Quantum Mechanics

\(\let\vec\mathbf\) \(\newcommand{\v}{\vec}\) \(\newcommand{\d}{\dot{}}\) \(\newcommand{\del}{\nabla}\) \(\newcommand{\abs}{\rvert}\) \(\newcommand{\h}{\hat}\) \(\newcommand{\f}{\frac}\) \(\newcommand{\~}{\widetilde}\) \(\newcommand{\<}{\langle}\) \(\newcommand{\>}{\rangle}\) \(\newcommand{\eo}{\epsilon_{0}}\) \(\newcommand{\hb}{\hbar}\) \(\newcommand{\pd}{\partial}\) \(\newcommand{\h}{\hat}\) \(\newcommand{\ket}[1]{\lvert #1 \rangle}\)

Postulates

1st Postulate

The state of a quantum mechanical system is described by a wavefunction \(\Psi(\v{r},t)\), which is mathematically interpreted as a state vector \(\ket{\Psi}\) that belongs to a state space called the Hilbert space and is a function of position \(\v{r}\) and time \(t\). For an isolated system, the probability of finding the particle somewhere must be equal to one.

\[\int_{-\infty}^{+\infty} \Psi^{*}(\v{r},t) \Psi(\v{r},t) d\tau = 1\]

The term \(\Psi^{*}(\v{r},t) \Psi(\v{r},t) d\tau\) defines the probability of finding the particle in volume \(d\tau\) at \(\v{r}\) and \(t\).

2nd Postulate

Physical observables are represented as operators in quantum mechanics. Operators have the property of being linear and Hermitian. In other words, a physical quantity \(A\) is described by a Hermitian operator \(\h{A}\) acting in state space \(H\), forming a basis for \(H\). The result of measuring \(A\) results in

3rd Postulate

\[\h{A}\Psi = a\Psi\] \[\Psi = \sum_{i = 1}^{n} c_{i} \Psi_{i}\]

4th Postulate

\[\< \h{A} \> = \int_{-\infty}^{+\infty} \Psi^{*} \h{A} \Psi d\tau\]

5th Postulate

The wavefunction or state function evolves in time according to the time independent Schrodinger equation:

\[i\hbar \f{\pd \Psi}{\pd t} = \h{H} \Psi\]

Time Independent Schrodinger Equation

For a given potential energy function \(V(x,t)\), the Schrodinger equation, independent of \(t\) is written as:

\[i \hb \f{\pd \Psi}{\pd t} = - \f{\hb^{2}}{2m} \f{\pd^{2} \Psi}{\pd x^{2} } + V \Psi\]

Infinite Square Well

Consider a situation where a particle is trapped in a potential well with infinitely high walls. The potential is described mathematically as,

\[V(x) = \left\{ \begin{array}{ll} 0 & 0 \leq x \leq a, \\ \infty & \verb|otherwise| \\ \end{array} \right.\]

The time indpendent Schrodinger equation can be set up using known conditions of the system.

The Schrodinger equation then takes the form,

\[- \f{\hb^{2}}{2m} \f{d^{2} \psi}{dx^{2}} = E \psi\]

We can divide through by \(-\f{\hb^{2}}{2m}\) and where \(k \equiv \f{\sqrt{2mE}}{\hb}\)

\[\f{d^{2} \psi}{dx^{2}} = - k^{2} \psi\]

The general solution this differential equation is,

\[\psi(x) = A\sin{kx} + B\cos{kx}\]

With both the wave function, \(\psi\), and its derivative ,\(\f{d\psi}{dx}\), being continuous, this just means that both are differentiable.

