Question 5.38: The harmonic chain consists of N equal masses arranged along...

The harmonic chain consists of N equal masses arranged along a line and connected to their neighbors by identical springs:

\hat{H}=-\frac{\hbar^{2}}{2 m} \sum_{j=1}^{N} \frac{\partial}{\partial x_{j}^{2}}+\sum_{j=1}^{N} \frac{1}{2} m \omega^{2}\left(x_{j+1}-x_{j}\right)^{2} ,

where x_j is the displacement of the jth mass from its equilibrium position. This system (and its extension to two or three dimensions—the harmonic crystal) can be used to model the vibrations of a solid. For simplicity we will use periodic boundary conditions: x_{N+1}=x_{1} , and introduce the ladder operators.

\hat{a}_{k \pm} \equiv \frac{1}{\sqrt{N}} \sum_{j=1}^{N} e^{\pm i 2 \pi j k / N}\left[\sqrt{\frac{m \omega_{k}}{2 \hbar}} x_{j} \mp \sqrt{\frac{\hbar}{2 m \omega_{k}}} \frac{\partial}{\partial x_{j}}\right]           (5.75).

where k=1, \ldots, N-1 and the frequencies are given by

\omega_{k}=2 \omega \sin \left(\frac{\pi k}{N}\right) .

(a) Prove that, for integers k and k′ between 1 and N-1,

\frac{1}{N} \sum_{j=1}^{N} e^{i 2 \pi j\left(k-k^{\prime}\right) / N}=\delta_{k^{\prime}, k} ,

\frac{1}{N} \sum_{i=1}^{N} e^{i 2 \pi j\left(k+k^{\prime}\right) / N}=\delta_{k^{\prime}, N-k} .

Hint: Sum the geometric series.

(b) Derive the commutation relations for the ladder operators:

\left[\hat{a}_{k-}, \hat{a}_{k^{\prime}+}\right]=\delta_{k, k^{\prime}} \text { and }\left[\hat{a}_{k-}, \hat{a}_{k^{\prime}-}\right]=\left[\hat{a}_{k+}, \hat{a}_{k^{\prime}+}\right]=0           (5.76).

(c) Using Equation 5.75, show that

x_{j}=R+\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{\hbar}{2 m \omega_{k}}}\left(\hat{a}_{k-}+\hat{a}_{N-k+}\right) e^{i 2 \pi j k / N} ,

\frac{\partial}{\partial x_{j}}=\frac{1}{N} \frac{\partial}{\partial R}+\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{m \omega_{k}}{2 \hbar}}\left(\hat{a}_{k-}-\hat{a}_{N-k+}\right) e^{i 2 \pi j k / N} .

where R=\sum_{j} x_{j} / N is the center of mass coordinate.

(d) Finally, show that

\hat{H}=-\frac{\hbar^{2}}{2(N m)} \frac{\partial^{2}}{\partial R^{2}}+\sum_{k=1}^{N-1} \hbar \omega_{k}\left(\hat{a}_{k+} \hat{a}_{k-}+\frac{1}{2}\right) .

Comment: Written in this form above, the Hamiltonian describes N-1 independent oscillators with frequencies \omega_{k} (as well as a center of mass that moves as a free particle of mass Nm). We can immediately write down the allowed energies:

E=-\frac{\hbar^{2} K^{2}}{2(N m)}+\sum_{k=1}^{N-1} \hbar \omega_{k}\left(n_{k}+\frac{1}{2}\right) .

where \hbar K is the momentum of the center of mass and n_{k}=0,1, \ldots is the energy level of the kth mode of vibration. It is conventional to call n_k the number of phonons in the kth mode. Phonons are the quanta of sound (atomic vibrations), just as photons are the quanta of light. The ladder operators a_{k+} \text { and } a_{k-} are called phonon creation and annihilation operators
since they increase or decrease the number of phonons in the kth mode.

The Blue Check Mark means that this solution has been answered and checked by an expert. This guarantees that the final answer is accurate.
Learn more on how we answer questions.

