## Abstract

The measurement of the spectral broadening, or temporal coherence property of very narrow linewidth lasers is not an easy task, while such a measurement is essential in any interferometric applications of the lasers. The beat note between two assumingly identical lasers only provides the convolutional spectral profile of the two lasers, but not characterizes the single laser. The delayed self-heterodyne interferometer (DSHI) would not be effective for kHz linewidth range because the finite delay cannot realize complete de-correlation. Here, we demonstrate, for the first time to our knowledge, the complete characterization of the modulus of the degree of coherence (DOC) of kHz linewidth lasers, with a self-referenced fashion where any other reference beam is not used, accordingly, characterize the spectral profile. The method is based on speckle statistical analysis of the Rayleigh scattering in the coherent fiber reflectometry, and would be a novel strong tool to characterize very narrow linewidth lasers.

© 2012 OSA

## 1. Introduction

Fourier spectroscopy [1–3] is a powerful tool particularly in relation to infrared spectroscopy. As with any types of spectrometric diagnosis, the resolution is determined by the maximum delay, hence, the size of the interferometer. This fact imposes a large burden when deploying the technique with narrow bandwidth lasers. Here, we present a novel Fourier spectroscopic approach to evaluate the modulus of the degree of coherence (DOC) over a very long range by using a speckle analysis in coherent fiber reflectometry. The term “quasi-“ in the title is used to indicate the missing phase modulation of the DOC, with the result that only the even-functional information of the spectral profile is observed. We have already provided a preliminary result and intuitive theoretical description [4], and here we confirm the validity of the technique in a delay of hundreds of microseconds (several tens of km in a vacuum), which enables kHz-level resolution.

The laser speckle is a well-known phenomenon closely related to the coherence of light [5–8], and it has been intensively investigated in such areas as imaging applications, metrology for detecting displacement, deformation, and vibration. Generally, since the ability to form speckle patterns stems from a coherent interaction, an appropriate optics set-up can measure the DOC by monitoring the speckle’s contrast [9, 10]. In silica optical fibres, a micro-size fluctuation in refractive index originates from such amorphous nature features as “rough surfaces”. In particular, our interest here is focused on a reflectometric measurement of such scattered lights (Rayleigh scattering) in single-mode fiber (SMF) [11]. With coherent fibre reflectometry, the reflected intensity is jagged along the propagation direction [12]. It is known that observing the alternation of the speckle pattern with respect to the reference pattern allows us to measure local changes in the optical length of the fibre caused by strain and/or temperature [13].

## 2. Theory of principle

The system set-up is shown in Fig. 1
. The beam emitted from the laser under the test (LUT) is used as a light source for coherent optical frequency domain reflectometry (C-OFDR) [13–16], which observes the Rayleigh backscattered beam in the fiber as a function of the distance or roundtrip delay to the scattering point in a SMF. The linear frequency sweep of the LUT is accomplished by sweeping the modulation sideband of an external single-sideband (SSB) modulator. Let ${a}_{q}(t){e}^{j2\pi ({\nu}_{q}+gt/2)t}$ be the modulated beam, i.e., the input into the SMF, where *g* is the sweep rate (Hz/s), ${\nu}_{q}$ is the optical frequency, and ${a}_{q}(t)={e}^{j{\theta}_{q}(t)}$ is the random fluctuation of the phase of the laser beam, which degrades the coherence. A single point reflection at the roundtrip time *τ'* in the fiber is detected as the spectrum (Fourier transfer), ${F}_{q,\tau \text{'}}(g\tau ),$ of the beat signal between the local beam and the reflected beam (replica of the local beam delayed by *τ*), $a{}_{q}(\tau \text{'}){a}_{q}^{*}(t-\tau \text{'}){e}^{j(2\pi g\tau t+2\pi {\nu}_{q}\tau \text{'}-\pi g\tau {\text{'}}^{2})}.$ The amplitude of the speckle, ${e}_{q}(\tau )$ observed at *τ* is the superposition of the spectrum, ${F}_{q,\tau \text{'}}(g\tau ),$ weighted by the random but time-invariant scattering coefficient $r(\tau \text{'}),$ where $\tau \text{'}$ is a scattering point located in the vicinity of *τ* [12].

