IPhO 2015 · Theory problem 2

The Extremum Principle

A The Extremum Principle in Mechanics

Consider a horizontal frictionless \(x-y\) plane shown in Fig. 1. It is divided into two regions, I and II, by a line AB satisfying the equation \(x=x_1\). The potential energy of a point particle of mass \(m\) in region I is \(V=0\) while it is \(V=V_0\) in region II. The particle is sent from the origin O with speed \(v_1\) along a line making an angle \(\theta_1\) with the \(x\)-axis. It reaches point P in region II traveling with speed \(v_2\) along a line that makes an angle \(\theta_2\) with the \(x\)-axis. Ignore gravity and relativistic effects in this entire task T-2 (all parts).

Figure 1

Figure 1

A1 Obtain an expression for \(v_2\) in terms of \(m\), \(v_1\) and \(V_0\). 0.2
A2 Express \(v_2\) in terms of \(v_1\), \(\theta_1\) and \(\theta_2\). 0.3

We define a quantity called action \(A=m\int v(s)ds\), where \(ds\) is the infinitesimal length along the trajectory of a particle of mass \(m\) moving with speed \(v(s)\). The integral is taken over the path. As an example, for a particle moving with constant speed \(v\) on a circular path of radius \(R\), the action \(A\) for one revolution will be \(2\pi mRv\). For a particle with constant energy \(E\), it can be shown that of all the possible trajectories between two fixed points, the actual trajectory is the one on which \(A\) defined above is an extremum (minimum or maximum). Historically this is known as the Principle of Least Action (PLA).

A3 PLA implies that the trajectory of a particle moving between two fixed points in a region of constant potential will be a straight line. Let the two fixed points O and P in Fig. 1 have coordinates \((0,0)\) and \((x_0,y_0)\) respectively and the boundary point where the particle transits from region I to region II have coordinates \((x_1,\alpha)\). Note that \(x_1\) is fixed and the action depends on the coordinate \(\alpha\) only. State the expression for the action \(A(\alpha)\). Use PLA to obtain the relationship between \(v_1/v_2\) and these coordinates. 1.0

B The Extremum Principle in Optics

A light ray travels from medium I to medium II with refractive indices \(n_1\) and \(n_2\) respectively. The two media are separated by a line parallel to the \(x\)-axis. The light ray makes an angle \(i_1\) with the \(y\)-axis in medium I and \(i_2\) in medium II (see Fig. 2). To obtain the trajectory of the ray, we make use of another extremum (minimum or maximum) principle known as Fermat’s principle of least time.

Figure 2

Figure 2

B1 The principle states that between two fixed points, a light ray moves along a path such that time taken between the two points is an extremum. Derive the relation between \(\sin i_1\) and \(\sin i_2\) on the basis of Fermat’s principle. 0.5

Shown in Fig. 3 is a schematic sketch of the path of a laser beam incident horizontally on a solution of sugar in which the concentration of sugar decreases with height. As a consequence, the refractive index of the solution also decreases with height.

Figure 3

Figure 3: Tank of Sugar Solution

B2 Assume that the refractive index \(n(y)\) depends only on \(y\). Use the equation obtained in B1 to obtain the expression for the slope \(dy/dx\) of the beam’s path in terms of refractive index \(n_0\) at \(y=0\) and \(n(y)\). 1.5
B3 The laser beam is directed horizontally from the origin \((0,0)\) into the sugar solution at a height \(y_0\) from the bottom of the tank as shown in figure 3. Take \(n(y)=n_0-ky\) where \(n_0\) and \(k\) are positive constants. Obtain an expression for \(x\) in terms of \(y\) and related quantities for the actual trajectory of the laser beam. 1.2

\(^{1}\) Manoj Harbola (IIT-Kanpur) and Vijay A. Singh (ex-National Coordinator, Science Olympiads) were the principal authors of this problem. The contributions of the Academic Committee, Academic Development Group and the International Board are gratefully acknowledged.

You may use: \(\int\sec\theta\,d\theta=\ln(\sec\theta+\tan\theta)+\text{constant}\), where \(\sec\theta=1/\cos\theta\) or
\(\displaystyle\int\frac{dx}{\sqrt{x^2-1}}=\ln\!\left(x+\sqrt{x^2-1}\right)+\text{constant}\)
B4 Obtain the value of \(x_0\), the point where the beam meets the bottom of the tank. Take \(y_0=10.0\ \mathrm{cm}\), \(n_0=1.50\), \(k=0.050\ \mathrm{cm^{-1}}\) (\(1\ \mathrm{cm}=10^{-2}\ \mathrm{m}\)). 0.8

C The Extremum Principle and the Wave Nature of Matter

We now explore the connection between the PLA and the wave nature of a moving particle. For this we assume that a particle moving from O to P can take all possible trajectories and we will seek a trajectory that depends on the constructive interference of de Broglie waves.

