Polarisation noise in supercontinuum


Supercontinuum light sources have traditionally been ignored in many OCT imaging applications owing to poor noise performance1,2. The development of super luminescent diodes (SLD) and tuneable lasers such as swept sources have seen potential growth and adoption for OCT systems in the last decades. However, many of these conventional light sources struggle to provide sufficient optical bandwidth needed to explore increasing demand in high resolution imaging.

As recently as in 2012, the use of supercontinuum sources for OCT has again been researched. The main argument this time around isn’t on noise but the promising imaging results SC sources can deliver based upon its broadband and high power performances.

A trade off between noise and bandwidth have resulted to massive deployment of SC sources in clinical applications. We at UBAPHODESA believe that the future of OCT relies heavily on the successful development of a low-noise version of the SC source. This is the reason why there are many concurrent projects of low-noise characterisation in our research group.

The ultimate aim is to design a high powered SC source capable of exploiting the noise performance and polarisation properties of existing SC sources for commercial use.

Polarisation noise

In most commercial laser sources, polarisation properties aren’t taken into account seriously. In conventional sources with narrow optical bandwidth and low power, there isn’t much interest in studying the polarisation effects as a specific polarisation can easily be manipulated at will. However, due to the very random nature of supercontinuum generation process, polarisation properties of produced light are hard to quantify and characterise. For that reason, this field has never been fully explored before.

We credit Z. Zhu and T. G. Brown3, for their discovery of interesting polarization properties in supercontinuum generation process. We take particular interest in their numerical simulation and earlier experimental results showing the different noise properties observed when pumping a birefringent photonic crystal fibre with different input pulse parameters.

At UBAPHODESA, we believe that by having access to NKT expertise in supercontinuum generation and combining Prof. Ole Bang’s group experience in numerical analysis, we could exploit the effect of polarisation to our advantage and to improve the noise performance of supercontinuum sources in imaging applications.

Initial findings

We have discovered so far the work on characterising polarisation noise in supercontinuum sources have tremendous benefits. The significances of our initial findings are:

  • Polarisation varies across SC spectrum
  • Stable polarisation extinction ratio of 10 dB
  • Distinct difference in intensity noise on long wavelengths


We characterise polarisation effects not only on polarisation extinction ratios (PERs), but also extensively on noise across the supercontinuum spectrum. One of our noise measurement methods involves the use of narrowband band pass filters (BPF) to evaluate noise properties at specific wavelengths of interest. We used BPF with pass band of 10 – 12 nm at full width at half maximum, at every 100 nm wavelengths interval between 500 nm and 2200 nm (visible/IR 450-1600 nm filters and mid-IR filters 1810-2310 nm). We have also studied the noise as two separate polarisation components by looking into the spectral response.

A broadband supercontinuum white light (BSCWL) source (SuperK Extreme EXR 20, NKT Photonics, Denmark) is used for this experiment. At the output of the source, a non-linear fibre (NLF) measuring 10 m was initially used. A test bench is set up for the measurement. The output beam from NLF is fed into a collimating lens and through a broadband double Glan-Taylor Calcite polarizer (350-2300 nm) mounted on a manual rotation mount. The polarizer is used to select the two orthogonal polarization components of light from the BSCWL source. A bandpass filter is placed after the polarizer to filter out a narrow band of the supercontinuum spectrum for noise analysis. The filtered light is sent into a high-speed photodiode connected to a GHz oscilloscope.

To facilitate broadband noise analysis, three separate DC/AC coupled photoreceivers (PRs) are used. They consist of a UV/visible PR for measurement in 450-1000 nm, a near IR PR for measurement in 1000-1700 nm (Si and InGaAs combined) as well as a long wavelength PR for measurement between 1700-2400 nm with an extended InGaAs receiver.

Supercontinuum output spectrum was sampled at several power levels using two optical spectrum analyzers (visible analyser from ANDO and model long-wavelength analyser from Yokogawa). As the supercontinuum spectrum spans across broad wavelength range, a good coverage from 400 nm up to 2400 nm is essential. On the long wavelength measurement, an additional long pass filter with cut off wavelength at 1250 nm was inserted after the polarizer in the system setup, to remove the effect of spectrum folding effect.


Figure: Illustration of system setup used to characterise noise and polarisation abnormity in the supercontinuum spectrum.

Noise power can be characterised by simply looking into the supercontinuum spectrum. The polarisation extinction ratio is simply the power difference between two polarisation components.


For our initial measurements, 12 sets of spectral data were collected and analysed. We deliberately sampled at three distinct pump power levels: full, two third, and one third. For each of these power settings, spectrum at four polarizer rotations was recorded and analysed.

Each of these angles is separated at 90 degrees apart. Two angles are orthogonal to each other. In the case of measuring the polarisation extinction ratio [dB], the spectrum with highest overall power spectral density [dBm/nm] is subtracted against the one with lower power.

Our spectral measurement shows no observable PER at wavelengths shorter than pump (λ < 1064 nm). At low pump power, Ppump = 500 mW, PER variation begins at wavelengths above 1200 nm and peaks at around 1450 nm. When pump power is tuned up in steps, we observed PER peaks are shifted further into the longer wavelength range. At maximum pump power, Ppump = 6 W, maximum PER was observed at around 2250 nm.

Futher results will be published here once they are reviewed.


Keywords: Supercontinuum, noise, optical coherence tomography, spectroscopy, ultrahigh resolution, swept source, SLD.


  1. J. Brown, S. Kim, and A. Wax, “Noise characterization of supercontinuum sources for low-coherence interferometry applications,” J. Opt. Soc. Am. A 31, 2703-2710 (2014) Link
  1. You, C. Wang, Y. Lin, A. Zaytsev, P. Xue and C. Pan, “Ultrahigh-resolution optical coherence tomography at 1.3 μm central wavelength by using a supercontinuum source pumped by noise-like pulses,” Laser Phys. Lett. 13 (2016) 025101 Link
  1. Zhu and T. G. Brown, “Experimental studies of polarization properties of supercontinua generated in a birefringent photonic crystal fiber,” Opt. Express 12, 791-796 (2004) Link
  1. Møller, S. T. Sørensen, C. Jakobsen, J. Johansen, P. M. Moselund, C. L. Thomsen, and O. Bang, “Power dependence of supercontinuum noise in uniform and tapered PCFs,” Opt. Express 20, 2851-2857 (2012) Link