Applying the boundary conditions, the continuity of \(\psi(x)\) require that

\[\psi(0) = \psi(a) = 0\]

Thus at \(\psi(0)\),

\[\psi(0) = 0 = A \sin{0} + B\cos{0} = B\]

Plugging \(B = 0\) back into the general solution yields

\[\psi(x) = A \sin{kx}\]

At \(\psi(a)\), we come across two cases in order to satify \(\psi(a) = 0\). The wavefunction at \(a\) is,

\[\psi(a) = A \sin{ka}\]

The two cases being,

The latter is only true for certain values of \(ka\),

\[ka = 0, \pm \pi, \pm 2\pi, \pm 3\pi, ...\]

Again, we come across another “bad” solution, \(k = 0\), since this would imply that \(\psi(x) = 0\). The minus sign from negative solutions such as \(\sin(-\theta) = -\sin(\theta)\) can be absorbed into the constant \(A\).

Thus, we arrive at distinct solutions where,

\[k_{n} = \f{n\pi}{a}\]

with values of \(n = 1, 2, 3, ...\). Combining this with definition of \(k\) in terms of energy \(E\),

\[E_{n} = \f{\hbar^{2} k_{n}^{2}}{2m} = \f{n^{2} \pi^{2} \hb^{2}}{2ma^{2}}\]

To interpret this physically, the energy \(E_{n}\) can only take on discrete values of \(n\). We also attach the subscript \(n\) to denote that \(k_{n}\) represents the set of allowed energies of the wavefunction, which will be important to evaluating the upcoming integral.

However, \(A\) is still a missing part of the equation, which is determined by \(k\). We can solve for \(A\) by normalizing \(\psi\).

With the help of a trigonometric identity: \(\sin^{2} (\theta) = \f{1-\cos(2\theta)}{2}\)…

\[\int_{0}^{a} \abs A \abs^{2} \sin^{2}(k_{n} x) dx = \abs A \abs^{2} \int_{0}^{a} \f{1- \cos(2k_{n} x)}{2} dx = 1\]

Using u-subsitution, let \(u=2k_{n} x\), where \(du = 2k_{n} dx\),

\[\abs A \abs^{2} \bigg[ \f{x}{2} \bigg|_{0}^{a} - \int_{0}^{a} \f{\cos{u}}{2} \f{du}{2} \bigg] = 1\] \[\abs A \abs^{2} \bigg[ \f{x}{2} - \f{\sin{2k_{n} x}}{4k} \bigg]_{0}^{a} = 1\]

Now it is important to recall that \(k= \f{n \pi}{a}\). \(k\) is restricted by \(n = 1,2,3,...\) and for this reason, the \(\sin2kx\) term is zero at all values of \(n\). Now simply solving for \(A\) will yield the magnitude of \(A\), picking the positive solution.

\[\abs A \abs^{2} \f{a}{2} = 1\] \[A = \sqrt{\f{2}{a}}\]

Thus, the wavefunction solutions inside the well is,

\[\psi_{n} (x) = \sqrt{\f{2}{a}} \sin\bigg(\f{n\pi x}{a}\bigg)\]

Note that the solutions alternate between even and odd between the intervals [0, a] . \(\psi_{1}\) is even, \(\psi_{2}\) is odd, \(\psi_{3}\) is even. This also related to how many nodes, or zero crossings, for successive states \(\psi_{n}\). The wave function that carries the lowest energy ,\(n=1\), is the ground state, whereas wavefunctions of energies proportional to \(n^{3}\) are the excited states.

Due to the \(\sin\) terms in states \(\psi_{n} (x)\) end up being mutually orthogonal. In other words,

\[\int \psi_{m} (x)^{*} \psi_{n}(x) dx = 0\]

when \(m \neq n\).

