PHYS 623 Introduction to Quantum Mechanics II
Spring 2018, University of Maryland
Prof. Ted Jacobson

Notes



Thursday Mar. 1:
Today:

1. The simple product of hydrogenic variational wavefunctions for a proton with two electrons yields an upper bound -0.945 Ry to the ground state energy. Since this is not less than -1 Ry, it is higher that the ionized state, H + e-, so it fails to reveal the existence of the hydrogen anion, H-. Challenge: try to find a trial wavefunction that establishes the existence of the bound state. It will presumably need to incorporate correlations between the two electrons, so that the electrons tend to be on opposite sides of the proton...If you find one, please share it with the class.

2. Excited states of helium, as covered in Schwabl. I harped on the fact that while the exchange integral "K" can be shown to be positive, it isn't manifestly positive, and I conjectured that if the Coulomb potential were modified, it might not be positive. For instance: suppose the denominator is replaced by d + w, where d is the distance between the electrons (the usual Coulomb denominator), and w is a constant length. This would decrease the advantage of the two electrons avoiding being in the same location, which leads me to suspect that it might invalidate the positivity of the integral. Let me know if you test this (for different w's)...

3. Multi-electron atoms: I tried to summarize the salient points. Many more details, and in particular explanations of the Hartree-Fock method of the self-consistent field approximation, can be found in Schwabl or Littlejohn. I don't plan to delve into those details, but wanted to emphasize a conceptual point: the ground state, being an energy eigenstate of a rotationally invariant Hamiltonian, must be an eigenstate of F2 (unless it's degenerate, which it ain't except wrt the mF value), hence has a definite total angular momentum quantum number F. To the extent that the hyperfine interaction can be neglected, it can also be taken to be an eigenstate of J,L and S, but that is only a (very good for many purposes) approximation.

4. I explained the notion of electron configurations and shells, along the lines in Schwabl. I also explained the meaning of the "term symbol", (2S+1)LJ, and how Hund's rules allow you to infer --- usually correctly --- the term symbol, given the configuration. I explained the example of oxygen, showing it is
3P2. I'd like to do another couple of examples here: iron (Fe), and ytterbium (Yb).

Example: The configuration of Fe is [Ar]
4s2 3d6. The 4s shell is filled but the 3d shell has room for 2x5=10 electrons, and contains only 6. To maximize the spin, we can distribute 5 electrons with spin up among the 5 ml values, and place the 6th electron with spin down in any one of the orbitals. That yields S = 2. The maximal ml would be obtained if we placed the 6th electron in the ml=2 orbital, yielding a maximal total ml=2, and therefore L = 2 is the maximal L accessible (given that S was already maximized). Finally, since the shell is more than half-filled, we should maximize J given S=2 and L=2, which yields J=4, and therefore the term symbol is 5D4.

The states with higher J values but the same L and S are excited states, and it's interesting to inspect the energy differences. You can see them here: https://physics.nist.gov/PhysRefData/Handbook/Tables/irontable5.htm.  If you compute the energy differences and divide by J, for adjacent levels, [E(J) - E(J-1)]/J, you find that they are fairly constant: -104, -96, -92, -90, in units of inverse cm. This near-constancy follows from the Landé interval rule, which in turn follows from the fact that
, in first order perturbation theory, the net effect of the spin-orbit coupling can be seen to be equal, for each configuration and L and S,  to an LS dependent constant times the expectation value of L.S = (J^2 -L^2 - S^2)/2 = [J(J+1) -L(L+1) - S(S+1)]/2. The change of this quantity between J and J-1, for fixed L and S, is J. The adjacent energy differences are thus proportional to J, so if you divide them by J, the result should be J independent. As to the magnitude of the energy differences, for instance, between the ground state (J=4) and the next excited state (J=3) the energy goes up by 416 inverse cm. Now 1/cm is about 0.12 meV, so 416/cm is 50 meV. Compared to 10 eV this is 50 10-4, which is ~ 10-3. This is ~ 10 times larger than the relative size of relativistic corrections in hydrogen. Apparently, as the atomic number grows, spin-orbit coupling strengths do increase, I suppose because the participating electrons are moving faster and/or they are sometimes seeing a larger central charge.

