Categories: Physics, Plasma physics.

Rutherford scattering

Rutherford scattering or Coulomb scattering is an elastic pseudo-collision of two electrically charged particles. It is not a true collision, and is caused by Coulomb repulsion.

The general idea is illustrated below. Consider two particles 1 and 2, with the same charge sign. Let 2 be initially at rest, and 1 approach it with velocity v1\vb{v}_1. Coulomb repulsion causes 1 to deflect by an angle θ\theta, and pushes 2 away in the process:

Two-body repulsive 'collision'

Here, bb is called the impact parameter. Intuitively, we expect θ\theta to be larger for smaller bb.

By combining Coulomb’s law with Newton’s laws, these particles’ equations of motion are found to be as follows, where r=r1r2r = |\vb{r}_1 - \vb{r}_2| is the distance between 1 and 2:

m1dv1dt=F1=q1q24πε0r1r2r3m2dv2dt=F2=F1\begin{aligned} m_1 \dv{\vb{v}_1}{t} = \vb{F}_1 = \frac{q_1 q_2}{4 \pi \varepsilon_0} \frac{\vb{r}_1 - \vb{r}_2}{r^3} \qquad \quad m_2 \dv{\vb{v}_2}{t} = \vb{F}_2 = - \vb{F}_1 \end{aligned}

Using the reduced mass μm1m2/(m1 ⁣+ ⁣m2)\mu \equiv m_1 m_2 / (m_1 \!+\! m_2), we turn this into a one-body problem:

μdvdt=q1q24πε0rr3\begin{aligned} \mu \dv{\vb{v}}{t} = \frac{q_1 q_2}{4 \pi \varepsilon_0} \frac{\vb{r}}{r^3} \end{aligned}

Where vv1 ⁣ ⁣v2\vb{v} \equiv \vb{v}_1 \!-\! \vb{v}_2 is the relative velocity, and rr1 ⁣ ⁣r2\vb{r} \equiv \vb{r}_1 \!-\! \vb{r}_2 is the relative position. The latter is as follows in cylindrical polar coordinates (r,φ,z)(r, \varphi, z):

r=rcosφe^x+rsinφe^y+ze^z=re^r+ze^z\begin{aligned} \vb{r} = r \cos{\varphi} \:\vu{e}_x + r \sin{\varphi} \:\vu{e}_y + z \:\vu{e}_z = r \:\vu{e}_r + z \:\vu{e}_z \end{aligned}

These new coordinates are sketched below, where the origin represents r1=r2\vb{r}_1 = \vb{r}_2. Crucially, note the symmetry: if the “collision” occurs at t=0t = 0, then by comparing t>0t > 0 and t<0t < 0 we can see that vxv_x is unchanged for any given ±t\pm t, while vyv_y simply changes sign:

Equivalent one-body deflection

From our expression for r\vb{r}, we can find v\vb{v} by differentiating with respect to time:

v=(rcosφrφsinφ)e^x+(rsinφ+rφcosφ)e^y+ze^z=r(cosφe^x+sinφe^y)+rφ( ⁣ ⁣sinφe^x+cosφe^y)+ze^z=re^r+rφe^φ+ze^z\begin{aligned} \vb{v} &= \big( r' \cos{\varphi} - r \varphi' \sin{\varphi} \big) \:\vu{e}_x + \big( r' \sin{\varphi} + r \varphi' \cos{\varphi} \big) \:\vu{e}_y + z' \:\vu{e}_z \\ &= r' \: \big( \cos{\varphi} \:\vu{e}_x + \sin{\varphi} \:\vu{e}_y \big) + r \varphi' \: \big( \!-\! \sin{\varphi} \:\vu{e}_x + \cos{\varphi} \:\vu{e}_y \big) + z' \:\vu{e}_z \\ &= r' \:\vu{e}_r + r \varphi' \:\vu{e}_\varphi + z' \:\vu{e}_z \end{aligned}

Where we have recognized the basis vectors e^r\vu{e}_r and e^φ\vu{e}_\varphi. If we choose the coordinate system such that all dynamics are in the (x,y)(x,y)-plane, i.e. z(t)=0z(t) = 0, we have:

r=re^rv=re^r+rφe^φ\begin{aligned} \vb{r} = r \: \vu{e}_r \qquad \qquad \vb{v} = r' \:\vu{e}_r + r \varphi' \:\vu{e}_\varphi \end{aligned}

Consequently, the angular momentum L\vb{L} is as follows, pointing purely in the zz-direction:

L(t)=μr×v=μ(re^r×rφe^φ)=μr2φe^z\begin{aligned} \vb{L}(t) = \mu \vb{r} \cross \vb{v} = \mu \big( r \vu{e}_r \cross r \varphi' \vu{e}_\varphi \big) = \mu r^2 \varphi' \:\vu{e}_z \end{aligned}

Now, from the figure above, we can argue geometrically that at infinity t=±t = \pm \infty, the ratio b/rb/r is related to the angle χ\chi between v\vb{v} and r\vb{r} like so:

br(±)=sinχ(±)χ(t)(r,v)\begin{aligned} \frac{b}{r(\pm \infty)} = \sin{\chi(\pm \infty)} \qquad \quad \chi(t) \equiv \measuredangle(\vb{r}, \vb{v}) \end{aligned}