(a) Consider

\frac{1}{N} \sum_{j=1}^{N} e^{i 2 \pi j r / N}=\frac{1}{N} \sum_{j=1}^{N}\left(e^{i 2 \pi r / N}\right)^{j}=\frac{1}{N} \frac{e^{2 i \pi r}-1}{1-e^{-2 i \pi r / N}} .

For any integer r the numerator vanishes and the sum must be zero unless the denominator also vanishes. The denominator vanishes for r = m N where m is any integer; when r = m N it is easy to see that the expression evaluates to 1. Now, to prove the first identity, k-k^{\prime}=m N only when m = 0 and hence k = k′. In the second case, k+k^{\prime}=m N only when m = 1 and k′ = N – k.

(b) We first consider

\left[a_{k-}, a_{k^{\prime}+}\right]=\frac{1}{N} \sum_{j, j^{\prime}} e^{i 2 \pi\left(j^{\prime} k^{\prime}-j k\right) / N} \frac{1}{2}\left\{-\sqrt{\frac{\omega_{k}}{\omega_{k^{\prime}}}}\left[x_{j}, \frac{\partial}{\partial x_{j^{\prime}}}\right]+\sqrt{\frac{\omega_{k}^{\prime}}{\omega_{k}}}\left[\frac{\partial}{\partial x_{j}}, x_{j^{\prime}}\right]\right\} .

where I’ve dropped the two terms involving commutators of two coordinates or two derivatives. The remaining commutators are -\delta_{j j^{\prime}} \text { and } \delta_{j j^{\prime}} respectively and we have

\left[a_{k-}, a_{k^{\prime}+}\right]=\left(\frac{1}{N} \sum_{j} e^{i 2 \pi j\left(k^{\prime}-k\right) / N}\right) \frac{1}{2}\left[\sqrt{\frac{\omega_{k}}{\omega_{k^{\prime}}}}+\sqrt{\frac{\omega_{k^{\prime}}}{\omega_{k}}}\right] .

We can now do the sum on j using the result from part (a), and that gives the first commutator. Next we take

\left[a_{k-}, a_{k^{\prime}-}\right]=\frac{1}{N} \sum_{j, j^{\prime}} e^{-i 2 \pi\left(j^{\prime} k^{\prime}+j k\right) / N} \frac{1}{2}\left\{\sqrt{\frac{\omega_{k}}{\omega_{k^{\prime}}}}\left[x_{j}, \frac{\partial}{\partial x_{j^{\prime}}}\right]+\sqrt{\frac{\omega_{k}^{\prime}}{\omega_{k}}}\left[\frac{\partial}{\partial x_{j}}, x_{j^{\prime}}\right]\right\} .

Evaluating the two commutators we have

\left[a_{k-}, a_{k^{\prime}-}\right]=\left(\frac{1}{N} \sum_{j} e^{-i 2 \pi j\left(k^{\prime}+k\right) / N}\right) \frac{1}{2}\left[-\sqrt{\frac{\omega_{k}}{\omega_{k^{\prime}}}}+\sqrt{\frac{\omega_{k}^{\prime}}{\omega_{k}}}\right] .

Doing the sum on j now gives the result that k′= N – k (or k′ = k if k = N) but since \omega_{N-k}=\omega_{k} this vanishes, proving the second result. The final commutator works out the same way.

(c)

\left(a_{k_{-}}+a_{N-k_{+}}\right)=\frac{1}{\sqrt{N}} \sum_{j=1}^{N}\left\{e^{-i 2 \pi j k / N}\left[\sqrt{\frac{m \omega_{k}}{2 \hbar}} x_{j}+\sqrt{\frac{\hbar}{2 m \omega_{k}}} \frac{\partial}{\partial x_{j}}\right]\right.

\left.+e^{i 2 \pi j(N-k) / N}\left[\sqrt{\frac{m \omega_{N-k}}{2 \hbar}} x_{j}-\sqrt{\frac{\hbar}{2 m \omega_{N-k}}} \frac{\partial}{\partial x_{j}}\right]\right\} .