Suppose that another separate but identical measurement is accomplished, represented by the suffix “*s*”, with respect to “*q*”. The sole difference is the fluctuation of the tested beam represented by ${a}_{q}(t)={e}^{j{\theta}_{q}(t)}$ and ${a}_{s}(t)={e}^{j{\theta}_{s}(t)}:$ the two beams are modulated with an identical waveform, and the speckle amplitudes ${e}_{(q,s)}(\tau )$ are superposition of ${F}_{(q,s),\tau \text{'}}(g\tau )$ with an identical random coefficient, $r(\tau \text{'})$. Therefore, their correlation $\overline{{e}_{q}(\tau ){e}_{s}^{*}(\tau )}$ should directly reflect that between ${F}_{q,\tau \text{'}}(g\tau )$ and ${F}_{s,\tau \text{'}}(g\tau ),$ or between their inverse Fourier transforms (Parceval’s theorem). Thus, we obtain

At the final stage, the exchange of $\tau \text{'}$ for τ is based on the assumption that the correlation is constant in the vicinity of $\tau \text{'}$. If the integration time *T* is sufficiently long compared with the coherence time so that the time average equals the ensemble average, and the average product of the independent *q*-th and *s*-th processes equals the product of each average. Hence, we obtain

## 3. Experimental set-up and laser under test

Two fiber lasers and a laser diode emitting at 1.55 μm were prepared as LUTs. LUT#1 was an NKT Photonics Koheras E15, LUT#2 was an Orbits Lightwave Ethernal and LUT#3 was a RIO PLANEX (RIO0194-1-01-1). The linewidth (full width at half maximum, FWHM) estimated by the self-delayed heterodyne method with a 160 km delay was about 110 kHz in LUT#3.

The frequency of the LUT was swept by using a single sideband (SSB) modulator. A frequency synthesizer with a ramp function (Agilent E8257D) generated the sinusoidal RF signal whose frequency was linearly swept. The modulation sideband of the optical beam was swept accordingly.

The sweep rate *g* and integration time *T*, which determine the sweep range *ΔF = gT*, can be selected very flexibly, but the following considerations are necessary. First, *T* must be sufficiently large compared with the expected coherence time so that it involves every random fluctuation. To realize this, we tested several different *T*s, and looked for one that seemed to provide sufficient randomisation. The second issue stems from the requirement that the DOC should be constant over the range in which the speckle correlation is obtained. The reflectometer's resolution is *Δτ* = 1/*ΔF*. If we use *N* samples, or the *NΔτ* range to obtain the correlation, the spectral width *Δν* of the LUT must be smaller than 1/ *NΔτ* so that the DOC is constant in the range. Hence, *ΔF > NΔν* is required. Therefore, if the sweep is accomplished with an electrical modulator, its bandwidth restricts the maximum spectral width that can be measured. The parameters used in the experiment are shown in the figure captions.

The output powers of the LUTs were typically 10 dBm. The SSB modulator imposed a total loss of ~-19 dB, so an optical amplifier was used to obtain a power of ~6 dBm at the entrance of the SMF. The C-OFDR provided a signal to noise ratio of about 25 dB even at a 40 km range.

The measurements were performed every 0.2 s, and a succeeding pair of obtained speckle patterns was used for the correlation analysis. The x- and y- polarizations of the scattering were obtained with polarization diversity, and the speckle correlation was analysed by using the inner product of the Jones vectors,

## 4. Experimental results

We tested three LUTs experimentally. Figure 2
shows an example of the observed reflection speckle at delays both shorter and longer than the laser's expected coherence time (*τ _{c}*). Since the SMF we used was a non-polarization-hold type, the polarization diversity was incorporated in the coherent receiver. The intensity plot of the scattering is the square sum of the x- and y-polarizations. The blue plots show the results of the first (

*q*-th) measurement, and red dashed plots shows those of the second (

*s*-th) measurement that was performed successively with the same LUT. As shown in the figure, when

*τ*<

*τ*, these two results exhibit a strong correlation, whereas when

_{c}*τ*>

*τ*, the correlation diminishes. This tendency was observed for all the LUTs used in the experiment, and was well predicted by the theoretical analysis.

