Quantum Mechanics for Mathematicians: Angular Momentum and Central Potentials

Peter Woit Department of Mathematics, Columbia University woit@math.columbia.edu December 13, 2012
We will now turn to the physical case of 3-dimensional space, and will be working in the Schr odinger representation, so we have H = L2 (R3 ). Using the action of SO(3) on R3 we get an induced representation on H by q1 (g ) (q) = (g 1 q) = (g 1 q2 ) q3 This representation was studied in detail in chapter 8, where we found that the Lie algebra representation q1 d d (X ) (q) = (etX )t=0 (q) = (etX q2 ) dt dt q3 acted as follows for l1 , l2 , l3 a basis of so(3). ( l1 ) = q3 d d q2 dq2 dq3

d d ( l2 ) = q1 q3 dq3 dq1 d d ( l3 ) = q2 q1 dq1 dq2 In our general study of the action of the Heisenberg group H2n+1 and the symplectic group Sp(2n, R) on the state space H, we saw that, at least at the Lie algebra level, this came from just using the operators Pj = i , Qj = qj , j = 1, 2, 3 qj

and considering polynomials in these operators of degree 0, 1, 2. Our SO(3) rotation group acts on both position and momentum variables, preserving inner products, so it will also preserve the symplectic form ( p p , q q )=qp pq

and thus we have SO(3) Sp(6, R), the group whose Lie algebra acts on H by quadratic polynomials in the Qj , Pj . We will nd that certain specic quadratic polynomials (the angular momenta) give the so(3) action, recovering the same Lie algebra representation we found earlier just from the action of SO(3) on functions on R3 . When the Hamiltonian function is invariant under rotations, we then expect eigenfunctions of the corresponding Hamiltonian operator to carry representations of SO(3). The eigenfunctions of a given energy break up into irreducible representations of SO(3), and we have seen that these are labeled by a integer l = 0, 1, 2, . . .. One can use this to nd properties of the solutions of the Schr odinger equation whenever one has a rotation-invariant potential energy, and we will work out what happens for the case of the Coulomb potential, describing the hydrogen atom. We will see that this specic case is exactly solvable because it has a second not-so-obvious SO(3) symmetry, in addition to the one coming from rotations in R3 .

Angular momentum
used the following basis elements 0 1 0 1 0 0 0 l3 = 1 0 0 0 0 0 0 0

In our study of the Lie algebra so(3) we 0 0 0 0 l1 = 0 0 1 l2 = 0 0 1 0 1 which satisfy the commutation relations

[ l1 , l2 ] = l3 , [ l2 , l3 ] = l1 , [ l3 , l1 ] = l2 or in a more compact notation [ la , lb ] =

abc lc

where the indices take values 1, 2, 3, we are using the Einstein summation convention of summing the repeated index c on the right,
123

=1

and the sign of abc changes when one interchanges two of its indices. We would like to nd quadratic polynomials in q1 , q2 , q3 , p1 , p2 , p3 that satisfy these relations as Poisson bracket relations. Then quantization will give us 2

a unitary representation of so(3) on H = L2 (R3 ). From elementary classical physics, one knows that the angular momentum vector is dened to be l = q p, which has components l1 = q2 p3 q3 p2 , l2 = q3 p1 q1 p3 , l3 = q1 p2 q2 p1 One can easilly check that these satisfy the Poisson bracket relations {l1 , l2 } = l3 , {l2 , l3 } = l1 , {l3 , l1 } = l2 Quantization gives us corresponding self-adjoint operators on L2 (R3 ) L1 = Q2 P3 Q3 P2 = i (q2 L2 = Q3 P1 Q1 P3 = i (q3 d d q3 ) dq3 dq2

d d q1 ) dq1 dq3 d d q2 ) L3 = Q1 P2 Q2 P1 = i (q1 dq2 dq1 which satisfy [L1 , L2 ] = i L3 , [L2 , L3 ] = i L3 , [L3 , L1 ] = i L2 Note that the relation of these operators Lj to the operators that give the Lie algebra representation is i ( lj ) = Lj with the i there to make the operator skew-adjoint, and the due to the relative normalization of the Poisson and Lie brackets. The Lie algebra representation operators are i d d ( l1 ) = L1 = q3 q2 dq2 dq3 d i d q3 ( l2 ) = L2 = q1 dq3 dq1 i d d ( l3 ) = L3 = q2 q1 dq1 dq2 and these are exactly the ones we found in chapter 8 as coming from the rotation action of so(3) on functions on R3 . It is the operators i e (lj ) = e Lj that give the operator corresponding to a rotation by about the j -axis. From the point of view of quantum mechanics, this story is independent of , with the eigenvalues of observables just the integer weights one nds in the action of SO(3) on its representations. It is only if one wants to interpret these eigenvalues in terms of the classical notion of angular monentum that one needs to introduce (which sets the scale of the Poisson bracket), and then one can characterize the eigenvalues as integer multiples of . 3