\text { But } \omega_{N-k}=\omega_{k}, \text { and } e^{i 2 \pi j(N-k) / N}=e^{i 2 \pi j} e^{-i 2 \pi j k / N}=e^{-i 2 \pi j k / N} ,  so

\left(a_{k_{-}}+a_{N-k_{+}}\right)=\frac{2}{\sqrt{N}} \sum_{j=1}^{N} \sqrt{\frac{m \omega_{k}}{2 \hbar}} e^{-i 2 \pi j k / N} x_{j} .

\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{\hbar}{2 m \omega_{k}}}\left(a_{k_{-}}+a_{N-k_{+}}\right) e^{i 2 \pi j k / N}=\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{\hbar}{2 m \omega_{k}}} \frac{2}{\sqrt{N}} \sum_{j^{\prime}=1}^{N} \sqrt{\frac{m \omega_{k}}{2 \hbar}} e^{-i 2 \pi j^{\prime} k / N} x_{j^{\prime}} e^{i 2 \pi j k / N} .

=\frac{1}{N} \sum_{j^{\prime}=1}^{N} \sum_{k=1}^{N-1} e^{i 2 \pi\left(j-j^{\prime}\right) k / N} x_{j^{\prime}}=\frac{1}{N} \sum_{j^{\prime}=1}^{N}\left(\sum_{k=1}^{N} e^{i 2 \pi\left(j-j^{\prime}\right) k / N}-e^{i 2 \pi\left(j-j^{\prime}\right)}\right) x_{j^{\prime}} .

=\sum_{j^{\prime}=1}^{N}\left(\frac{1}{N} \sum_{k=1}^{N} e^{i 2 \pi\left(j-j^{\prime}\right) k / N}\right) x_{j^{\prime}}-\frac{1}{N} \sum_{j^{\prime}=1}^{N} x_{j^{\prime}}=\sum_{j^{\prime}=1}^{N} \delta_{j j^{\prime}} x_{j^{\prime}}-R=x_{j}-R .

That proves the first relation; to prove the second, we repeat these steps, starting with

\left(a_{k_{-}}-a_{N-k_{+}}\right)=\frac{1}{\sqrt{N}} \sum_{j=1}^{N}\left\{e^{-i 2 \pi j k / N}\left[\sqrt{\frac{m \omega_{k}}{2 \hbar}} x_{j}+\sqrt{\frac{\hbar}{2 m \omega_{k}}} \frac{\partial}{\partial x_{j}}\right]\right.

\left.-e^{i 2 \pi j(N-k) / N}\left[\sqrt{\frac{m \omega_{N-k}}{2 \hbar}} x_{j}-\sqrt{\frac{\hbar}{2 m \omega_{N-k}}} \frac{\partial}{\partial x_{j}}\right]\right\} .

=\frac{2}{\sqrt{N}} \sum_{j=1}^{N} \sqrt{\frac{\hbar}{2 m \omega_{k}}} e^{-i 2 \pi j k / N} \frac{\partial}{\partial x_{j}} .

\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{m \omega_{k}}{2 \hbar}}\left(a_{k_{-}}-a_{N-k_{+}}\right) e^{i 2 \pi j k / N}=\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{m \omega_{k}}{2 \hbar}} \frac{2}{\sqrt{N}} \sum_{j^{\prime}=1}^{N} \sqrt{\frac{\hbar}{2 m \omega_{k}}} e^{-i 2 \pi j^{\prime} k / N} \frac{\partial}{\partial x_{j^{\prime}}} e^{i 2 \pi j k / N} .

=\frac{1}{N} \sum_{j^{\prime}=1}^{N} \sum_{k=1}^{N-1} e^{i 2 \pi\left(j-j^{\prime}\right) k / N} \frac{\partial}{\partial x_{j^{\prime}}}=\frac{1}{N} \sum_{j^{\prime}=1}^{N}\left(\sum_{k=1}^{N} e^{i 2 \pi\left(j-j^{\prime}\right) k / N}-e^{i 2 \pi\left(j-j^{\prime}\right)}\right) \frac{\partial}{\partial x_{j^{\prime}}} .