_{c}The results of the correlation $\overline{{e}_{q}(\tau ){e}_{s}^{*}(\tau )}$ for the three LUTs are shown in Fig. 3 . The real and imaginary parts of the correlation are displayed in Fig. 3(a). Their correlation value should theoretically be real, but the practical value was complex and includes oscillation, as shown in the figures. The oscillation may be attributed to the bias of the laser frequency or a change in the fiber temperature imposed between the two measurements ${e}_{q,s}(\tau )$. The modulus of the complex numbers is calculated from these real and imaginary parts, and shown in Fig. 3(b), where the modulus shows the squared DOC. Three typical measurement results for LUT#1 (light-green, blue and brown) and #2 (red, green and purple) are plotted, and the results for LUT#3 are plotted in orange in the same figure (also shown separately on a larger scale). These DOC curves decrease monotonically with respect to the delay, but we see that the rates of decrease are unique to each LUT.

By taking the squared-root of $\overline{{e}_{q}(\tau ){e}_{s}^{*}(\tau )},$ we obtain the modulus of the DOC. From the Wiener-Khinchin relationship, the even-functional information of the power spectrum can be obtained, while the lack of the phase of the DOC veils the odd-functional information. Spectral profiles thus obtained are depicted in Fig. 4 . The measurements were performed every 0.2 s, and the succeeding pair of speckle charts were processed to examine their correlation. We found no significant deviation of the observed spectra for the same LUT throughout the measurement. The full-widths at half-maximum (FWHM) were about 6, 10, and 120 kHz, for LUT#1 to #3, respectively.

To examine the dependence of the above results on the sampling time *T*, we collected data for several *T*s, as shown in Fig. 5
. The observed tendency in LUT#1 in particular, which has the longest expected coherence time of the three, is that the obtained $\overline{{e}_{q}(\tau ){e}_{s}^{*}(\tau )}$ curves exhibited the curvature perturbation in every experiment when the integration time *T* was short, while the perturbation disappeared when *T* was long. As described in the theoretical consideration, we presumed that the observed perturbation is a result of the fact that there is insufficient sampling time to contain all possible random events. In such cases, the light sometimes appears coherent even with very large delays, while the interference may diminish at a specific delay, and the cases are dependent on each individual event.

Finally, to examine the validity of the results, we conducted a reference experiment, in which we used a Mach-Zehnder interferometer with a few fixed delays. Figure 6(a) shows the reference measurement set-up we used to characterize the phase noise of a laser using a Mach-Zehnder interferometer with a fixed delay. The continuous light produced from the laser under test (LUT) was divided into two paths. One path was delayed by time τ through a single-mode fiber (SMF), and frequency-shifted by an acousto-optical frequency shifter (AOF) $\Delta {\omega}_{0}$ to allow the use of an ac-coupled photodetector. The two beams were mixed by a 90-degree hybrid to produce the constellation diagram. After removing the constant rotation, we evaluated the phase variance $\sigma {\left(\tau \right)}^{2}$ at a few fixed delays (0 ~200 μs). The DOC was estimated by

By obtaining the phase-error variance at delays of 0, ~25, ~50, ~100 and ~200 μs, we were able to analyze the DOC as indicated within the red and grey bars, for LUT#1 and #2, respectively, for each delay in Fig. 3(b). We can clearly see that there is very good agreement with the results obtained with the proposed method. In addition, we describe the effect of the mechanical vibration of the SMF, which could lead to an underestimation of the coherence. The amplitude of the sound pressure density was typically 58 dB (16 mPa) outside the soundproof box, and 54 dB (10 mPa) inside it. SMFs with identical 40 km lengths (~200 μs delay) were inserted in both paths of the interferometer. The two SMFs were placed apart from each other in different soundproof boxes, so we believe that their vibrations were uncorrelated. The observed phase variance was much smaller than that observed when the path length was different and laser’s phase noise was incorporated.

We examined the phase variance caused by the mechanical vibration of SMFs with identical 40 km lengths (~200 μs delay) in soundproof boxes (Fig. 7(b)
). The amplitude of the sound pressure density was approximately 54 dB (10 mPa) in the soundproof box. Clearly, the broadening of the constellation is much smaller than that caused by the LUT’s phase fluctuation. On the assumption that the vibrations of the two SMFs are uncorrelated, we estimate the variance caused by a single SMF to be ~0.1 rad^{2}; this corresponds to ${\left|\gamma \left(\tau \right)\right|}^{2}$≅ 0.95, whereas the observed DOC at the corresponding delay was almost eliminated. Thus, we conclude that the DOC was slightly underestimated, but the effect was not significant.