With this, we can rewrite the magnitude of the angular momentum L\vb{L} as follows, where the total velocity v|\vb{v}| is a constant, thanks to conservation of energy:

L(±)=μr×v=μrvsinχ=μbv\begin{aligned} \big| \vb{L}(\pm \infty) \big| = \mu \big| \vb{r} \cross \vb{v} \big| = \mu r |\vb{v}| \sin{\chi} = \mu b |\vb{v}| \end{aligned}

However, conveniently, angular momentum is also conserved, i.e. L\vb{L} is constant in time:

L(t)=μ(r×v+v×v)=r×(μv)=r×(q1q24πε0rr3)=0\begin{aligned} \vb{L}'(t) &= \mu \big( \vb{r} \cross \vb{v}' + \vb{v} \cross \vb{v} \big) = \vb{r} \cross (\mu \vb{v}') = \vb{r} \cross \Big( \frac{q_1 q_2}{4 \pi \varepsilon_0} \frac{\vb{r}}{r^3} \Big) = 0 \end{aligned}

Where we have replaced μv\mu \vb{v}' with the equation of motion. Thanks to this, we can equate the two preceding expressions for L\vb{L}, leading to the relation below. Note the appearance of a new minus, because the sketch shows that φ<0\varphi' < 0, i.e. φ\varphi decreases with increasing tt:

μr2dφdt=μbv    dt=r2bvdφ\begin{aligned} - \mu r^2 \dv{\varphi}{t} = \mu b |\vb{v}| \quad \implies \quad \dd{t} = - \frac{r^2}{b |\vb{v}|} \dd{\varphi} \end{aligned}

Now, at last, we turn to the main equation of motion. Its yy-component is given by:

μdvydt=q1q24πε0yr3    μdvy=q1q24πε0yr3dt\begin{aligned} \mu \dv{v_y}{t} = \frac{q_1 q_2}{4 \pi \varepsilon_0} \frac{y}{r^3} \quad \implies \quad \mu \dd{v_y} = \frac{q_1 q_2}{4 \pi \varepsilon_0} \frac{y}{r^3} \dd{t} \end{aligned}

We replace dt\dd{t} with our earlier relation, and recognize geometrically that y/r=sinφy/r = \sin{\varphi}:

μdvy=q1q24πε0bvyrdφ=q1q24πε0bvsinφdφ=q1q24πε0bvd(cosφ)\begin{aligned} \mu \dd{v_y} = - \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}|} \frac{y}{r} \dd{\varphi} = - \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}|} \sin{\varphi} \dd{\varphi} = \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}|} \dd{(\cos{\varphi})} \end{aligned}

Integrating this from the initial state ii at t=t = -\infty to the final state ff at t=t = \infty yields:

Δvyifdvy=q1q24πε0bvμ(cosφfcosφi)\begin{aligned} \Delta v_y \equiv \int_{i}^{f} \dd{v_y} = \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}| \mu} \big( \cos{\varphi_f} - \cos{\varphi_i} \big) \end{aligned}

From symmetry, we see that φi=π ⁣ ⁣φf\varphi_i = \pi \!-\! \varphi_f, and that Δvy=vy,f ⁣ ⁣vy,i=2vy,f\Delta v_y = v_{y,f} \!-\! v_{y,i} = 2 v_{y,f}, such that:

2vy,f=q1q24πε0bvμ(cosφfcos(π ⁣ ⁣φf))=q1q24πε0bvμ(2cosφf)\begin{aligned} 2 v_{y,f} = \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}| \mu} \big( \cos{\varphi_f} - \cos(\pi \!-\! \varphi_f) \big) = \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}| \mu} \big( 2 \cos{\varphi_f} \big) \end{aligned}

Furthermore, geometrically, at t=t = \infty we notice that vy,f=vsinφfv_{y,f} = |\vb{v}| \sin{\varphi_f}, leading to:

2vsinφf=q1q22πε0bvμcosφf\begin{aligned} 2 |\vb{v}| \sin{\varphi_f} = \frac{q_1 q_2}{2 \pi \varepsilon_0 b |\vb{v}| \mu} \cos{\varphi_f} \end{aligned}

Rearranging this yields the following equation for the final polar angle φfφ()\varphi_f \equiv \varphi(\infty):

tanφf=sinφfcosφf=q1q24πε0bv2μ\begin{aligned} \tan{\varphi_f} = \frac{\sin{\varphi_f}}{\cos{\varphi_f}} = \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}|^2 \mu} \end{aligned}

However, we want θ\theta, not φf\varphi_f. One last use of symmetry and geometry tells us that θ=2φf\theta = 2 \varphi_f, and we thus arrive at the celebrated Rutherford scattering formula:

tan ⁣(θ2)=q1q24πε0bv2μ\begin{aligned} \boxed{ \tan\!\Big( \frac{\theta}{2} \Big) = \frac{q_1 q_2}{4 \pi \varepsilon_0 b |\vb{v}|^2 \mu} } \end{aligned}

In fact, this formula is also valid if q1q_1 and q2q_2 have opposite signs; in that case particle 2 is simply located on the other side of particle 1’s trajectory.


  1. P.M. Bellan, Fundamentals of plasma physics, 1st edition, Cambridge.
  2. M. Salewski, A.H. Nielsen, Plasma physics: lecture notes, 2021, unpublished.