=\sum_{j^{\prime}=1}^{N}\left(\frac{1}{N} \sum_{k=1}^{N} e^{i 2 \pi\left(j-j^{\prime}\right) k / N}\right) \frac{\partial}{\partial x_{j^{\prime}}}-\frac{1}{N} \sum_{j^{\prime}=1}^{N} \frac{\partial}{\partial x_{j^{\prime}}}=\sum_{j^{\prime}=1}^{N} \delta_{j j^{\prime}} \frac{\partial}{\partial x_{j^{\prime}}}-\frac{1}{N} \sum_{j^{\prime}=1}^{N} \frac{\partial}{\partial x_{j^{\prime}}}=\frac{\partial}{\partial x_{j}}-\frac{1}{N} \frac{\partial}{\partial R} .

In the last step I used \frac{\partial}{\partial R}=\sum_{j=1}^{N} \frac{\partial x_{j}}{\partial R} \frac{\partial}{\partial x_{j}}, \text { and } \frac{\partial x_{j}}{\partial R}=1   (for instance, if N = 2; R=\frac{1}{2}\left(x_{1}+x_{2}\right) , and r=\frac{1}{2}\left(x_{1}-x_{2}\right) \text {, so } x_{1}=R+r \text { and } x_{2}=R-r \Rightarrow \partial x_{i} / \partial R=1 \text { ) } .

(d)

\left(x_{j+1}-x_{j}\right)=\frac{1}{\sqrt{N}} \sum_{k=1}^{N-1} \sqrt{\frac{\hbar}{2 m \omega_{k}}}\left(a_{k_{-}}+a_{N-k_{+}}\right) e^{i 2 \pi j k / N}\left(e^{i 2 \pi k / N}-1\right) .

\left(x_{j+1}-x_{j}\right)^{2}=\frac{1}{N} \sum_{k, k^{\prime}}^{N-1} \frac{\hbar}{2 m \sqrt{\omega_{k} \omega_{k^{\prime}}}}\left(a_{k_{-}}+a_{N-k_{+}}\right)\left(a_{k_{-}^{\prime}}+a_{N-k^{\prime}+}\right) e^{i 2 \pi j\left(k+k^{\prime}\right) / N}

\times\left(e^{i 2 \pi k / N}-1\right)\left(e^{i 2 \pi k^{\prime} / N}-1\right) ,

\sum_{j=1}^{N} \frac{1}{2} m \omega^{2}\left(j+1-x_{j}\right)^{2}=\sum_{k, k^{\prime}}^{N-1} \frac{\hbar \omega^{2}}{4 \sqrt{\omega_{k} \omega_{k^{\prime}}}}\left(a_{k_{-}}+a_{N-k_{+}}\right)\left(a_{k_{-}^{\prime}}+a_{N-k^{\prime}+}\right) \delta_{k^{\prime}, N-k}

\times\left(e^{i 2 \pi k / N}-1\right)\left(e^{i 2 \pi k^{\prime} / N}-1\right) .

=\sum_{k=1}^{N-1} \frac{\hbar \omega^{2}}{4 \omega_{k}}\left(a_{k_{-}}+a_{N-k_{+}}\right)\left(a_{N-k_{-}}+a_{k_{+}}\right) \frac{\omega_{k}^{2}}{\omega^{2}} .

=\sum_{k=1}^{N-1} \frac{\hbar \omega_{k}}{4}\left(a_{k_{-}} a_{k+}+a_{N-k_{+}} a_{k_{+}}+a_{k_{-}} a_{N-k_{-}}+a_{N-k_{+}} a_{N-k_{-}}\right) .

