IPhO 2000 · Theory problem 3

Gravitational-wave detector (Part A) and gravitational deflection of light by the Sun (Part B)

Gravitational waves and the effects of gravity on light.

Part A

This part is concerned with the difficulties of detecting gravitational waves generated by astronomical events. It should be realised that the explosion of a distant supernova may produce fluctuations in the gravitational field strength at the surface of the Earth of about \(10^{-19}\ \mathrm{N\,kg^{-1}}\).

A model for a gravitational wave detector (see figure 3.1) consists of two metal rods each 1m long, held at right angles to each other. One end of each rod is polished optically flat and the other end is held rigidly. The position of one rod is adjusted so there is a minimum signal received from the photocell (see figure 3.1).

Figure 1

Figure 3.1

The rods are given a short sharp impulse by a piezoelectric device. As a result the free ends of the rods oscillate with a longitudinal displacement \(\Delta x_t\), where

\[ \Delta x_t=ae^{-\mu t}\cos(\omega t+\phi), \]

and \(a,\mu,\omega\) and \(\phi\) are constants.

(a) If the amplitude of the motion is reduced by 20% during a 50s interval determine a value for \(\mu\).

(b) Determine also a value for \(\omega\) given that the rods are made of aluminium with a density (\(\rho\)) of \(2700\ \mathrm{kg\,m^{-3}}\) and a Young modulus (\(E\)) of \(7.1\times10^{10}\ \mathrm{Pa}\).

(c) It is impossible to make the rods exactly the same length so the photocell signal has a beat frequency of \(0.005\ \mathrm{Hz}\). What is the difference in length of the rods?

(d) For the rod of length \(l\), derive an algebraic expression for the change in length, \(\Delta l\), due to a change, \(\Delta g\), in the gravitational field strength, \(g\), in terms of \(l\) and other constants of the rod material.

(e) The light produced by the laser is monochromatic with a wavelength of 656nm. If the minimum fringe shift that can be detected is \(10^{-4}\) of the wavelength of the laser, what is the minimum value of \(l\) necessary if such a system were to be capable of detecting variations in \(g\) of \(10^{-19}\ \mathrm{N\,kg^{-1}}\)?

A non-directional form of gravitational wave detector consists of a sphere of copper-alloy of mass 1168 kg, suspended in a vacuum from a vibration-reducing assembly. Transducers, containing tuned circuits, are attached to the sphere to detect changes in its dimensions. The transducers will, however, pick up all spurious vibrations due to, for example, temperature effects and noise due to electric pick up.

(f) To reduce vibrations due to temperature effects the sphere is maintained at a temperature of 100 mK. By what factor will the amplitude of the atomic vibrations been reduced in cooling the assembly from 300 K?

(g) The sphere is initially cooled to 4.2 K using liquid nitrogen and liquid helium. The temperature, \(T\), is further reduced to 100 mK by a refrigeration process, which removes energy from the system at an average rate of 1 mW. Given that the specific thermal capacity, \(s\), of the copper-alloy varies directly as \(T^3\) at these low temperatures, estimate the time taken for the system to cool from 4.2 K to 100 mK, given that \(s=0.072\ \mathrm{J\,kg^{-1}\,K^{-1}}\) at 4.2 K.

Part B

This part is concerned with the effect of a gravitational field on the propagation of light in space.

(a) A photon emitted from the surface of the Sun (mass \(M\), radius \(R\)) is red-shifted. By assuming a rest-mass equivalent for the photon energy, apply Newtonian gravitational theory to show that the effective (or measured) frequency of the photon at infinity is reduced (red-shifted) by the factor \((1-GM/Rc^2)\).

(b) A reduction of the photon’s frequency is equivalent to an increase in its time period, or, using the photon as a standard clock, a dilation of time. In addition, it may be shown that a time dilation is always accompanied by a contraction in the unit of length by the same factor.

We will now try to study the effect that this has on the propagation of light near the Sun. Let us first define an effective refractive index \(n_r\) at a point \(r\) from the centre of the Sun. Let

\[ n_r=\frac{c}{c'_r}, \]

where \(c\) is the speed of light as measured by a co-ordinate system far away from the Sun’s gravitational influence (\(r\to\infty\)), and \(c'_r\) is the speed of light as measured by a co-ordinate system at a distance \(r\) from the centre of the Sun.

Show that \(n_r\) may be approximated to:

\[ n_r=1+\frac{\alpha GM}{rc^2}, \]

for small \(GM/rc^2\), where \(\alpha\) is a constant that you determine.

(c) Using this expression for \(n_r\), calculate in radians the deflection of a light ray from its straight path as it passes the edge of the Sun.