Orthogonality:

\[\int \psi_{m} (x)^{*} \psi_{n}(x) dx = \f{2}{a} \int_{0}^{a} \sin{\bigg( \f{m\pi x}{a} \bigg)} \sin{\bigg( \f{n\pi x}{a} \bigg)} dx\]

Using a product-to-sum trigonometric identity: \(\sin(\alpha)\sin(\beta) = \f{1}{2} [\cos(\alpha - \beta) - \cos(\alpha + \beta)\)

\(= \f{1}{a} \int_{0}^{a} \cos{\bigg(\f{ (m-n) \pi }{a} \bigg)} - \cos{\bigg(\f{ (m+n) \pi }{a} \bigg)} dx\) \(= \f{1}{a} \bigg[ \f{1}{ (m-n) \pi } \sin{\bigg(\f{ (m-n) \pi }{a} \bigg)} - \f{1}{(m+n) \pi} \sin{\bigg(\f{ (m+n) \pi }{a} \bigg)} \bigg] \bigg|_{0}^{a}\)

When evaluating from \(0\) to \(a\), it should be clear to see that only \(a\) survives, leaving

\(= \f{1}{\pi} \bigg[ \f{ \sin{\bigg(\f{ (m-n) \pi }{a} \bigg)}}{(m-n)} - \f{\sin{\bigg(\f{ (m+n) \pi }{a} \bigg)}}{(m+n)} \bigg] = 0\) (for \(m\neq n\))

Rewritten in terms of the Kronecker Delta,

\[\int \psi_{m} (x)^{*} \psi_{n}(x) dx = \delta_{mn}\]

Electron in a Vacuum

Consider an electron in a vacuum, without any external influences (i.e. electromagnetic fields). The linear momentum operator is:

\[- i \hbar \nabla \psi = \v{p} \psi\]

with \(\v{p}\) being the associated eigenvalues. The wavefunction of an electron in a vacuum is:

\[\psi = e^{i(\v{k}\cdot \v{r} - \omega t)}\]

Expanding \(\v{k}\), \(\v{r}\), and \(\nabla\) into their linear components gives,

\[- i \hbar \bigg( \f{\pd{}}{\pd{x}} \h{i} + \f{\pd{}}{\pd{y}} \h{j} + \f{\pd{}}{\pd{z}} \h{k} \bigg) e^{i (k_{x}x + k_{y} y + k_{z} z)} = \v{p} e^{i(\v{k} \cdot {r} - \omega t)}\]

The momentum operator acting on \(\psi\) demonstrates that the eigenvalue \(\v{p}\) is,

\[\v{p} = \hbar \bigg( k_{x} \h{i} + k_{y} \h{j} + k_{z} \h{k} \bigg) = \hbar \v{k}\]

Using some basic equations from wave mechanics,

\[\lambda = \f{h}{p}\] \[k = | \v{k} | = \f{2\pi}{\lambda}\]

It shows that,

\[p = \hbar k = \bigg(\f{h}{2\pi} \bigg) \bigg(\f{2\pi}{\lambda} \bigg)\]

Quantum Fluxes

Many of the following derivations are rooted in the continuity equation,

\[\frac{\pd \rho_{P}}{\pd t} + \del \cdot \v{J}_{P} = 0\] \[\frac{\pd \rho_{M}}{\pd t} + \del \v{J}_{M} = 0\] \[\frac{\pd \rho_{E}}{\pd t} + \del \v{J}_{E} = 0\]

where \(P\), \(M\), and \(E\) represent probability, momentum, and energy currents.

Before deriving flux quantities, it will be useful to mention the following relationships and identities:

\[\frac{\pd \psi}{\pd t} = \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \del^{2} \psi + V \psi \bigg]\]

The complex conjugate of \(\frac{\pd \psi}{\pd t}\) is simply,

\[\frac{\pd \psi^{*}}{\pd t} = \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \del^{2} \psi^{*} + V \psi^{*} \bigg]\]

Moreover, the two operator identities that will be of use,

\[\del \cdot (\psi^{*} \del \psi) = \del \psi^{*} \cdot \del \psi - \psi^{*} \cdot \del^{2} \psi\] \[\del \cdot (\psi \del\psi^{*}) = \del \psi \cdot \del \psi^{*} - \psi \cdot \del^{2} \psi^{*}\]