C1 As the particle moves along its trajectory by an infinitesimal distance \(\Delta s\), relate the change \(\Delta\varphi\) in the phase of its de Broglie wave to the change \(\Delta A\) in the action and the Planck constant. 0.6
C2 Recall the problem from part A where the particle traverses from O to P (see Fig. 4). Let an opaque partition be placed at the boundary AB between the two regions. There is a small opening CD of width \(d\) in AB such that \(d\ll(x_0-x_1)\) and \(d\ll x_1\).

Consider two extreme paths OCP and ODP such that OCP lies on the classical trajectory discussed in part A. Obtain the phase difference \(\Delta\varphi_{CD}\) between the two paths to first order.
1.2
Figure 4

Figure 4

D Matter Wave Interference

Consider an electron gun at O which directs a collimated beam of electrons to a narrow slit at F in the opaque partition \(\mathrm{A_1B_1}\) at \(x=x_1\) such that OFP is a straight line. P is a point on the screen at \(x=x_0\) (see Fig. 5). The speed in I is \(v_1=2.0000\times10^7\ \mathrm{m\,s^{-1}}\) and \(\theta=10.0000^\circ\). The potential in II is such that speed \(v_2=1.9900\times10^7\ \mathrm{m\,s^{-1}}\). The distance \(x_0-x_1\) is \(250.00\ \mathrm{mm}\) (\(1\mathrm{mm}=10^{-3}\mathrm{m}\)). Ignore electron-electron interaction.

Figure 5

Figure 5

D1 If the electrons at O have been accelerated from rest, calculate the accelerating potential \(U_1\). 0.3
D2 Another identical slit G is made in the partition \(\mathrm{A_1B_1}\) at a distance of \(215.00\ \mathrm{nm}\) (\(1\mathrm{nm}=10^{-9}\mathrm{m}\)) below slit F (Fig. 5). If the phase difference between de Broglie waves arriving at P through the slits F and G is \(2\pi\beta\), calculate \(\beta\). 0.8
D3 What is the smallest distance \(\Delta y\) from P at which null (zero) electron detection maybe expected on the screen? [Note: you may find the approximation \(\sin(\theta+\Delta\theta)\approx\sin\theta+\Delta\theta\cos\theta\) useful] 1.2
D4 The beam has a square cross section of \(500\mathrm{nm}\times500\mathrm{nm}\) and the setup is \(2\ \mathrm{m}\) long. What should be the minimum flux density \(I_{\min}\) (number of electrons per unit normal area per unit time) if, on an average, there is at least one electron in the setup at a given time? 0.4

Official English text of the problem.

Answers

  • A1: \(\displaystyle v_2=\sqrt{v_1^2-\frac{2V_0}{m}}\).
  • A2: \(\displaystyle v_2=v_1\frac{\sin\theta_1}{\sin\theta_2}\).
  • A3:
    \[ A(\alpha)=m v_1\sqrt{x_1^2+\alpha^2} +m v_2\sqrt{(x_0-x_1)^2+(y_0-\alpha)^2}, \] \[ \frac{v_1}{v_2} =\frac{(y_0-\alpha)\sqrt{x_1^2+\alpha^2}} {\alpha\sqrt{(x_0-x_1)^2+(y_0-\alpha)^2}}. \]
  • B1: \(\displaystyle n_1\sin i_1=n_2\sin i_2\).
  • B2: \(\displaystyle \frac{dy}{dx}=-\sqrt{\frac{n(y)^2}{n_0^2}-1}\).
  • B3:
    \[ x(y)=\frac{n_0}{k}\ln\!\left[ \frac{n_0-ky+\sqrt{(n_0-ky)^2-n_0^2}}{n_0} \right]. \] Equivalently, \(\displaystyle y(x)=\frac{n_0}{k}\left[1-\cosh\!\left(\frac{kx}{n_0}\right)\right]\).
  • B4: \(\displaystyle x_0=23.86\ldots\ \mathrm{cm}\approx24\ \mathrm{cm}\).
  • C1: \(\displaystyle \Delta\varphi=\frac{2\pi}{h}\Delta A=\frac{\Delta A}{\hbar}\).
  • C2: \(\displaystyle \Delta\varphi_{CD}=0\) to first order in \(d\); the leading correction is quadratic.
  • D1: \(\displaystyle U_1\approx1.14\times10^3\ \mathrm{V}\).
  • D2: \(\displaystyle \beta=\frac{m_e d(v_1-v_2)\sin\theta}{h}\approx5.13\).
  • D3: \(\displaystyle \Delta y\approx1.6\times10^{-5}\ \mathrm{m}=16\ \mu\mathrm{m}\), below P.
  • D4: \(\displaystyle I_{\min}\approx4\times10^{19}\ \mathrm{m^{-2}\,s^{-1}}\).