Data:

Gravitational constant, \(G=6.67\times10^{-11}\ \mathrm{N\,m^2\,kg^{-2}}\).
Mass of Sun, \(M=1.99\times10^{30}\ \mathrm{kg}\).
Radius of Sun, \(R=6.95\times10^8\ \mathrm{m}\).
Velocity of light, \(c=3.00\times10^8\ \mathrm{m\,s^{-1}}\).

You may also need the following integral:

\[ \int_{-\infty}^{+\infty}\frac{dx}{(x^2+a^2)^{3/2}} =\frac{2}{a^2}. \]

Official English text of the problem.

Answers

  • A(a): \(\boxed{\mu=4.46\times10^{-3}\ \mathrm{s^{-1}}}\).
  • A(b): For the fundamental longitudinal mode, \(\boxed{\omega\simeq8.06\times10^3\ \mathrm{rad\,s^{-1}}}\).
  • A(c): \(\boxed{|l_1-l_2|\simeq3.90\times10^{-6}\ \mathrm{m}=3.90\ \mu\mathrm{m}}\).
  • A(d): \(\boxed{\Delta l=\dfrac{\rho l^2\Delta g}{2E}}\), with \(\Delta g\) taken along the rod, positive towards its free end.
  • A(e): \(\boxed{l_{\min}\simeq1.31\times10^8\ \mathrm{m}}\), interpreting the threshold as an optical path difference and taking one arm’s axial response relative to the other.
  • A(f): The amplitude decreases by a factor \(\boxed{54.8}\), to \(\boxed{0.0183}\) of its initial value, in the classical harmonic model.
  • A(g): \(\boxed{t\simeq8.83\times10^4\ \mathrm{s}\simeq24.5\ \mathrm{h}}\).
  • B(a): \(\boxed{\nu_\infty\simeq\nu_R\left(1-\dfrac{GM}{Rc^2}\right)}\).
  • B(b): \(\boxed{\alpha=2,\qquad n_r\simeq1+\dfrac{2GM}{rc^2}}\).
  • B(c): \(\boxed{\theta=\dfrac{4GM}{Rc^2}=8.49\times10^{-6}\ \mathrm{rad}}\), towards the Sun.

Solution

A(a)

The amplitude envelope is

\[ A(t)=ae^{-\mu t}. \]

A reduction by \(20\%\) means that the amplitude after \(50\ \mathrm{s}\) is \(0.80\) times its previous value:

\[ \frac{A(t+50)}{A(t)}=e^{-50\mu}=0.80. \]

Therefore,

\[ \boxed{\mu=-\frac{\ln(0.80)}{50} =4.46\times10^{-3}\ \mathrm{s^{-1}}}. \]

A(b)

Treat the rod as a uniform elastic continuum, and take the observed oscillation to be its fundamental longitudinal mode.

Let \(u(x,t)\) be the longitudinal displacement and \(S\) the cross-sectional area. The longitudinal force is \(ES\,\partial u/\partial x\). Newton’s second law for a short element gives

\[ \rho S\,dx\,\frac{\partial^2u}{\partial t^2} =ES\,dx\,\frac{\partial^2u}{\partial x^2}. \]

Thus longitudinal waves travel at

\[ v=\sqrt{\frac{E}{\rho}}. \]

The fixed end has zero displacement, while the free end has zero longitudinal stress:

\[ u(0,t)=0, \qquad \left.\frac{\partial u}{\partial x}\right|_{x=l}=0. \]

For a mode with spatial dependence \(\sin(kx)\), the free-end condition requires

\[ \cos(kl)=0. \]

For the fundamental mode, \(kl=\pi/2\), so the rod is one quarter of a wavelength long. Hence

\[ \omega_0=vk=\frac{\pi}{2l}\sqrt{\frac{E}{\rho}}. \]

Numerically,

\[ v=\sqrt{\frac{7.1\times10^{10}}{2700}} =5.128\times10^3\ \mathrm{m\,s^{-1}}, \]

and, for \(l=1\ \mathrm{m}\),

\[ \omega_0=\frac{\pi}{2}(5.128\times10^3) =8.055\times10^3\ \mathrm{rad\,s^{-1}}. \]

For linear damping, the observed angular frequency is \(\omega=\sqrt{\omega_0^2-\mu^2}\). Since \(\mu\ll\omega_0\), this correction is negligible:

\[ \boxed{\omega\simeq8.06\times10^3\ \mathrm{rad\,s^{-1}}}. \]

The corresponding frequency is

\[ f=\frac{\omega}{2\pi}=\frac{v}{4l} \simeq1.282\times10^3\ \mathrm{Hz}. \]

A(c)

For rods made of the same material and oscillating in the same fundamental mode,

\[ f(l)=\frac{v}{4l}. \]

Consequently, their beat frequency is

\[ f_{\mathrm b}=|f_1-f_2| =\frac{v}{4}\left|\frac1{l_1}-\frac1{l_2}\right|. \]

