EUREKA MARCH 2001 COVER STORY

Light measures light with femto precision

Fibre optics show extraordinary advances in the precision measurement of strain, temperature and, under the same development umbrella, the wavelength of light. Tom Shelley reports

The wavelength of light, the building block for all kinds of advanced communications and measurement systems, can now be measured to ten femtometres – about one ten thousandth of the width of an atom.


Measuring instrument screen display

The technique used depends on the analysis of a series of spectral lines (and much fainter lines on each side caused by scattering effects) produced by reflected light emerging from a long length of optical fibre.

This analysis is performed using a Fabry Perot (FP) interferometer. It comes as a spinoff development of a tool for measuring strain and temperatures over distances of up to 10km. Distributed sensing systems have already been installed down oil wells and in dams and customers in communications are clamouring for the wavelength meters.

The team developing the technologies, headed by Mahmoud Farhadiroushan and Dr Tom Parker, trades as independent company, Sensornet.

The faint lines, called Brillouin peaks, are caused by light being inelastically scattered by acoustic phonons in the glass. They will either acquire or lose the energy of one phonon (Phonons are quantum packets of natural vibration just as photons are quantum packets of light).

The height of the lines is dependent on the phonon population and the temperature – the higher the temperature, the more the glass atoms move about and so the more phonons there are. The spacing of the lines from the central peaks depends on the magnitude of the phonons, which is affected by the strain in the glass. If the glass is embedded in some larger mass of material it will also record the strain in the material under study. The central lines, called Rayleigh peaks, represent light which is elastically 'backscattered'.

Precision measurement

The wavelength measuring technique has spun directly out of the strain measurement research. The team was trying to purchase a suitable laser and found it had to develop its own technique to discover how much frequency variation there was between individual pulses from the same laser.

The first challenge was to turn the laser light pulses, which may only be 10mm long, into a pseudo-continuous signal. This is done by sending the light into a very long length of optical fibre and looking at the light which is 'backscattered' from different distances along the fibre, which will arrive at different times.

As in conventional analysis methods, the technique measures the apparent separation of spectral lines from a test and a reference light source as seen in an FP device. Such devices are routinely used to measure light wavelength and frequency but there is always a problem in deciding which number of reflections a particular line has gone through between the two mirrors in the interferometer.

In the new technique a comparison is made between the test and reference light sources and the fact that the Brillouin shift in frequency is inversely proportional to the source wavelength is used. The shift can only be resolved to 0.03nm. This is, however, sufficient to find the order of any multiply reflected line and establish which one, or which order, it has.

One way in which the system is made more sensitive is to turn the fibre into a laser, injecting light at the Brillouin frequency. The fibre then resonates, producing a very much larger Brillouin peak. This is possible because the light is modulated in the range 1 to 20 GHz, generating side bands whose frequency can be very precisely controlled.

The final measurement required is the calibration of the FP device. This is achieved by directing modulated light in to the unit. As the modulation frequency is swept upwards, along the x-axis of a computer-based plotting system, positions of all the detected modulation (side band) peaks are recorded. By calculating values as slopes of plots, determined from a large number of measurements, it is possible to calculate absolute wavelength to the required (approximate) 10fm accuracy. Despite the complexity of the method, the whole procedure can presently be completed in about ten minutes.

The economic importance of the device, first and foremost, is that it will allow fibre optic communication systems running at 1.5 micron wavelengths to pack in even more channels at even closer frequencies. The other aspect to its importance is in very accurate distance measurement – of increasing importance in microelectronics which usually depend on optical interferometric techniques. Knowing light wavelengths and source variability to even greater degrees of accuracy will bring about commensurate increases in distance measurement accuracy.

Sensornet has currently won SMART, London Business Innovation and Metrology Awards. The FemtoMeter is currently in the process of being turned into a packaged instrument.

Strain, pressure and temperature

Commercially available fibre optic systems are able to measure temperature distributions. However, it has not been possible to measure temperature and strain at the same time.

Conventional systems rely on measuring Raman scattering by optical phonons. Sensornet, however, knows that the height of Brillouin peaks increases by 0.3 per cent for each degree centigrade rise in temperature and that their precise position depends on strain.

In a system, experimentally tested with a view to use in measuring pressure in oil pipelines, light from a distributed feedback laser (DFB) was externally modulated to produce 100ns pulses at a repetition rate of 20kHz.

The pulsed light was coupled into an erbium-doped fibre amplifier which amplified it to give pulses of around 1 to 10W. The amplified light output was then coupled, first into two 100m lengths of reference fibre and then into the sensing fibre. The signal was spectrally and temporally analysed using a 10GHz Free Spectral Range (FSR) scanning FP.

The technique involved scanning the FP relatively slowly (2Hz) so that data was effectively stored as a series of optical time domain reflectometry (OTDR) traces – each one taken for a different optical frequency. The traces were then re-arranged to form frequency spectra corresponding to the Brillouin spectrum at various locations in the fibre. A peak detection system measured the amplitudes and frequencies of the peaks. From this temperature, strain and pressure could be deduced.

Since then refinement and improvement has taken place and the team is "very confident" that it will soon be able to demonstrate the measurement of temperatures to an accuracy of 1 deg C and strain to an accuracy of 20micro strains. Measurement and averaging times are down to 10 to 15 minutes instead of an hour and it is expected to be possible to spatially locate temperature and strain distributions to within tens of metres in a 10km-long fibre.

Test fibres are currently deployed in two dams in northern Sweden. The system is also on test on oil wells in California, where flow is being maintained by steam injection. There is a recognised need in the oil industry to be able to monitor strains down hole, since taking oil out leads to considerable ground movements which may risk damaging well casings and other equipment.

Sensornet is engaged in further developing and packaging its distributed strain and temperature measuring systems for commercial deployment. The company has filed a number of patent applications. www.sensornet.co.uk

Design Pointers

The wavelength of pulsed or continuous infra red light can be measured to an accuracy of 10fm.

The technique is expected to greatly increase the number of data channels which can be sent down a single optical fibre

It should be possible to simultaneously measure temperature and strain to an accuracy of 1 deg C and 20 micro strains respectively, with measurement times of 10 to 15 minutes.

Return to list of stories page