Quantum particle in a central potential

The Hamiltonian operator for the quantum system describing a particle moving in a potential V (q1 , q2 , q3 ) is H= 1 2 3 (P 2 + P2 + P3 ) + V (Q1 , Q2 , Q3 ) 2m 1 2 2 2 2 = ( 2 + 2 + 2 ) + V (q1 , q2 , q3 ) 2m q1 q2 q3 2 = + V (q1 , q2 , q3 ) 2m

We will be interested in so-called central potentials, potential functions that 2 2 2 are functions only of q1 + q2 + q3 , and thus only depend upon r, the radial distance to the origin. For such V , both terms in the Hamiltonian will be SO(3) invariant, and eigenfunctions of H will be representations of SO(3). Using the expressions for the angular momentum operators in spherical coordinates derived in chapter 8, one can show that the Laplacian has the following expression in spherical coordinates = 2 2 1 + 2 L2 2 r r r r

where L2 is the Casimir operator in the representation , which we have shown has eigenvalues l(l +1) on irreducible representations of dimension 2l +1 (integral spin l). So, restricted to such an irreducible representation, we have = d2 2 d l(l + 1) + 2 dr r dr r2

To solve the Schr odinger equation, we need to nd the eigenfunctions of H . These will be irreducible representations of SO(3) acting on H, which we have seen can be explicitly expressed in terms of the spherical harmonic functions Ylm (, ) in the angular variables. So, to nd eigenfunctions of the Hamiltonian
2

H=

2m

+ V (r )

we need to nd functions glE (r) depending on l = 0, 1, 2, . . . and the energy eigenvalue E satisfying ( 2 d2 2 d l(l + 1) ( + ) + V (r))glE (r) = EglE (r) 2m dr2 r dr r2

Given such a glE (r) we will have HglE (r)Ylm (, ) = EglE (r)Ylm (, ) and the (r, , ) = glE (r)Ylm (, ) 4

will be a 2l + 1 dimensional (since m = l, l + 1, . . . , l 1, l) space of energy eigenfunctions for H of eigenvalue E . For a general potential function V (r), exact solutions for the eigenvalues E and corresponding functions glE (r) cannot be found in closed form. One special case where we can nd such solutions is for the three-dimensional harmonic 1 m 2 r2 . These are much more easily found though oscillator, where V (r) = 2 using the creation and annihilation operator techniques discussed in the last chapter. The other well-known and physically very important case is the case of a 1 r potential, called the Coulomb potential. This describes a light charged particle moving in the potential due to the electric eld of a much heavier charged particle, a situation that corresponds closely to that of a hydrogen atom. In this case we have e2 V = r where e is the charge of the electron, so we are looking for solutions to ( 2 d2 2 d l(l + 1) e2 ( 2+ ) )glE (r) = EglE (r) 2 2m dr r dr r r d2 (rg ) = Erg dr2 is equivalent to 2 d2 + )g = Eg dr2 r for any function g , glE (r) will satisfy ( ( 2 d2 l(l + 1) e2 ( 2 ) )rglE (r) = ErglE (r) 2m dr r2 r

Since having

The solutions to this equation can be found through a rather elaborate process. For E 0 there are non-normalizable solutions that describe scattering phenomena that we wont study here. For E < 0 solutions correspond to an integer n = 1, 2, 3, . . ., with n l + 1. So, for each n we get n solutions, with l = 0, 1, 2, . . . , n 1, all with the same energy En = me4 2 2 n2

This degeneracy in the energy values leads one to suspect that there is some extra group action in the problem commuting with the Hamiltonian, so energy eigenfunctions will come in irreducible representations of some larger group than SO(3), one such that when one restricts to the SO(3) subgroup, the representation is reducible, giving n copies of our SO(3) representation of spin l. In the next section we will see that this is the case, and there use representation theory to derive the above formula for En . 5

We wont go through the process of showing how to explicitly nd the functions glEn (r) but just quote the result. Setting
2

a0 =

me2

(this has dimensions of length and is known as the Bohr radius), and dening gnl (r) = glEn (r) the solutions are of the form gnl (r) e na0 (
r

