Given the absolute pressures of air on both sides of a titanium membrane of known thickness (0.1mm) and diameter (3mm), how would one calculate the rate of gas exchange/escape through the titanium material assuming a pressure gradient? Ideal units are in cc/sec.

  • 1
    $\begingroup$ Not sure how to approach this problem. But I do know the rate of diffusion across a metal membrane will also be a function of the membrane temperature. Such membranes are used as 'valves' to inject gas particles into particle accelerators. Perhaps if you research in that area. I imagine for 0.1 mm thickness the rate would be very low. $\endgroup$ – docscience Oct 13 '15 at 19:56

The Diffusion Flux of a gas through a metal membrane is theoretically given by:



$p_1$ and $p_2$ the pressures on the respective sides on the membrane.

$\delta$ the thickness of the metal membrane.

$K$ is a proportionality constant that connects gas concentration in the metal with pressure of the gas via:

$K^2=\frac{c^2}{p}$, where $c$ is the concentration of the gas in the metal and $p$ the pressure of the gas.

The main problem now becomes finding reliable values for $K$ and $D$. Both are likely to be very small, for example for hydrogen/iron Wikipedia lists a value of:

$D=1.66 \times 10^{-9}\:\mathrm{cm^2/s}$.

$K$ is also likely to be very small, as $c=K\sqrt{p}$ and the solubility of air in titanium metal is very small.

This means that at modest pressure differences and a membrane thickness of $\delta =1 \times 10^{-4}\:\mathrm{m}$, $J$ is likely to be negligible.

But even if the pressure was $100$ times larger on one side of the membrane than the other then J would still be only approx. $\sqrt{100}=10$ times higher.

A $0.1\:\mathrm{mm}$ thickness a titanium membrane is likely not to leak at all, at least not in 'reasonable' conditions of pressure.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.