Lastly, we define the “density”, \(\rho\), with units of “amount per unit volume” as,

\[\rho = \psi^{*} \psi\]

Probability Flux

The probability density given as,

\[\rho_{P} (\v{r}, t) = \psi^{*}(\v{r},t) \psi(\v{r},t)\]

take the partial with respect to time,

\[\f{\pd \rho_{P}}{\pd t} = \bigg( \f{\pd \psi^{*}}{\pd t} \bigg) \psi + \psi^{*} \bigg( \f{\pd \psi }{\pd t} \bigg)\]

we plug the expressions of \(\f{\pd \psi}{\pd t}\) and its complex conjugate defined above into the equation,

\[\f{\pd \rho_{P}}{\pd t} = \f{\pd \psi^{*}\psi}{\pd t} = \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \del^{2} \psi^{*} + V \psi^{*} \bigg] \psi + \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \del^{2} \psi + V \psi \bigg] \psi^{*}\] \[= -\f{\hbar}{2mi} (\del^{2} \psi) \psi^{*} + V \psi \psi^{*} + \f{\hbar}{2mi} (\del^{2} \psi^{*} ) \psi - V \psi^{*} \psi\] \[= -\f{\hbar}{2mi} \bigg( (\del^{2} \psi )\psi^{*} + (\del^{2} \psi^{*} )\psi \bigg)\] \[\f{\pd \rho_{P}}{\pd t} = -\f{\hbar}{2mi} \del \bigg( (\del \psi )\psi^{*} + (\del \psi^{*} )\psi \bigg)\]

Now that we have an expression for \(\f{\pd \rho_{P}}{\pd t}\), we can apply the continuity equation, which is also a statement of conservation,

\[\f{\pd \rho_{P}}{\pd t} + \f{\hbar}{2mi} \del \bigg( (\del \psi )\psi^{*} + (\del \psi^{*} )\psi \bigg) = 0\]

Thus, the second term in the right hand side of the equation tells us that the probability current is,

\[\v{J}_{P} = \f{\hbar}{2mi} \bigg( (\del \psi )\psi^{*} + (\del \psi^{*} )\psi \bigg)\]

Momentum Current

In a similiar fashion, we can also derive the “momentum current”, \(j_{M}\), given a momentum density,

\[\rho_{M} (\v{r}, t) = \Re{ (\psi^{*} (\v{r},t) \h{p} \psi(\v{r},t) )}\]

Again, we take the partial derivative with respect to time,

\[\f{\pd \rho_{M}}{\pd t} = \bigg( \f{\pd \psi^{*}}{\pd t} \bigg) \h{p} \psi + \psi^{*} \h{p} \bigg( \f{\pd \psi }{\pd t} \bigg)\]

where the momentum operator is \(\h{p} = -i \hbar \del\).

We substitute the definitions of \(\f{\pd \psi^{*}}{\pd t}\) and \(\f{\pd \psi}{\pd t}\) from above. Keep in mind that we are only considering the real component of the entire expression (the momentum operator happens to be complex).

\[= \Re \biggl\{ \f{1}{i\hbar} \bigg[-\f{\hbar^{2}}{2m} \del^{2} \psi^{*} \h{p} \psi + V \psi^{*} \h{p} \psi \bigg] \biggr\} + \Re \biggl\{ \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \psi^{*} \h{p} \del^{2} \psi + \psi^{*} \h{p} V \psi \bigg] \biggr\}\]

substituting the definition of the momentum operator,

\[\f{\pd \rho_{M}}{\pd t} = \Re \biggl\{ \f{\hbar^{2}}{2m} \del^{2} \psi^{*} \del \psi - i\hbar V \psi^{*} \del \psi + \f{\hbar^{2}}{2m} \psi^{*} \del(\del^{2} \psi) - i\hbar \psi^{*} \del V \psi \biggr\}\]

and grouping terms involving potential together,

\[= \Re \biggl\{ \f{\hbar^{2}}{2m} \bigg( \del^{2} \psi^{*} \del \psi + \psi^{*} \del (\del^{2} \psi ) \bigg) - \bigg( i \hbar ( V \psi^{*} \del \psi + \psi^{*} \del (V \psi) ) \bigg) \biggr\}\]