2r l 2l+1 2r ) Ln+l ( ) na0 na0

l+1 where the L2 n+l are certain polynomials known as associated Laguerre polynomials. So, nally, the energy eigenfunctions will be

nlm (r, , ) = gnl (r)Ylm (, ) for n = 1, 2, . . . l = 0, 1, . . . , n 1 m = l, l + 1, . . . , l 1, l The rst few of these, properly normalized, are 100 = 1 a3 0 e a0
r

(called the 1S state, S meaning l = 0) 200 =

r r 2a 0 ) e (1 2a0 8a3 0

(called the 2S state), and the three dimensional l = 1(called 2P , P meaning l = 1 ) states with basis elements 211 = 1 8
r r 2a i 0 sin e e a 0 a3 0

211 = 211 =

1 4 1 2a3 0 a3 0

r r 2a e 0 cos a0

r r 2a e 0 sin ei a0

so(4) symmetry and the Coulomb potential

The Coulomb potential problem is very special in that it has an additional symmetry, of a non-obvious kind. This symmetry appears even in the classical problem, where it is responsible for the relatively simple solution one can nd to the essentially identical Kepler problem. This is the problem of nding the classical trajectories for bodies orbiting around a central object exerting a gravitational force, which also has a 1 r potential. Keplers second law for such motion comes from conservation of angular momentum, which corresponds to the Poisson bracket relation {la , h} = 0 Here well take the Coulomb version of the Hamiltonian that we need for the hydrogen atom problem e2 1 |p|2 h= 2m r One can read the relation {la , h} = 0 in two ways: it says both that the hamiltonian h is invariant under the action of the group (SO(3)) whose innitesimal generators are la , and that the components of the angular momentum (la ) are invariant under the action of the group (R of time translations) whose innitesimal generator is h. Keplers rst and third laws have a dierent origin, coming from the existence of a new conserved quantity for this special choice of Hamiltonian. This quantity is, like the angular momentum, a vector, often called the Lenz (or sometimes Runge-Lenz, or even Laplace-Runge-Lenz) vector. Denition. Lenz vector The Lenz vector is the vector-valued function on the phase space R6 given by 1 q w = (p l) e2 2 m |q| Simple manipulations of the cross-product show that one has lw =0 We wont here explicitly calculate the various Poisson brackets involving the components wa of w, since this is a long and unilluminating calculation, but will just quote the results, which are {wa , h} = 0 This says that, like the angular momentum, the vector wa is a conserved quantity under time evolution of the system, and its components generate symmetries of the classical system.

 {wa , lb } =
abc wc

These relations say that the generators of the SO(3) symmetry act on wa the way you would expect, since wa is a vector. 2h ) m This is the most surprising relation, and it has no simple geometrical explanation (although one can change variables in the problem to try and give it one). It expresses a highly non-trivial relationship between the two sets of symmetries generated by the vectors l, w and the Hamiltonian h. {wa , wb } =
abc lc (

The wa are cubic in the q and p variables, so one would expect that the Groenewold-van Hove no-go theorem would tell one that there is no consistent way to quantize this system by nding operators Wa corresponding to the wa that would satisfy the commutation relations corresponding to these Poisson brackets. It turns out though that this can be done, although not over the entire phase-space. One gets around the no-go theorem by doing something that only works where the Hamiltonian h is positive (well be taking a square root of h). The choice of operator Wa that works is W= Q 1 (P L L P) e2 2m |Q|2

which (by elaborate and unenlightening computation) one can show satisfy [Wa , H ] = 0 [Wa , Lb ] = i 2 H) m [Wa , Lb ] = i abc Wc
abc Lc (

as well as LW =WL=0 Most importantly, one has the following relation between W 2 , H and the Casimir operator L2 2 W 2 = e4 + H (L2 + 2 ) m and it is this which will allow us to nd the eigenvalues of H , since we know those for L2 , and can nd those of W 2 by changing variables to identify a second so(3) Lie algebra. To do this, rst change normalization by dening K= m W 2E 8

where E is the eigenvalue of the Hamiltonian that we are trying to solve for. Note that it is at this point that we violate the conditions of the no-go theorem, since we must have E < 0 to get a K with the right properties, and this restricts the validity of our calculations to a subset of phase space. For E > 0 one can proceed in a similar way, but the Lie algebra one gets is dierent (so(3, 1) instead of so(4)). One then has the following relation between operators 2H (K 2 + L2 + and the following commutation relations [La , Lb ] = i [Ka , Lb ] = i [Ka , Kb ] = i Dening M= one has [Ma , Mb ] = i [Na , Nb ] = i
abc Mc abc Nc abc Lc abc Kc abc Lc 2