## 5. Conclusion

In summary, we have described a Fourier spectroscopic measurement with a range of several hundreds of microseconds with the aim of characterizing the coherence of narrowband laser beams, and have successfully determined the spectral profile of kHz-level lasers. The unique feature of this new technique makes it attractive for characterizing narrowband beams. This is because the existing technique using a delayed self-heterodyne interferometer [18, 19], which only observes a beat spectrum with a finitely delayed portion, includes a fundamental error when it is adapted for such narrowband sources.

## References and links

**1. **P. Fellgett, “Thesis,” in University of Cambridge, (1951).

**2. **P. L. Richards, “High-resolution Fourier transform spectroscopy in the far-infrared,” J. Opt. Soc. Am. **54**(12), 1474–1484 (1964). [CrossRef]

**3. **J. Connes and P. Connes, “Near-infrared planetary spectra by Fourier spectroscopy. I. Instruments and results,” J. Opt. Soc. Am. **56**(7), 896–910 (1966). [CrossRef]

**4. **M. Inoue, Y. Koshikiya, X. Fan, and F. Ito, “Coherence characterization of narrow-linewidth beam by C-OFDR based Rayleigh speckle analysis,” Opt. Express **19**(21), 19790–19796 (2011). [CrossRef] [PubMed]

**5. **J. D. Rigden and E. I. Gorden, “The granularity of scattered optical maser light,” *Proc. IRE***50**, 2367–2368 (1962).

**6. **B. M. Oliver, “Sparkling spots and random diffraction,” Proc. IEEE **51**(1), 220–221 (1963). [CrossRef]

**7. **J. W. Goodman, “Speckle phenomena in optics,” in *Roberts and Company*, (Englewood, 2007).

**8. **J. C. Dainty, “Some statistical properties of random speckle patterns in coherent and partially coherent illumination,” Opt. Acta (Lond.) **17**(10), 761–772 (1970). [CrossRef]

**9. **F. M. Mottier and R. Dandliker, “A simple spectrum analyzer for laser light using speckles,” Opt. Commun. **3**(5), 366–368 (1971). [CrossRef]

**10. **R. A. Dandliker and F. M. Mottier, “Determination of coherence length from speckle contrast on a rough surface,” Z. Angew. Math. Phys. **22**(3), 369–381 (1971) (ZAMP). [CrossRef]

**11. **E. Brinkmeyer, “Backscattering in single-mode fibres,” Electron. Lett. **16**(9), 329–330 (1980). [CrossRef]

**12. **P. Healey, “Fading in heterodyne OTDR,” Electron. Lett. **20**(1), 30–32 (1984). [CrossRef]

**13. **M. Froggatt and J. Moore, “High-spatial-resolution distributed strain measurement in optical fiber with Rayleigh scatter,” Appl. Opt. **37**(10), 1735–1740 (1998). [CrossRef] [PubMed]

**14. **W. Eickhoff and R. Ulrich, “Optical frequency domain reflectometry in single-mode fiber,” Appl. Phys. Lett. **39**(9), 693–695 (1981). [CrossRef]

**15. **K. Tsuji, K. Shimizu, T. Horiguchi, and Y. Koyamada, “Coherent optical frequency domain reflectometry using phase-decorrelated reflected and reference lightwaves,” J. Lightwave Technol. **15**(7), 1102–1109 (1997). [CrossRef]

**16. **X. Fan, Y. Koshikiya, and F. Ito, “Phase-noise-compensated optical frequency domain reflectometry with measurement range beyond laser coherence length realized using concatenative reference method,” Opt. Lett. **32**(22), 3227–3229 (2007). [CrossRef] [PubMed]

**17. **M. Born and E. Wolf, *Principles of Optics*, 7th ed. (Cambridge University Press, 1999).

**18. **T. Okoshi, K. Kikuchi, and A. Nakayama, “Novel method for high resolution measurement of laser output spectrum,” Electron. Lett. **16**(16), 630–631 (1980). [CrossRef]

**19. **L. E. Richter, H. I. Mandelberg, M. S. Kruger, and P. A. McGrath, “Linewidth determination from self-heterodyne measurements with subcoherence delay times,” IEEE J. Quantum Electron. **22**(11), 2070–2074 (1986). [CrossRef]