Solution

For the numerical electron calculations, use \[ m_e=9.1093837\times10^{-31}\ \mathrm{kg},\qquad e=1.602176634\times10^{-19}\ \mathrm{C},\qquad h=6.62607015\times10^{-34}\ \mathrm{J\,s}, \] where \(e\) is the magnitude of the electron charge.

A1

The potential is time independent, so mechanical energy is conserved: \[ \frac12 mv_1^2=\frac12 mv_2^2+V_0. \] Therefore, \[ \boxed{v_2=\sqrt{v_1^2-\frac{2V_0}{m}}}. \] The problem assumes that the particle is transmitted into region II.

A2

The potential depends only on \(x\), so the force has no \(y\)-component. Consequently, the momentum parallel to the boundary is conserved: \[ mv_1\sin\theta_1=mv_2\sin\theta_2. \] Thus, \[ \boxed{v_2=v_1\frac{\sin\theta_1}{\sin\theta_2}}. \]

A3

Let the boundary crossing point be \(Q=(x_1,\alpha)\). The two straight segments have lengths \[ L_1=\sqrt{x_1^2+\alpha^2},\qquad L_2=\sqrt{(x_0-x_1)^2+(y_0-\alpha)^2}. \] At fixed total energy, the speeds in the two regions are fixed. Hence \[ \boxed{A(\alpha)=m(v_1L_1+v_2L_2)}. \]

The actual trajectory satisfies \[ \frac{dA}{d\alpha} =m\left[ v_1\frac{\alpha}{L_1} -v_2\frac{y_0-\alpha}{L_2} \right]=0. \] Therefore, \[ \boxed{ \frac{v_1}{v_2} =\frac{(y_0-\alpha)L_1}{\alpha L_2} =\frac{(y_0-\alpha)\sqrt{x_1^2+\alpha^2}} {\alpha\sqrt{(x_0-x_1)^2+(y_0-\alpha)^2}} }. \]

Since \(\alpha/L_1=\sin\theta_1\) and \((y_0-\alpha)/L_2=\sin\theta_2\), this reproduces A2. Moreover, \[ \frac{d^2A}{d\alpha^2} =m\left[ \frac{v_1x_1^2}{L_1^3} +\frac{v_2(x_0-x_1)^2}{L_2^3} \right]>0, \] so this stationary action is a minimum.

B1

Write the horizontal interface as \(y=b\), and let the ray cross it at \((\alpha,b)\). The speed of light in medium \(j\) is \(c/n_j\). Its travel time from O to P is therefore \[ T(\alpha)=\frac{n_1}{c}\sqrt{\alpha^2+b^2} +\frac{n_2}{c}\sqrt{(x_0-\alpha)^2+(y_0-b)^2}. \]

Fermat’s principle requires \(dT/d\alpha=0\): \[ n_1\frac{\alpha}{\sqrt{\alpha^2+b^2}} =n_2\frac{x_0-\alpha} {\sqrt{(x_0-\alpha)^2+(y_0-b)^2}}. \] Because the angles are measured from the interface normal, the two geometric ratios are \(\sin i_1\) and \(\sin i_2\). Thus \[ \boxed{n_1\sin i_1=n_2\sin i_2}, \qquad \boxed{\frac{\sin i_1}{\sin i_2}=\frac{n_2}{n_1}}. \]

B2

Regard the liquid as a succession of thin horizontal layers. Applying B1 between adjacent layers shows that \[ n(y)\sin i(y)=\text{constant}. \] The initial ray is horizontal, so \(i(0)=90^\circ\), giving \[ n(y)\sin i(y)=n_0. \]