) = me4

1 1 (L + K), N = (L K) 2 2

[Ma , Nb ] = 0 which is just two commuting copies of so(3). Recall from our discussion of rotations in three dimensions that representations of so(3) = su(2) correspond to representations of Spin(3) = SU (2), the double cover of SO(3) and the irreducible ones have dimension 2l + 1, with l half-integral. Only for l integral does one get representations of SO(3), and it is these that occur in the SO(3) representation on functions on R3 . For four dimensions, we found that Spin(4), the double cover of SO(4), is SU (2) SU (2), and one thus has spin(4) = so(4) = su(2) su(2) = so(3) so(3). This is exactly the Lie algebra we have found here, so one can think of the Coulomb problem as having an so(4) symmetry. The representations that will occur will include the half-integral ones, since for this larger symmetry one of our so(3) factors no longer corresponds to the action of physical rotations on functions on R3 . Using the fact that LK=KL=0 one nds that M2 = N2 so on our energy eigenstates the values of the Casimirs for the two so(3) factors must match. The relation between the Hamiltonian and these Casimirs is 2H (K 2 + L2 +
2

) = 2H (2M 2 + 2N 2 + 9

) = 2H (4M 2 +

) = me4

On irreducible representations, we will have M 2 = ( + 1) for some half-integral , so we get the following equation for the energy eigenvalues me4 me4 E= 2 = 2 2 (4( + 1) + 1) 2 (2 + 1)2 Letting n = 2 + 1, for = 0, 1 2 , 1, . . . we get n = 1, 2, 3, . . . and precisely the same equation for the eigenvalues described earlier En = me4 2 2 n2

The irreducible representations of a product like so(3) so(3) are just tensor products of irreducibles, and in this case the two factors of the product are identical due to the equality of the Casimirs M 2 = N 2 . The dimension of the so(3) so(3) irreducibles is thus (2 + 1)2 = n2 , explaining the multiplicity of states one nds at energy eigenvalue En .

The Hydrogen atom

The Coulomb potential problem provides a good description of the quantum physics of the hydrogen atom, but it is missing an important feature of that 1 systems. To describe this, one really system, the fact that electrons are spin 2 needs to take as space of states H = L2 (R3 ) C2 The action of the rotation group is now an SU (2) = Spin(3) action on a tensor product. It is given by the homomorphism to SO(3) and the rotation group action of SO(3) on the rst factor of the tensor product. On the second factor it is the spinor representation. Physicists generally write the tensor product Lie algebra representation of the SU (2) operators on H as J=L+S where the rst term acts just on L2 (R3 ) by the rotation action, the second on C2 as the spinor representation, so should perhaps be written as J=L1+1S One can explicitly write elements of H as two-component wave-functions | = + (q) (q)

where the + (q) is thought of as the wave-function for the electron to be in the 1 eigenvalue of S3 ), (q) the wave-function for the spin spin up state (+ 2 10

down state ( 1 2 eigenvalue of S3 ). The Hamiltonian operator is just the identity on the spinor factor, so the only eect of this factor is to double the number of energy eigenstates at each energy. Electrons are fermions, so anti-symmetry of multi-particle wave functions implies the Pauli principle that states can only be occupied by a single particle. As a result, one nds that when adding electrons to an atom described by the Coulomb potential problem, the rst two ll up the lowest Coulomb energy eigenstate (the 100 or 1S state at n = 1), the next eight ll up the n = 2 states ( two each for 200 , 211 , 210 , 211 ), etc. This goes a long ways towards explaining the structure of the periodic table of elements. When one puts a hydrogen atom in a constant magnetic eld B , the Hamiltonian then acts non-trivially on the spinor factor of H, through a term e BS mc

This is exactly the sort of Hamiltonian we began our study of quantum mechanics with for a simple two-state system. It causes a shift in energy eigenvalues proportional to |B| depending on the eigenvalue of S3 (the Zeeman eect), giving dierent energies to the states in + (q) and (q).

For further reading

This is a standard topic in all quantum mechanics books. For example, see chapters 12 and 13 of [3]. The so(4) calculation is not in [3], but is in some of the other such textbooks, a good example is chapter 7 of [1]. For an extensive discussion of the symmetries of the 1 r potential problem, see [2].

References
[1] Baym, G., Lectures on Quantum Mechanics, Benjamin, 1969. [2] Guillemin, V., and Sternberg, S., Variations on a Theme of Kepler, AMS, 1990. [3] Shankar, R., Principles of Quantum Mechanics, 2nd Ed., Springer, 1994.

11