For a small length difference \(\delta l=|l_1-l_2|\), with both lengths approximately \(l\),

\[ f_{\mathrm b}\simeq\frac{v}{4l^2}\delta l =\frac{f}{l}\delta l. \]

Therefore,

\[ \delta l\simeq\frac{l f_{\mathrm b}}{f} =\frac{(1)(0.005)}{1281.998}\ \mathrm{m}, \]

giving

\[ \boxed{\delta l\simeq3.90\times10^{-6}\ \mathrm{m}}. \]

The ratio \(\delta l/l\simeq3.9\times10^{-6}\) confirms the validity of the small-difference approximation.

A(d)

Use linear elasticity and a quasi-static gravitational perturbation. Let \(x\) run from the fixed end, \(x=0\), to the free end, \(x=l\), and let \(\Delta g\) denote the signed component of the field change along this direction.

The material beyond a cross-section at \(x\) has mass

\[ m(x)=\rho S(l-x). \]

The additional axial force transmitted through that section is therefore

\[ \Delta F(x)=\rho S(l-x)\Delta g. \]

By Young’s modulus, the change in local strain is

\[ \Delta\epsilon(x)=\frac{\Delta F(x)}{ES} =\frac{\rho(l-x)\Delta g}{E}. \]

Integrating the extension of all elements,

\[ \Delta l =\int_0^l\Delta\epsilon(x)\,dx =\frac{\rho\Delta g}{E}\int_0^l(l-x)\,dx. \]

Thus

\[ \boxed{\Delta l=\frac{\rho l^2\Delta g}{2E}}. \]

A field change directed towards the fixed end gives compression, represented by a negative \(\Delta g\).

A(e)

A displacement \(\Delta l\) of a reflecting end changes the round-trip optical path by \(2\Delta l\). Taking the field change along one rod, with the other acting as the longitudinal reference,

\[ |\Delta D|=2|\Delta l|. \]

A change of optical path by one wavelength moves the interference condition through one fringe. The stated sensitivity therefore corresponds to

\[ |\Delta D|_{\min}=10^{-4}\lambda =10^{-4}(656\times10^{-9}) =6.56\times10^{-11}\ \mathrm{m}. \]

The minimum detectable rod-end displacement is consequently

\[ |\Delta l|_{\min}=3.28\times10^{-11}\ \mathrm{m}. \]

Using A(d),

\[ 2|\Delta l|=\frac{\rho l^2|\Delta g|}{E}. \]

At the detection threshold,

\[ l_{\min} =\sqrt{\frac{E\,10^{-4}\lambda}{\rho|\Delta g|}}. \]

Substitution gives

\[ l_{\min} =\sqrt{ \frac{(7.1\times10^{10})(10^{-4})(656\times10^{-9})} {(2700)(10^{-19})} }, \]

so

\[ \boxed{l_{\min}\simeq1.31\times10^8\ \mathrm{m}}. \]

More generally, if both rods respond, \(\Delta g\) in this expression is replaced by the difference between their signed axial field changes.

A(f)

In the classical harmonic approximation, an atomic displacement \(x\) has elastic potential energy \(\tfrac12\kappa x^2\). Equipartition gives

\[ \frac12\kappa\langle x^2\rangle=\frac12 k_{\mathrm B}T. \]

For unchanged effective stiffness \(\kappa\), the characteristic vibration amplitude therefore scales as

\[ A\propto\sqrt{T}. \]

Since \(100\ \mathrm{mK}=0.100\ \mathrm{K}\),

\[ \frac{A_{\mathrm{cold}}}{A_{\mathrm{warm}}} =\sqrt{\frac{0.100}{300}} =\frac1{\sqrt{3000}} =0.0183. \]

Thus the amplitude is reduced by a factor

\[ \boxed{\frac{A_{\mathrm{warm}}}{A_{\mathrm{cold}}} =\sqrt{3000}\simeq54.8}. \]

This estimate uses classical thermal vibrations and neglects quantum zero-point motion.

