Lecture 20 Mar 21#
Variational method applied to helium#
The variational method uses the fact that for any state \(\ket{\Psi}\),
where \(E_1\) is the ground state energy for the Hamiltonian \(\hat{H}\). The denominator is there in case \(\ket{\Psi}\) is not normalized. This is straightforward to see: if \(\ket{\Psi}\) was the true ground state, then we would get equality. For any other state, we have introduced some contributions from higher order states which have higher energies, so the value increases.
This is a powerful method which allows an estimate of the ground state energy to be obtained for Hamiltonians \(\hat{H}\) that can’t be solved exactly. A common approach is to consider a parametrized family of possible wavefunctions, and find the values of the parameter(s) that minimize the energy. This is then the best estimate (upper limit) for the ground state energy.
Exercise:
Apply this idea to the ground state of helium. For the trial wavefunction, replace the charge \(Z\) in the hydrogen ground state \(\psi_{100}\) with an effective charge \(Z_{\rm eff}\) – the idea is that the electron feels a smaller force from the nucleus because of screening from the other electron. The energy can then be minimized with respect to the parameter \(Z_{\rm eff}\). What value of effective charge minimizes the energy and what is the ground state energy that you obtain?
To help you avoid doing a lot of integrals, here are some useful formulae. For a wavefunction \(\psi(r) \propto e^{-\beta(r_1+r_2)/a_0}\),
Derivation of these expectation values
Let’s do first \(\langle 1/r_1\rangle\). The full integration we need to do is
where the trial wavefunction depends only on \(r_1\) and \(r_2\), not on the angles \(\theta_1\), \(\phi_1\), \(\theta_2\) or \(\phi_2\), and I’ve explicitly included the normalization factor. Since the quantity we’re averaging depends only on \(r_1\), the integrals cancel except for the \(r_1\) integration, giving
The answer is the same for \(\langle 1/r_2\rangle\).
Similarly,
and the same for \(r_2\).
The last one is more tricky. We want
Choose coordinates such that the \(z\)-axis lies along \(\mathbf{r}_1\). Then \(\mathbf{r}_1\cdot\mathbf{r}_2 = r_1r_2\cos\theta\) and we can write the integral as
The angular part is
giving
The normalization factor is
Therefore
Solution
We need to evaluate the expectation value of the full Hamiltonian for helium using the trial wavefunction \(\psi(r_1,r_2)\propto e^{-Z_\mathrm{eff} (r_1+r_2)/a_0}\).
The Coulomb terms are straightforward using the integrals given. They are
We have to be a bit more careful with the kinetic energy term. Since the two kinetic energy terms give equal contributions, we can write the total kinetic energy as
(so we get a positive kinetic energy term, which makes sense).
Adding all terms, we have
Minimizing with respect to \(Z_{\rm eff}\), ie. setting \(dE/dZ_{\rm eff}=0\) gives
and
This is closer to the correct answer than perturbation theory.
Further reading#
Townsend Chapter 12.2 discusses the helium atom.
We’re going to move on to Chapter 13 next week, but if you are interested you can look at 12.3 which talks about the filling of shells in multi-electron atoms, and section 12.4 on the H\(_2^+\) molecule, which has an interesting analogy to the double square well and ammonia problems that we looked at earlier in the course.
See Kinoshota (1957) for a more sophisticated 39-parameter estimate of the helium ground state energy.