IPhO 1976 · Theory problem 3

Measuring an air bubble inside a glass sphere

Somewhere in a glass sphere there is an air bubble. Describe methods how to determine the diameter of the bubble without damaging the sphere.

Official English text of the problem.

Answers

Problem 3. Three possible nondestructive methods are:

  • Weighing: Measure the sphere’s outer diameter \(D\) and mass \(m\). If the glass density is \(\rho_g\), neglecting the mass of air, \[ \boxed{d=\left(D^3-\frac{6m}{\pi\rho_g}\right)^{1/3}}. \]
  • Optical immersion: Immerse the sphere in a liquid with the same refractive index as the glass. Measure the bubble’s transverse outline with calibrated optics. If its image diameter is \(L\) and the transverse magnification is \(M\), \[ \boxed{d=\frac{L}{M}}. \]
  • Radiography: Use a parallel X-ray beam and measure the bubble’s projected diameter. With a calibrated image scale, this gives its actual diameter directly.

No numerical data are supplied.

Solution

We assume that the glass is homogeneous and that the bubble is spherical, as is implicit in asking for its diameter. Write the outer radius as \(R=D/2\) and the bubble radius as \(r=d/2\). The bubble need not be at the centre of the glass sphere.

Weighing

Measure \(D\) with calipers and \(m\) with a balance. The glass density must be known independently, for example from a separate specimen of the same glass.

The outer volume and bubble volume are \[ V=\frac{\pi D^3}{6}, \qquad v=\frac{\pi d^3}{6}. \] If the density of the air inside the bubble is \(\rho_a\), conservation of mass gives \[ m=\rho_g(V-v)+\rho_a v =\rho_gV-(\rho_g-\rho_a)v. \] Consequently, \[ \boxed{ d=\left[ \frac{\rho_gD^3-6m/\pi}{\rho_g-\rho_a} \right]^{1/3} }. \] Ordinarily \(\rho_a\ll\rho_g\), giving \[ \boxed{d=\left(D^3-\frac{6m}{\pi\rho_g}\right)^{1/3}}. \]

The result is independent of the bubble’s position. For a small bubble, this method requires precise measurements: the missing mass is only approximately the fraction \((d/D)^3\) of the mass of a solid glass sphere.

Optical measurement in an index-matching liquid

Place the sphere in a transparent tank containing a liquid whose refractive index equals that of the glass. Use monochromatic light if necessary so that dispersion does not prevent accurate matching.

At the outer surface, Snell’s law becomes \[ n_g\sin i=n_\ell\sin t. \] Since \(n_\ell=n_g\), we have \(t=i\): the outer spherical boundary no longer bends the rays. The bubble remains visible because the air and glass have different refractive indices.

View through a flat tank window, perpendicular to the viewing direction. Use parallel viewing rays—for example, a telecentric imaging system—or a travelling microscope whose axis remains perpendicular to the window.

The limiting rays defining the bubble’s outline are tangent to the bubble. Approached from outside the outline, these rays remain entirely in the glass and matching liquid, so they are not refracted by the air cavity. The perpendicular projection of a sphere is a circle of its true diameter, regardless of its depth.

Thus, if a travelling microscope records the two opposite edges at transverse positions \(x_L\) and \(x_R\), \[ \boxed{d=|x_R-x_L|}. \] Alternatively, a calibrated image with transverse magnification \(M\) gives \[ \boxed{d=L/M}. \]

X-ray transmission

Illuminate the sphere with a sufficiently penetrating, approximately parallel X-ray beam and record its transmitted intensity. Neglect ray bending and scattering. Air attenuates the beam much less than glass, so the bubble produces a region of enhanced transmission.

For a ray at transverse distance \(s\) from the projected bubble centre, the length of glass replaced by air is \[ \ell_b(s)= \begin{cases} 2\sqrt{r^2-s^2},&s<r,\\ 0,&s\ge r. \end{cases} \] If \(\mu_g\) and \(\mu_a\) are the linear attenuation coefficients, the Beer–Lambert law gives \[ \frac{I}{I_{\mathrm{solid}}} =\exp\!\left[(\mu_g-\mu_a)\ell_b(s)\right], \] where \(I_{\mathrm{solid}}\) is the transmission through an otherwise identical solid glass sphere.

The enhanced-transmission region therefore ends at \(s=r\), giving a circular projection of diameter \(d\). With parallel rays, the bubble’s position along the beam does not affect its projected size. If the recorded outer-sphere and bubble diameters are \(L_s\) and \(L_b\), respectively, \[ \boxed{d=D\frac{L_b}{L_s}}. \]

The solution was checked against the official solution.