Again, we apply the continuity equation,

\[\f{\pd \rho_{M}}{\pd t} + \f{\hbar^{2}}{2m} \bigg( \del^{2} \psi^{*} \del \psi + \psi^{*} \del (\del^{2} \psi ) \bigg) = V \psi^{*} \del \psi + \psi^{*} \del (V\psi)\]

This introduces a source potential,

\[- \rho_{P} \del V = V \psi^{*} \del \psi + \psi^{*} \del (V\psi)\]

The continuity equation for momentum takes the form of a momentum conservation law,

\[\f{\pd \rho_{M}}{\pd t} + \f{\hbar^{2}}{2m} \del \bigg( \del \psi^{*} \del \psi + \psi^{*} (\del^{2} \psi ) \bigg) = - \rho_{P} \del V\]

From here, we find that the momentum current is,

\[j_{M} = \f{\hbar^{2}}{2m} \bigg( \del \psi^{*} \del \psi + \psi^{*} \del^{2} \psi \bigg)\]

Energy Current

Energy desnity is given by:

\[\rho_{E} (\v{r}, t) = \Re (\psi^{*} (\v{r}, t) \h{H} \psi (\v{r}, t))\]

Much like the taking the time derivative of the energy density, \(\rho_{E}\), with the Hamiltonian operator gives

\[\f{\pd \rho_{M}}{\pd t} = \bigg( \f{\pd \psi^{*}}{\pd t} \bigg) \h{H} \psi + \psi^{*} \h{H} \bigg( \f{\pd \psi }{\pd t} \bigg)\]

The Hamiltonian when operating on a wavefunction is,

\[\h{H} = i \hbar \f{\pd \psi}{\pd t}\]

plugging \(\f{\pd \psi}{\pd t}\) and \(\f{\pd \psi^{*}}{\pd t}\) terms,

\[= \Re \biggl\{ \f{1}{i\hbar} \bigg[-\f{\hbar^{2}}{2m} \del^{2} \psi^{*} \h{H} \psi + V \psi^{*} \h{H} \psi \bigg] \biggr\} + \Re \biggl\{ \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \psi^{*} \h{H} \del^{2} \psi + \psi^{*} \h{H} V \psi \bigg] \biggr\}\]

Applying the definition of the Hamiltonian,

\[= \Re \biggl\{ \f{1}{i\hbar} \bigg[-\f{\hbar^{2}}{2m} \del^{2} \psi^{*} i\hbar\f{\pd \psi}{\pd t}+ V \psi^{*} i\hbar \f{\pd \psi}{\pd t} \bigg] \biggr\} + \Re \biggl\{ \f{1}{i\hbar} \bigg[ -\f{\hbar^{2}}{2m} \psi^{*} i\hbar\f{\pd}{\pd t} \del^{2} \psi + \psi^{*} i\hbar \f{\pd}{\pd t} V \psi \bigg] \biggr\}\]

Rearranging for terms involving potentials,

\[= -\f{\hbar^{2}}{2m} \bigg[ \psi^{*} \del^{2} \f{\pd \psi}{\pd t} - \f{\pd \psi}{\pd t} \del^{2} \psi^{*} \bigg] + \bigg[ V \psi^{*} \frac{\pd \psi}{\pd t} + \psi^{*} V \f{\pd \psi}{\pd t} \bigg]\] \[\v{j}_{E} = \f{\hbar^{2}}{2m} \Re \biggl\{ \psi^{*} \del \bigg( \f{\pd \psi}{\pd t}\bigg) - \f{\pd \psi}{\pd t} \del \psi^{*} \biggr\}\]