A(g)

Let

\[ T_i=4.2\ \mathrm{K},\qquad T_f=0.100\ \mathrm{K},\qquad s_i=0.072\ \mathrm{J\,kg^{-1}\,K^{-1}}. \]

The specified heat-capacity law is

\[ s(T)=s_i\left(\frac{T}{T_i}\right)^3. \]

The heat that must be removed from the sphere is

\[ Q=m\int_{T_f}^{T_i}s(T)\,dT =\frac{ms_i}{T_i^3}\int_{T_f}^{T_i}T^3\,dT. \]

Therefore,

\[ Q=\frac{ms_i}{4T_i^3}(T_i^4-T_f^4) =\frac{ms_iT_i}{4} \left[1-\left(\frac{T_f}{T_i}\right)^4\right]. \]

Numerically,

\[ Q= \frac{(1168)(0.072)(4.2)}4 \left[1-\left(\frac{0.100}{4.2}\right)^4\right] \simeq88.3\ \mathrm{J}. \]

At average cooling power \(P=1.00\times10^{-3}\ \mathrm{W}\),

\[ t=\frac QP =\frac{88.3}{10^{-3}}\ \mathrm{s}. \]

Hence

\[ \boxed{t\simeq8.83\times10^4\ \mathrm{s} \simeq24.5\ \mathrm{h}}. \]

B(a)

Let the emitted frequency be \(\nu_R\). Following the prescribed Newtonian model, assign the photon an energy-equivalent gravitational mass

\[ m_\gamma=\frac{h\nu_R}{c^2}. \]

The work required to move this mass from the solar surface to infinity is

\[ W=\int_R^\infty\frac{GMm_\gamma}{r^2}\,dr =\frac{GMm_\gamma}{R}. \]

Conservation of energy gives

\[ h\nu_\infty=h\nu_R-\frac{GMm_\gamma}{R}. \]

Substituting \(m_\gamma=h\nu_R/c^2\),

\[ h\nu_\infty =h\nu_R\left(1-\frac{GM}{Rc^2}\right), \]

and therefore

\[ \boxed{\nu_\infty\simeq \nu_R\left(1-\frac{GM}{Rc^2}\right)}. \]

This is a first-order weak-field result. Using the emitted energy to calculate \(m_\gamma\) introduces only higher-order corrections. For the Sun,

\[ \frac{GM}{Rc^2}=2.12\times10^{-6}\ll1, \]

so the approximation is well justified.

B(b)

Define

\[ \epsilon=\frac{GM}{rc^2}, \qquad q=1-\epsilon. \]

The frequency reduction by \(q\) corresponds to an increase in the clock period by \(1/q\). According to the length-contraction rule supplied in the question, the associated length scale decreases by \(q\).

Thus, in the problem’s comparison with distant coordinates, the effective speed scales as

\[ \frac{c'_r}{c}\simeq\frac{q}{1/q}=q^2. \]

It follows that

\[ n_r=\frac{c}{c'_r} \simeq\frac1{(1-\epsilon)^2}. \]

Expanding to first order,

\[ (1-\epsilon)^{-2}=1+2\epsilon+O(\epsilon^2). \]

Therefore,

\[ \boxed{n_r\simeq1+\frac{2GM}{rc^2}}, \qquad \boxed{\alpha=2}. \]

Here \(c'_r\) is the effective coordinate speed used in the question’s refractive-index model. A local measurement using proper lengths and times still gives the vacuum speed \(c\).

B(c)

Take the incoming light ray to travel approximately along the \(x\)-direction, with impact parameter \(b\). Along its unperturbed path,

\[ y=b,\qquad r=\sqrt{x^2+b^2}. \]

The refractive index is

\[ n(x,y)=1+\frac{\alpha GM}{c^2\sqrt{x^2+y^2}}. \]

Snell’s law for a continuously varying refractive index gives the ray equation

\[ \frac{d}{ds}(n\hat{\boldsymbol t})=\boldsymbol\nabla n, \]

where \(\hat{\boldsymbol t}\) is the ray’s unit tangent. For a small deflection, \(ds\simeq dx\) and \(n\simeq1\), so its transverse component becomes

\[ \frac{d\theta}{dx}\simeq\frac{\partial n}{\partial y}. \]

Differentiation gives

\[ \frac{\partial n}{\partial y} =-\frac{\alpha GM\,y} {c^2(x^2+y^2)^{3/2}}. \]

To first order, this may be evaluated along the straight path \(y=b\):

\[ \Delta\theta =-\frac{\alpha GMb}{c^2} \int_{-\infty}^{+\infty} \frac{dx}{(x^2+b^2)^{3/2}}. \]

Using the supplied integral,

\[ \Delta\theta =-\frac{\alpha GMb}{c^2}\frac2{b^2} =-\frac{2\alpha GM}{bc^2}. \]

The negative sign indicates bending towards the Sun. For a ray grazing its edge, \(b\simeq R\), and \(\alpha=2\), giving the deflection magnitude

\[ \theta=\frac{4GM}{Rc^2}. \]

Numerically,

\[ \theta= \frac{4(6.67\times10^{-11})(1.99\times10^{30})} {(6.95\times10^8)(3.00\times10^8)^2}. \]

Thus

\[ \boxed{\theta=8.49\times10^{-6}\ \mathrm{rad} \quad\text{towards the Sun}}. \]

This is approximately \(1.75\) arcseconds.

The solution was checked against the official solution.