On This Page

Chapter 3 Decay rates and cross sections

Recap: Fermi’s golden rule

Given a system which Hamiltonian $H = H_0 +H^\prime$, where $H^\prime \ll H_0$. The question is, what’s the average transition rate $\varGamma_{fi}$ between state f and state i, when $t\to 0$?

Due to the property of perturbation, we can simply calculate the change of probability amplitude of states under the eigen wave functions of Hamiltonian $H_0$. Assuming the wave function $ \Psi(t) = \sum_{k} c_{k}(t) \psi_{k}(0) \mathrm{e}^{-\mathrm{i} \omega_{k} t} $, we calculate the coefficients to the first order, and get

\[P_{fi}(t)=c_f(t)c_f^*(t)=|T_{fi}|^2\frac{\sin^2\frac{\omega t}{2}}{\left(\frac{\omega}{2}\right)^2}\]
where $T_{fi}=<\psi_f(0) H^\prime \psi_i(0)>$, and $\omega = E_f - E_i$ (set $\hbar = 1$). What we’re interested in is the transition rate $\varGamma$:
\[\varGamma_{fi}=\lim_{T \to \infty}\frac{P_{fi}(T)}{T}= |T_{fi}|^2\lim_{T\to\infty}\frac{1}{T}\frac{\sin^2{\frac{\omega T}{2}}}{\left(\frac{\omega}{2}\right)^2}\]

According to the definition of Dirac-delta function, the limitation above converge to

\[\varGamma_{fi}=2\pi|T_{fi}|^2\delta(\omega)\]

Proof for the delta function:

Going back to the integral form of $\text{sinc}(x)$, we got

$$

\begin{aligned} \lim_{T\to\infty}\frac{1}{T}\frac{\sin^2{\frac{\omega T}{2}}}{\left(\frac{\omega}{2}\right)^2} &= \lim_{T\to\infty}\frac{1}{T}\int_0^T\mathrm e^{-\mathrm i \omega t}\mathrm d t \int 0^T\mathrm e ^{\mathrm i \omega t^\prime}\mathrm d t^\prime
&= \lim
{T\to\infty}\frac{1}{T}\int_0^T\int 0^T \mathrm e ^{\mathrm i \omega (t^\prime-t)}\mathrm d t\mathrm d t^\prime
&=\lim
{T\to\infty}\frac{1}{T}\int_0^T\mathrm d t\int_{-t}^{T-t}\mathrm e^{\mathrm i \omega\tau}\mathrm d \tau
\text{(periodic function)}&=\lim_{T\to\infty}\frac{1}{T}\cdot T\int_{-T/2}^{T/2}\mathrm e^{\mathrm i \omega\tau}\mathrm d \tau
&=\lim_{T\to\infty}\int_{-T/2}^{T/2}\mathrm e^{\mathrm i \omega\tau}\mathrm d \tau
&= 2\pi \delta(\omega) \end{aligned} $$ Q.E.D.

For systems with continuous energy levels, the transition rate can be written in the integral form, $$

\varGamma_{fi} = \int\mathrm d \varGamma_{fi} = \int2\pi T_{fi} ^2\delta(\omega) \frac{\mathrm d n}{\mathrm d E_f}\mathrm d E_f = 2\pi T_{fi} ^2 \rho(E_i)
\[where $\rho(E) = \frac{\mathrm d n}{\mathrm d E}$ called the density of energy. ### Phase space and wavefunction normalization Using "box normalization", we can get the wave function integrated into 1 in volume $V=a^3$. According to the periodic boundary condition, we got\]

\vec p = \frac{2\pi}{a}\vec n, \vec n \in \mathbb Z^3

\[Since the large numeric value of a, we get "density of momentum"\]

\frac{\mathrm d n}{\mathrm d p}=\frac{V}{(2\pi)^3}\cdot 4\pi p^2

\[Due to normalization factor in $|T_{fi}|^2$, $V$ will not appear in the expression of the decay rate$\varGamma_{fi}$. One step more, we can set V=1 initially. As for multiple particle scenarios, we get\]