The final example I want to mention here is Yb, since a student mentioned using that atom in a quantum computer setting in the lab she works in, and since it is used in the most stable atomic clock. (To read about the clock, see the article linked on the Supplements page.) The ground state configuration is
[Xe]4f146s2. The letter f denotes l=3, whose shell has 2x7=14 states, so it is a filled shell, which has S=0 and L=0, and the term symbol of the ground state is 1S0. The two 6s electrons are like in helium, so among the excited states is a spin triplet (S=1) (like orthohelium) that also has orbital angular momentum L=1, and the fine structure of this level includes J = 0,1,2 sublevels. The J=0 term has term symbol 3P0. In this state, the spin and orbital angular momenta are each nonzero but they add to zero. The one photon transition from this state down to the ground state is "completely forbidden" as we'll see, because the ground state also has J=0, and there is no J=0 to J=0 one photon transition. If that were the whole story the lifetime of this state would be extremely long (years?). That would be the case for one of the I=0 isotopes of yttrium. However, the long lifetime also means the transition line "width" is extremely narrow, so it is too difficult to excite the transition to be practical. The Yb-171 isotope, on the other hand, has nuclear spin I=1/2, and therefore a magnetic dipole moment, which introduces a hyperfine interaction. This means that the 3P0(F=1/2) state mixes with the 3P1(F=1/2) state, and the latter state has an allowed decay mode. Although not as narrow, it's still a very narrow transition, and correspondingly very long lived, since a spin flip is needed to get from the triplet to the singlet spin state, and since (I presume) the amount of admixture of the latter state is small. 

Tuesday Feb. 20:
 
For a pdf file of these notes is here.
Today I finished discussing the hyperfine splitting of atomic energy levels. Since I approached the material rather differently than does Schwabl or Littlejohn, I wanted to make a brief synopsis here. The following main points were made:

1. Reviewed definition of gyromagnetic ratio and g-factor.

2. Hyperfine interactions are interactions of nuclear magnetic dipole or electric quadrupole moments with the field produced by the electrons, evaluated at the nucleus. One could also view this the other way around: the influence of nulcear multipole fields on the electron. Either way, it's the interaction energy that matters, and we'll view it the first way.

3. An electric dipole moment (EDM) is forbidden unless the Hamiltonian violates both time reversal symmetry and parity symmetry. (See my Supplement on Nuclear moments.) Parity is strongly violated by the weak interactions, and time reversal symmetry is violated by the complex phases in the quark mixing matrix of the standard model, but that would be a very small effect in nuclei, so some very small EDM is expected, but none has been detected as yet. The main search is for a neutron EDM. The present upper limit is 10^5 times larger than the value expected from the quark mixing matrix, but other theoretical considerations suggest that there should be a larger value, and finding it could be an important sign of new physics. See the Supplement for a few more details.

4. The nuclear magneton e_0/(2m_p c) is smaller than the Bohr magneton e_0/(2m_e c) by a factor m_e/m_p ~ 1/1836 ~ 10^-3.

5. The magnetic hyperfine interaction Hamiltonian is H_hf,mag = - µ_nucleus.B_electrons, where the electronic magnetic field is evaluated at the nuclear position. Using the Wigner-Eckart theorem, I showed that the expectation value of this Hamiltonian in states of definite |FIJm_F> is equal to the expectation value of I.J, times two factors that depend only on I and J.  (Here I and J are the nuclear and electronic total angular momenta, and F is the overall total angular momentum of the atom.) The first factor is the nuclear gyromagnetic ratio. The second factor depends on the electronic state. (The  key observation behind the argument is that µ_nucleus is a nuclear vector operator and B_electrons is an electronic vector operator. Using this, we may invoke the aspect of the Wigner-Eckart theorem that states that matrix elements of any two tensor operators of the same rank are proportional to each other.)

6. When treating the hyperfine interaction as a perturbation, we face the fact that the states labeled by different m_I & m_J are degenerate, so we must find the eigenvectors and eigenvalues of the hyperfine Hamiltonian truncated to the degenerate (IJ) subspace. In view of point 5., this is easy, because I.J = (F^2 - I^2 - J^2)/2 = [F(F+1) - I(I+1) - J(J+1)]/2, i.e.,
I.J is already diagonal in the |FIJm_F> basis. The level shifts thus have the form A(I,J)[F(F+1) - I(I+1) - J(J+1)].