For a ray traveling towards increasing \(x\), \[ \sin i=\frac{dx}{ds} =\frac{1}{\sqrt{1+(dy/dx)^2}}. \] Consequently, \[ \frac{n(y)}{\sqrt{1+(dy/dx)^2}}=n_0, \] and hence \[ \left(\frac{dy}{dx}\right)^2=\frac{n(y)^2}{n_0^2}-1. \]

The ray bends towards the larger refractive index, which is downward. Choosing the negative slope, \[ \boxed{\frac{dy}{dx} =-\sqrt{\frac{n(y)^2}{n_0^2}-1}}. \]

B3

Substitute \(n(y)=n_0-ky\): \[ \frac{dx}{dy} =-\frac{n_0}{\sqrt{(n_0-ky)^2-n_0^2}}. \] Integrating from the entry point \((0,0)\), \[ x=-n_0\int_0^y \frac{d\eta}{\sqrt{(n_0-k\eta)^2-n_0^2}}. \]

With \(u=(n_0-k\eta)/n_0\), this becomes \[ x=\frac{n_0}{k} \int_1^{(n_0-ky)/n_0}\frac{du}{\sqrt{u^2-1}}. \] Using the supplied integral, \[ \boxed{ x(y)=\frac{n_0}{k} \ln\!\left[ \frac{n_0-ky+\sqrt{(n_0-ky)^2-n_0^2}}{n_0} \right] }. \] This is the branch with \(x\ge0\) and \(-y_0\le y\le0\).

Equivalently, \[ \frac{kx}{n_0} =\operatorname{arcosh}\!\left(1-\frac{ky}{n_0}\right), \] so \[ \boxed{ y(x)=\frac{n_0}{k} \left[1-\cosh\!\left(\frac{kx}{n_0}\right)\right] }. \] It passes through the origin with a horizontal tangent and curves increasingly downward.

B4

At the bottom, \(y=-y_0\). Therefore, \[ x_0=\frac{n_0}{k} \operatorname{arcosh}\!\left(1+\frac{ky_0}{n_0}\right). \] Using centimetres consistently, \[ \frac{n_0}{k}=30.0\ \mathrm{cm}, \qquad 1+\frac{ky_0}{n_0} =1+\frac{0.050(10.0)}{1.50}=\frac43. \] Thus \[ x_0=30.0\ln\!\left(\frac{4+\sqrt7}{3}\right)\ \mathrm{cm} =23.86096\ldots\ \mathrm{cm}. \] Hence \[ \boxed{x_0\approx24\ \mathrm{cm}=0.24\ \mathrm{m}}. \]

C1

The de Broglie wavelength is \[ \lambda=\frac{h}{p}=\frac{h}{mv}. \] The spatial phase accumulated over an infinitesimal distance is \[ \Delta\varphi=\frac{2\pi}{\lambda}\Delta s =\frac{2\pi}{h}mv\,\Delta s. \] Since \(\Delta A=mv\,\Delta s\), \[ \boxed{\Delta\varphi=\frac{2\pi}{h}\Delta A =\frac{\Delta A}{\hbar}}. \] For stationary waves of the same energy, the common temporal phase cancels when comparing paths at the same observation time.

C2

Let C have coordinate \((x_1,\alpha)\), so D has coordinate \((x_1,\alpha+d)\). To first order, \[ OD-OC=d\sin\theta_1+O(d^2), \] while \[ DP-CP=-d\sin\theta_2+O(d^2). \] Therefore, \[ A_{ODP}-A_{OCP} =md\left(v_1\sin\theta_1-v_2\sin\theta_2\right)+O(d^2). \]

Because OCP is the classical trajectory, A2 gives \[ v_1\sin\theta_1=v_2\sin\theta_2. \] Using C1, \[ \Delta\varphi_{CD} =\frac{2\pi md}{h} \left(v_1\sin\theta_1-v_2\sin\theta_2\right)+O(d^2). \] Thus \[ \boxed{\Delta\varphi_{CD}=0\quad\text{to first order in }d}. \]

The phase difference is generally nonzero at second order. Stationary action makes neighboring paths have the same phase to first order.

D1

Acceleration from rest through a potential difference of magnitude \(U_1\) gives \[ eU_1=\frac12m_ev_1^2. \] Hence \[ U_1=\frac{m_ev_1^2}{2e} =1137.126\ldots\ \mathrm{V}, \] or \[ \boxed{U_1\approx1.14\times10^3\ \mathrm{V}}. \]

D2

Let \[ d=215.00\times10^{-9}\ \mathrm{m}, \qquad L=x_0-x_1=0.25000\ \mathrm{m}. \] We treat the incident collimated beam as a coherent plane wave and use the far-field path-length differences to first order in the slit separation.