\text{(number of states for N particles)}\mathrm d n= \prod_{i=1}^{N-1}\mathrm d n_i=\prod_{i=1}^{N-1}\frac{1}{(2\pi)^3}\mathrm d^3\vec p_i

\[The last term for $p_N$ is constrained by the law of conservation of momentum. For symmetry reasons, we can also finish this term in the form of delta-function, i.e.,\]

\mathrm d n =(2\pi)^3\delta(\vec P-\sum_{i=1}^N\vec p_i)\prod_{i=1}^{N}\frac{1}{(2\pi)^3}\mathrm d^3\vec p_i

\[where $\vec P$ is the total momentum of the system of particles. #### Lorentz-invariant phase space Apparently the integral factor $\mathrm d ^3\vec p$ is not Lorentz-invariant. Let's consider another way for integration, which is\]

\int {\psi^\prime}^*\psi^\prime\mathrm d^3\vec x = 2E

\[Then we can define\]
M_{fi} ^2 = \int {\psi^\prime_f}^*H^\prime\psi^\prime_i\mathrm d^3\vec x ^2=2 E_i\cdot2E_f \cdot T_{fi} ^2
\[which lead decay rate into\]

\begin{aligned} \varGamma_{fi}=&\int2\pi|T_{fi}|^2\delta(E-\sum_{i=1}^NE_i)\mathrm d n
&= \int(2\pi)^4|T_{fi}|^2\delta(E-\sum_{i=1}^NE_i)\delta(\vec P-\sum_{i=1}^N\vec p_i)\prod_{i=1}^{N}\frac{1}{(2\pi)^3}\mathrm d^3\vec p_i
&= \int\frac{(2\pi)^4}{2E_i}|M_{fi}|^2\delta(E-\sum_{i=1}^NE_i)\delta(\vec P-\sum_{i=1}^N\vec p_i)\prod_{i=1}^{N}\frac{1}{(2\pi)^3}\frac{\mathrm d^3\vec p_i}{2E_i}
\end{aligned}

$$ Obviously the p/E factor is Lorentz-invariant, so we can calculate this term in any frame we want.

Proof: \(\frac{\mathrm d p^\prime}{\mathrm d p}=\gamma( 1-\beta\frac{\mathrm dE}{\mathrm d p})\)

\[\frac{E^\prime}{E}=\gamma(1-\beta\frac{p}{E})\]

Using mass-energy equation: \(E^2 = p^2 +m^2\Rightarrow E\mathrm d E = p \mathrm d p\) Hence \(\frac{\mathrm d p^\prime}{E^\prime}=\frac{\mathrm d p}{E}\) Q.E.D.

Another footprint for LIPS(Lorentz-invariant phase space): \(LIPS = \int \prod_{i=1}^{N} \delta (E_i^2-\vec p_i^2-m_i^2)\mathrm d^4\vec p_i=\int \prod_{i=1}^{N}\frac{\mathrm d^3\vec p_i}{2E_i}\) Each delta meets a constrain condition.

### Particle Decays

As an example, let’s calculate the decay rate for the 2-body situation,i.e. particle interaction $a \to 1+2$. For simplicity, finish this calculation in center-of-mass frame, where total momentum equals to 0 and $E_a = m_a$ $$

\begin{aligned} \varGamma_{fi}&= \int\frac{(2\pi)^4}{2m_a}|M_{fi}|^2\delta(E_a-E_1-E_2)\delta(\vec p_1+\vec p_2)\frac{1}{(2\pi)^6}\frac{\mathrm d^3\vec p_1}{2E_1}\frac{\mathrm d^3\vec p_2}{2E_2}
&=\frac{1}{8\pi^2m_a}\int|M_{fi}|^2\cdot\frac{1}{4E_1E_2}\cdot\delta(E_a-\sqrt{m_1^2+p_1^2}-\sqrt{m_2^2+p_1^2})p_1^2\mathrm dp_1\mathrm d^2\Omega^
&=\frac{p^
}{32\pi^2m_a^2}\int|M_{fi}|^2\mathrm d^2\Omega^* \end{aligned}

\[where $p^*$ is the solution for $\delta(f(p))=0$, which is\]

p^*=\frac{\sqrt{(m_a^2-(m_1+m_2)^2)(m_a^2-(m_1-m_2)^2)}}{2m_a}

\[### Interaction cross sections Let's introduce the definition of **cross section**. For interactions between a and b, we have\]