7. The must fundamental example is the ground state of the hydrogen atom. The unperturbed electronic orbital state is 1s, and the spin state is arbitrary, while the nuclear spin state is also arbitrary. Both the electron and the proton have spin-1/2, so I = 1/2 and J = 1/2, hence the possible values of F are 1 and 0, the triplet and singlet. The corresponding values of
[F(F+1) - I(I+1) - J(J+1)] are 1/2 and -3/2. It turns out that the coefficient A here is positive, so the ground state is the singlet, which is depressed three times as much as the triplet is raised. The frequency of this transition (i.e. the energy difference divided by hbar) is around 1.4 GHz, and the wavelength is the famous "21 cm line".

8. A super important example to physics is the ground state of Cs-133. Cesium is an alkalai atom, with a 6s valence electron. All the other electrons make up closed shells, which are exactly rotationally invariant, and so contribute nothing to the magnetic moment of the electrons. The nuclear spin is I = 7/2, so the possible values of F are 4 and 3. The energy difference between these is radiation with a frequency around 10 GHz. In fact the SI unit of the second is *defined* to be exactly 9,192,631,770 periods corresponding to this frequency.

9. And now what about electric quadrupole moments? See the Supplement for the definition. The key thing is that the quadrupole moment operator is a rank 2 irreducible tensor operator. Hence the selection rule aspect of the Wigner-Eckart theorem tells us that a spin-1/2 nucleus cannot have an electric quadrupole moment: <1/2|Q_2|1/2> = 0, because 1/2 is not in 2 x 1/2. Any nuclear spin greater than 1/2 however does support an electric quadrupole, and in particular the Cs-133 nucleus does. So why don't we need to consider how that affects the fine structure of the ground state of Cs-133? Why can we conclude it is only split into a doublet, corresponding to F = 4 and F = 3?

The answer, again, comes from the selection rule. To understand this, we should first write down the interaction energy between the nuclear quadrupole moment and the electronic field. I gave an only partly true explanation of this in class, so let me try to fix it here: The electrostatic interaction energy of a charge distribution rho and an electrostatic field Phi is ∫ rho Phi dV. This leads to a quadrupole interaction energy 1/2 Q^ij Phi,ij(0) (see below for a derivation). Since Q^ij is traceless, the trace part of Phi,ij(0) doesn't contribute here, so we may replace this by

1/2 Q^ij W^ij, where W^ij = (Phi,ij - 1/3 Phi,kk delta^ij)(0).

W^ij is an operator on the electronic Hilbert space, since Phi is generated by the electron(s), and in fact it is an irreducible tensor operator of rank 2. Hence its expectation value vanishes in electronic states with J = 0 or J = 1/2. In particular, in the case of Cs-133, <6s|W^ij|6s> = 0. All of which is to say that, although the cesium-133 nucleus has a reasonably large electric quadrupole moment, it can't "feel" the electric field generated by the 6s electron!
 
Derivation of the quadrupolar interaction energy:
Let's expand Phi around the origin, taken to be the center of mass of the nucleus, viz,

Phi(x) = Phi(0) + Phi,i(0) x^i + 1/2 Phi,ij(0) x^i x^j + ....

The i and j indices are the cartesian coordiante indices, the comma indicates partial derivative, and repeated indices are summed over. Now we can carry out the interaction energy integral with this expansion, and we get

(*)      Q Phi(0) + p^i Phi,i(0) + 1/2 Q^ij Phi,ij(0) + 1/6 (∫ rho x^2 dV)Phi,ii(0) + ... 

The last term is needed because in order to express ∫ rho x^i x^j dV in terms of the quadrupole moment Q^ij we must subtract and then add back in the trace part:

∫ rho x^i x^j dV = ∫ rho (x^i x^j - 1/3 x^2 delta^ij) dV + 1/3 delta^ij ∫ rho x^2 dV.

The first term in this expansion (*) corresponds to the Coulomb interaction, the second term is the EDM discussed above, which nearly vanishes, and  the third term is the
electric quadrupole interaction. The last term is a second moment of the charge distribution, multiplied by the Laplacian of the potential evaluated at the origin. According to Maxwell's equations, the latter is proportional to the electron charge density at the origin. This term represents a nuclear finite size correction to the Coulomb potential, and is a scalar, so it doesn't break the degeneracy among the m_I m_J states.