In the penultimate line I used

\left(e^{i 2 \pi k / N}-1\right)\left(e^{i 2 \pi(N-k) / N}-1\right)=e^{i \pi k / N}\left(e^{i \pi k / N}-e^{-i \pi k / N}\right) e^{-i \pi k / N}\left(e^{-i \pi k / N}-e^{i \pi k / N}\right) .

=(2 i \sin (\pi k / N))(-2 i \sin (\pi k / N))=[2 \sin (\pi k / N)]^{2}=\left(\frac{\omega_{k}}{\omega}\right)^{2} .

Meanwhile,

-\frac{\hbar^{2}}{2 m} \frac{\partial^{2}}{\partial x_{j}^{2}}=\frac{1}{N} \sum_{k, k^{\prime}}^{N-1}\left(-\frac{\hbar^{2}}{2 m}\right) \frac{m}{2 \hbar} \sqrt{\omega_{k} \omega_{k^{\prime}}}\left(a_{k_{-}}-a_{N-k_{+}}\right)\left(a_{k_{-}^{\prime}}-a_{N-k^{\prime}+}\right) e^{i 2 \pi j\left(k+k^{\prime}\right) / N} -\frac{\hbar^{2}}{2 m} \frac{1}{N^{3 / 2}}\left[\frac{\partial}{\partial R} \sum_{k=1}^{N-1} \sqrt{\frac{m \omega_{k}}{2 \hbar}}\left(a_{k_{-}}-a_{N-k_{+}}\right) e^{i 2 \pi j k / N}\right.

\left.+\sum_{k=1}^{N-1} \sqrt{\frac{m \omega_{k}}{2 \hbar}}\left(a_{k_{-}}-a_{N-k_{+}}\right) e^{i 2 \pi j k / N} \frac{\partial}{\partial R}\right]-\frac{\hbar^{2}}{2 m} \frac{1}{N^{2}} \frac{\partial^{2}}{\partial R^{2}} .

The middle terms vanish when summed over j \left(\sum_{j=1}^{N} e^{i 2 \pi j k / N}=N \delta_{k, 0}=0\right) , leaving

-\frac{\hbar^{2}}{2 m} \sum_{j=1}^{N} \frac{\partial^{2}}{\partial x_{j}^{2}}=\sum_{k=1}^{N-1}\left(-\frac{\hbar \omega_{k}}{4}\right)\left(a_{k_{-}}-a_{N-k_{+}}\right)\left(a_{N-k_{-}}-a_{k_{+}}\right)-\frac{\hbar^{2}}{2 N m} \frac{\partial^{2}}{\partial R^{2}} .

=\sum_{k=1}^{N-1} \frac{\hbar \omega_{k}}{4}\left(a_{k_{-}} a_{k_{+}}-a_{N-k_{+}} a_{k_{+}}-a_{k_{-}} a_{N-k_{-}}+a_{N-k_{+}} a_{N-k_{-}}\right)-\frac{\hbar^{2}}{2 N m} \frac{\partial^{2}}{\partial R^{2}} .

Adding the two results:

\hat{H}=-\frac{\hbar^{2}}{2 N m} \frac{\partial^{2}}{\partial R^{2}}+\sum_{k=1}^{N-1} \frac{\hbar \omega_{k}}{2}\left(a_{k_{-}} a_{k_{+}}+a_{N-k_{+}} a_{N-k_{-}}\right)=-\frac{\hbar^{2}}{2 N m} \frac{\partial^{2}}{\partial R^{2}}+\sum_{k=1}^{N-1} \frac{\hbar \omega_{k}}{2}\left(a_{k_{-}} a_{k_{+}}+a_{k_{+}} a_{k_{-}}\right) .

(the change in the final two subscripts just means adding these terms in reverse order, given that \omega_{N-k}=\omega_{k} ).

Thus

\hat{H}=-\frac{\hbar^{2}}{2(N m)} \frac{\partial^{2}}{\partial R^{2}}+\sum_{k=1}^{N-1} \hbar \omega_{k}\left(a_{k_{+}} a_{k_{-}}+\frac{1}{2}\right) .

Related Answered Questions