\text{(the rate of interaction particle b)}r_b :=\frac{\mathrm d N_b}{\mathrm d t}= \text{(cross section)}\sigma\cdot\text{(flux)}\phi_a

\[where flux\]

\phi_a = \frac{\mathrm d N_a}{\mathrm d t \mathrm d S}

\[For instance, let's consider this definition in the interactions between a and b. Suppose the vertical area of the flux of particle a is $A$, and the velocity of the flux is $v$, and the still particle b has the uniform distribution of $n$, hence in time $\delta t$, the number of particle b encountered with particle a is\]

\delta N = Av\delta t n

\[By the definition of cross section, we get the probability of interaction\]

\delta P = \frac{\delta N\cdot \sigma}{A} = nv\sigma\delta t

\[As the classical explanation of cross section, we got the cross section by how much area where the particle in can interact; in the quantum world, all the processes happen with a probability distribution, so the cross section can be an effective equivalent for this quantum process. #### Lorentz-invariant flux Let's find the relations between interaction rate and cross section. For simplicity, let's consider the interaction $a+b\to1+2$. By the definition of cross section $\sigma$, we have\]

\text{(interaction rate)}r = \phi_a \sigma n_bV = n_an_b(v_a+v_b)\sigma V

\[Therefore\]

\varGamma_{fi}=(v_a+v_b)\sigma

\[Hence\]
\sigma = \int\frac{(2\pi)^4}{4(v_a+v_b)E_aE_b} M_{fi} ^2\delta(E_a+E_b-E_1-E_2)\delta(\vec p_a+\vec p_b-\vec p_1-\vec p_2)\frac{1}{(2\pi)^6}\frac{\mathrm d^3\vec p_1}{2E_1}\frac{\mathrm d^3\vec p_2}{2E_2}
\[Apparently the factor $(v_a+v_b)E_aE_b$ is Lorentz-invariant, we can do this calculation in center-of-mass frame, then we get\]
\sigma = \frac{p_f^}{64\pi^2(E_a^+E_b^)^2p_i^}\int M_{fi} ^2\mathrm d^2\Omega^*
\[### Differential cross sections\]
\frac{\mathrm d \sigma}{\mathrm d^2 \Omega^} =\frac{p_f^}{64\pi^2(E_a^+E_b^)^2p_i^*} M_{fi} ^2
\[--- <del>TO BE CONTINUED: the precise calculation for electron-proton scattering</del> To verify our postulation, we have to change this differential cross section into laboratory frame. For a simple example, let's consider the elastic scattering $e^-p\to e^-p$ (ignore the mass of electron since it much less than that of proton), where the target particle is still in L(short for laboratory) frame. We have the relationship of four momentum between L and C(Center-of-mass) frame:\]

E_3^* = \gamma(E_3-\beta p_3\cos\theta)
p_3^\cos \theta^ =\gamma(p_3\cos\theta-\beta E_3)

\[Take the derivative of these 2 equations above, we get\]

\frac{\mathrm d \cos \theta^}{\mathrm d \cos \theta}= \frac{\cos\theta-\beta\frac{\beta E_3}{1-\beta \cos \theta}+E_3}{p_3^}\gamma

\[Align with\]

\beta = \frac{E_1}{E_1+m_p}

\[We have\]
\frac{\mathrm d \sigma}{\mathrm d \Omega}= \frac{ M_{fi} ^2}{64\pi^2}\frac{1}{(m_p+E_1-E_1\cos\theta)^2}

$$

I believe this derivation is much simpler and more straightforward than that in the book.