depend on the strength of the Coulomb pull. The angular part Y(θ, φ) will encode the shape: which directions the electron prefers, how the lobes point, where the cloud thins to nothing. And here is the punchline. The equation for Y turns out to be exactly the equation Laplace solved in 1782. The angular part of any spherically symmetric quantum problem (any problem where the potential depends only on the distance from a center) is forced to be a spherical harmonic. The radial part carries the chemistry. The angular part is pure geometry, the same in hydrogen as in a vibrating brass bell.

This is one of the deepest unification stories in physics, and it is worth pausing over. The angular shapes of orbitals are not chosen by the electron. They are not consequences of the Coulomb law. They are not even consequences of quantum mechanics. They are consequences of the fact that the problem has the symmetry of a sphere. Replace the proton with any other spherically symmetric source of attraction (a square-well potential, a harmonic trap, a hard-sphere repulsion) and the same Y_ℓ^m will appear. The orbital labels s, p, d, f are not chemistry. They are a property of the round.

So we have a quantum number ℓ that tells us which spherical harmonic family we are in (ℓ = 0, 1, 2, 3, …), and within each family a second quantum number m that runs from −ℓ to +ℓ. That gives 2ℓ + 1 functions at each level: one s, three p, five d, seven f. The numbers are not negotiable. They come out of demanding that the wavefunction be single-valued on the sphere, that walking once around the equator brings you back to the same complex number you started with. Without single-valuedness the world would be multi-valued, which is to say not a world at all. The integer ℓ and the integer m are the smallest concession the mathematics will make to the fact that a sphere is a closed surface.

What do these functions look like? The simplest, Y_0^0, is just a constant. Everywhere on the sphere it takes the same value. Square it (the probability density of finding the electron at that angle) and you get a uniform glow over the whole sphere. That is the s-orbital: a single fuzzy ball, no preferred direction. For ℓ = 1 there are three functions, Y_1^{−1}, Y_1^0, Y_1^{+1}, and after a standard recombination into real-valued forms they become p_x, p_y, p_z: three dumbbells aligned with the three coordinate axes. Each one has a single nodal plane, a flat surface through the origin where the wavefunction is exactly zero. For ℓ = 2 there are five functions, and these are the d-orbitals. Four of them are cloverleaf shapes (d_{xy}, d_{yz}, d_{xz}, d_{x²−y²}); the fifth, d_{z²}, is the famous donut-around-a-dumbbell. Every d-orbital has two nodal surfaces. The pattern continues: ℓ nodal surfaces, 2ℓ + 1 independent shapes, ever more lobes as ℓ climbs.

The nodal structure is the easiest way to read a spherical harmonic by eye. A nodal surface is a place where the wavefunction is exactly zero, so the probability of finding the electron there is also exactly zero. The s-orbital has none. The p-orbitals each have one nodal plane: the xy plane for p_z, the xz plane for p_y, the yz plane for p_x. The d-orbitals each have two. The cloverleafs have two perpendicular nodal planes. The d_{z²}, the famous oddball, instead has a pair of nodal cones (think two ice-cream cones meeting at the origin, opening up and down) whose surfaces pinch the wavefunction to zero everywhere on those cones. Squint at any orbital picture and count the nodal surfaces; that is your ℓ.

∇² = (1/r²) ∂_r (r² ∂_r) + (1/r²) [angular piece]

L̂² Y = ℏ² ℓ(ℓ+1) Y

Y_ℓ^m(θ, φ) = N_{ℓm} · P_ℓ^m(cos θ) · e^{i m φ}

It is worth seeing one specific harmonic up close before we leave the chapter. The most distinctive of the five d-orbitals, the one that looks unlike its siblings, is Y_2^0. Its angular shape in spherical coordinates is proportional to (3 cos²θ − 1), which is the Legendre polynomial P_2(cos θ). Walk through what that means. At the north pole (θ = 0) cos θ is 1, and 3·1 − 1 = 2: the function is positive and as large as it gets. As you slide down toward the equator, cos θ shrinks; the function decreases, passes through zero when 3 cos²θ = 1 (that is, at θ ≈ 54.7° and again at θ ≈ 125.3°), turns negative, reaches its most negative value at the equator, and then mirrors back. The squared function (the probability density) is symmetric about the equator. It has two large lobes pointing up and down along z, a vanishing ring at the two cones, and a smaller torus of probability sitting around the waist. The signs are different in the two regions, which matters when this orbital overlaps with its neighbors during bonding, but in a probability picture they are both glowing.

The angle 54.7° (the magic angle, as NMR spectroscopists call it) is not arbitrary. It is the solution of 3 cos²θ = 1, the place where the Legendre polynomial P_2 changes sign. The same angle reappears whenever an experimentalist needs to suppress a d-orbital-like signal in a magnetic resonance experiment. Spin a sample around an axis tilted at 54.7° from the field and the dipolar interaction averages to zero, because the spatial part of the dipole-dipole coupling has the same P_2(cos θ) shape and so vanishes when integrated over a cone at the magic angle. A trick that looks like wizardry in a chemistry lab is, underneath, the same nodal cone visible in the d_z² picture above. The mathematics keeps surfacing in places nobody planned for.

We arrive, then, at a strange and beautiful situation. The same handful of shapes (one round, three dumbbell, five cloverleaf, seven f-pattern, and so on) keeps reappearing every time anyone in physics or chemistry writes down a spherically symmetric problem. They are the eigenfunctions of the angular Laplacian, the shapes that survive on a sphere. Laplace found them while computing the orbit of the Moon. Lord Kelvin renamed them while developing his theory of solid harmonics. Schrödinger imported them in 1926 because they were sitting in the toolbox, waiting. Every chemist who memorizes the s, p, d, f labels is memorizing a property of round things.

The next chapter will pick up the other half of the wavefunction, the radial part R(r). That is where Coulomb’s law shows up, where the energies of the hydrogen atom get fixed, where the principal quantum number n is born. The angular part we now know. The angles are the geometry. The radii are the physics.