Interpreting group velocity of free particle wave packet

I am trying to understand the concept of group velocity of a free particle wave packet: $$\Psi(x,t) = \frac{1}{\sqrt{2 \pi}}\int_{-\infty}^{\infty} \phi(k)e^{ikx}e^{-\frac{i \hbar k^2 t}{2m}}dk.$$

Where $$\omega(k) = \frac{\hbar k^2}{2m}.$$ Assuming that $\phi(k)$ is narrowly peaked about some particular value $k_0$ we can Taylor expand $\omega(k)$ about $k_0$ to get $$\omega(k) = \frac{\hbar k_0}{2m} + \frac{\hbar k_0}{m}(k-k_0)+ \frac{\hbar}{2m}(k-k_0)^2.$$ Then after defining variable $\kappa := k-k_0$ we get $$\Psi(x,t) \approx \frac{1}{\sqrt{2 \pi}}e^{\frac{-i \hbar k_0^2 t}{2m}}e^{\frac{i \hbar k_0^2 t}{m}} \int^{\infty}_{-\infty}\phi(\kappa + k_0)e^{i(\kappa + k_0)(x-\frac{\hbar k_0 t}{m})}dk.$$

This integral is the superposition of waves of the form $$e^{i(\kappa + k_0)(x-\frac{\hbar k_0 t}{m})}.$$ However notice that each of these waves have the same speed $\frac{\hbar k_0}{m}$. Then in terms of the energy $$v_{\text{group}} = \frac{d \omega(k_0)}{dk} = \frac{\hbar k_0}{m} = \sqrt{\frac{2 E_0}{m}}.$$ Hence for stationary state $\Psi_{k_0}$ we have that $$v_{\text{group}} = 2 v_{\text{phase}}.$$

Apparently we can then evaluate the group velocity for each $k$, hence $v_{\text{group}} = \frac{d \omega(k)}{dk}$.

Question: I understand the above calculation. But qualitatively I don't understand how the group velocity can be a function of $k$ as opposed to just a single value (independent of $k$) which describes how fast the envelope propagates? I know that each stationary state has it's own phase velocity but I thought that superimposing these waves of different speeds produces an envelope which travels at a single velocity $v_{\text{group}}$?

• Why would you expect it to be independent of $k$? btw Griffiths also gives an example of water waves, also this GIF would probably help. – Oswald May 6 '16 at 4:21
• @TheGhostOfPerdition Independent of $k$ was probably not the right way to phrase it. What I meant was that I expected the $v_{\text{group}}$ to be some scalar value rather than a function. In your animation, would you say that the dark blue wave which I'm assuming is the envelope is moving at one specific scalar valued speed? – user100411 May 6 '16 at 11:59
• @JohnDoe - When the group velocity is not a constant 3-vector but a function of $\mathbf{k}$ and/or $\omega$, then the mode is said to be dispersive. http://physics.stackexchange.com/a/143717/59023 – honeste_vivere May 6 '16 at 14:46

A qualitative understanding can be gained from the animation provided by 'TheGhostOfPerdition', each small peak on the dark blue wave (envelope of the wave packet) has an associated $k_{0}$ for which $\phi(k)$ peaks at $k = k_0$. So you can work out the velocity $v_g : = \frac{d \omega}{d k}|_{k = k_0}$ of the dark blue (envelope of wave packet) using the scheme that you outlined, you have thus found the group velocity for that peak. The velocity might differ for another peak. When there is distortion (the component wavelengths move at different speeds), you could get a envelope of the wave packet which changes shape as it moves as different points move at different speeds.