Since F is above G, the incident phase difference is \[ \varphi_F^{\mathrm{in}}-\varphi_G^{\mathrm{in}} =\frac{m_ev_1}{\hbar}d\sin\theta. \] At P, which lies on the straight continuation of OF, \[ FP-GP\simeq-d\sin\theta. \] Propagation in region II therefore contributes \[ -\frac{m_ev_2}{\hbar}d\sin\theta. \]

Taking the total phase difference as the phase through F minus that through G, \[ 2\pi\beta =\frac{m_ed}{\hbar}(v_1-v_2)\sin\theta. \] Thus \[ \boxed{\beta=\frac{m_ed(v_1-v_2)\sin\theta}{h}}. \] Numerically, \[ \beta=5.132650\ldots, \qquad \boxed{\beta\approx5.13}. \] This is the unreduced phase difference in cycles; modulo one cycle it is \(0.132650\ldots\).

D3

For identical slits with equal amplitudes, destructive interference occurs when the phase difference is an odd multiple of \(\pi\), or a half-integer number of cycles.

Let the outgoing direction towards a nearby screen point be \(\theta+\delta\theta\). The incident phase difference remains fixed, while the propagation phase changes. The phase difference in cycles becomes \[ b(\delta\theta) =\frac{m_ed}{h} \left[v_1\sin\theta-v_2\sin(\theta+\delta\theta)\right]. \] Using the given small-angle expansion, \[ b(\delta\theta) \simeq\beta-\frac{m_ev_2d}{h}\cos\theta\,\delta\theta. \]

The neighboring half-integers are \(4.5\) and \(5.5\). Their distances from \(\beta\) are \[ \beta-4.5=0.632650\ldots, \qquad 5.5-\beta=0.367350\ldots. \] The nearest minimum therefore has \(b=5.5\), giving \[ \delta\theta =-\frac{h(5.5-\beta)}{m_ev_2d\cos\theta}. \] The negative sign means the minimum lies below P.

The signed displacement on the vertical screen is \[ \delta y =L\left[\tan(\theta+\delta\theta)-\tan\theta\right] \simeq\frac{L}{\cos^2\theta}\delta\theta. \] Thus its magnitude is \[ \boxed{ \Delta y =\frac{hL(5.5-\beta)} {m_ev_2d\cos^3\theta} }. \]

Numerically, \[ \frac{hL}{m_ev_2d\cos^3\theta} =44.5001\ldots\ \mu\mathrm{m}, \] so the leading-order result is \[ \Delta y=(44.5001\ldots)(0.3673499\ldots)\ \mu\mathrm{m} =16.3471\ldots\ \mu\mathrm{m}. \] Therefore, \[ \boxed{\Delta y\approx16\ \mu\mathrm{m},\quad\text{below P}}. \]

The small-angle and far-field approximations are appropriate here: \(|\delta\theta|\simeq6.34\times10^{-5}\ \mathrm{rad}\), and \(d/L=8.60\times10^{-7}\). The quoted result neglects higher-order path-length corrections.

D4

The beam’s normal cross-sectional area is \[ S=(500\times10^{-9})^2 =2.50\times10^{-13}\ \mathrm{m^2}. \] For a \(2\ \mathrm{m}\) flight distance, using \(v\simeq2.0\times10^7\ \mathrm{m\,s^{-1}}\), the transit time is \[ \tau\simeq\frac{2}{2.0\times10^7} =1.0\times10^{-7}\ \mathrm{s}. \]

If the flux density is \(I\), the number entering per unit time is \(IS\). In steady operation, the average number in the apparatus is \[ \overline N=IS\tau. \] The minimum flux corresponds to \(\overline N=1\): \[ I_{\min}=\frac{1}{S\tau}. \] Therefore, \[ \boxed{ I_{\min}\simeq \frac{1}{(2.50\times10^{-13})(1.0\times10^{-7})} =4\times10^{19}\ \mathrm{m^{-2}\,s^{-1}} }. \]

More precisely, \(\tau=\ell_1/v_1+\ell_2/v_2\), where \(\ell_1+\ell_2\) is the flight distance. The two speeds differ by only \(0.5\%\), so replacing them by the common approximate speed does not affect the quoted precision.

The solution was checked against the official solution.