Abstract
Coherent optical frequency domain reflectometry has been widely used to locate static reflectors with high spatial resolution. Here, we present a new type of Doppler optical frequency domain reflectometry that offers simultaneous measurement of the position and speed of moving objects. The system is exploited to track optically levitated “flying” particles inside a hollow-core photonic crystal fiber. As an example, we demonstrate distributed temperature sensing with sub-mm-scale spatial resolution and a standard deviation of ∼10°C up to 200°C.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Corrections
Max Koeppel, Abhinav Sharma, Jasper Podschus, Sanju Sundaramahalingam, Nicolas Y. Joly, Shangran Xie, Philip St. J. Russell, and Bernhard Schmauss, "Doppler optical frequency domain reflectometry for remote fiber sensing: erratum," Opt. Express 29, 24193-24193 (2021)https://opg.optica.org/oe/abstract.cfm?uri=oe-29-15-24193
1. Introduction
Coherent optical frequency domain reflectometry (COFDR) was originally proposed as an alternative to optical time domain reflectometry (OTDR) [1–3]. Since it offers a spatial resolution several orders of magnitude better than OTDR, it also allows the analysis of small-scale components and integrated optical waveguides [4–7]. The high resolution and sensitivity is useful in many different sensing applications, for example, distributed temperature or strain sensing based on Rayleigh scattering fingerprints [8–10].
In many sensing situations there is also a great need for dynamic measurements. Although COFDR systems for detecting vibrations or acoustic signals have been reported [11,12], they have not yet been extended to dynamic measurements of moving targets. Here, we report for the first time a Doppler COFDR system in which the instantaneous position and speed of moving objects are measured with µm-scale spatial resolution and speed accuracy in the µm/s-range. Using a custom-built dynamic OFDR system, the position and speed of optically levitated flying particles in a hollow-core photonic crystal fiber (HC-PCF) were continuously monitored, permitting realization of a versatile distributed temperature sensor with high spatial resolution.
2. Principle of Doppler OFDR
Figure 1 shows a sketch of the Doppler COFDR system, which is based on a Mach-Zehnder topology. Analogously with radio-frequency-modulated radar systems, the laser frequency was linearly swept so as to remove the distance ambiguity inherent to conventional interferometers. The laser output was split up into a reference and a measurement arm incorporating the reflecting target. First, assuming a steady target with reflectivity R, the time varying signal (after filtering out the DC and sum-frequency components) collected by a square-law detector can be written as:
Frequency analysis of Eq. (1) yields a beat frequency:
For static objects, assuming negligible dispersion over the sweep bandwidth, the distance can be readily calculated from the beat frequency:
The two-point spatial resolution of COFDR is determined by the target main lobe width of the Fourier transform [4]:
Note that the exact target position can be determined much more precisely by approximating the peak with a second-order polynomial and calculating its maximum position. Using a conventional COFDR system, we have previously demonstrated localization of optically trapped particles with a precision of ∼2 µm, corresponding to ∼1/15 of the two-point resolution [13].
In order to analyze the effects of (uniformly) moving targets, τ in Eq. (1) is replaced by:
where v0 is the target speed. Since v0 is much smaller than c in our application, any relativistic effects can be neglected and the beat signal becomesA phase-gradient autofocus algorithm, initially proposed for synthetic aperture radar images [14], was used to refocus the target peak. First, coarse peak detection and windowing of the reflection profile were applied around the peak of interest. The window width was set wide enough to cover the full peak energy and yet small enough to exclude noise. Although this algorithm can be applied iteratively, a fixed window size and a single iteration were sufficient for our application. The windowed section of the reflection profile containing the peak of interest was shifted to the origin and the phase error derivative was determined using the relation given in [14,15]. The phase errors were then determined by numerical integration and any linear slope was compensated for, since this would cause a target position error. Eventually, the blur was mitigated by multiplying the windowed peak spectrum with the conjugate phase term and shifting back the peak to its original position in the reflection profile. Naturally, this correction not only compensates the chirp term (V) but also removes the phase distortions remaining after imperfect sweep nonlinearity correction (see Appendix A.1). Figure 2 depicts the reflection profiles (up and down sweep) of a moving target (2-µm-diameter silica particle trapped and propelled inside HC-PCF with a moving speed of ∼30.4 mm/s) before and after applying the autofocus algorithm. It can be seen that peak broadening due to the phase chirp effect can be effectively suppressed by re-focusing.
Once the chirp term is compensated, two consecutive independent measurements are combined to determine the position and speed of the target. Figure 3 depicts the impact of the Doppler frequency shift and the time delay related to the target distance on the measured beat frequency.
Assuming ideal up and down sweeps with the same absolute sweep rate |k| (but opposite signs) for both measurements, the target position and speed can be calculated from the beat frequencies |fb,up| and |fb,down|:
where movement of the target between the two measurements is neglected. Accounting for different sweep rates k1, k2 and durations T1, T2 and a measurement time delay Δt between the up and down ramps (starting at f01 and f02 respectively), the position and speed can be retrieved as (see Appendix A.2 for the derivation):Note that in contrast to conventional Doppler velocimetry based on fixed-wavelength laser [16], the system can measure the instantaneous absolute position of the moving objects and is not subject to integration error over time.
3. Tracking of optically trapped and propelled particles in HC-PCF
The Doppler COFDR system was used to track optically trapped particles in HC-PCF. Flying particle sensors in HC-PCF have inherent advantages such as galvanic isolation and coverage of long distances, while additionally offering reconfigurability, improved spatial resolution and multi-parameter sensing [17]. The experimental setup is shown in Fig. 4. The 1030 nm trapping laser emitted ∼200 fs pulses at a repetition rate of 70 MHz. The average total trapping power was ∼2 W. The use of a pulsed laser is necessary to suppress inter-modal beating between fundamental and higher order core modes [18]. The measured modal walk-off distance is ∼20 cm (see Appendix A.3). The high repetition frequency of the pulse train ensures that gravity-related particle displacement between pulses is negligible. The COFDR probe beam at 1550 nm wavelength was coupled into the beam path using a dichroic mirror. The power ratio of the trapping beams was set using a motorized half-wave plate to adjust the polarization state of the trapping laser in front of a polarizing beam-splitter. We used a commercially available swept-wavelength laser source (Luna Phoenix 1400 PM) with a maximum tuning range of 50 nm around a central wavelength of 1540 nm at a maximum tuning rate of 2000 nm/s (i.e. a sweep rate of approx. 2.5×1014 Hz s−1). The inset in Fig. 4 shows a scanning electron micrograph of the single-ring HC-PCF. It has a core diameter of ∼30 µm and can guide both the trapping beam and the COFDR probe beam with low loss. The microparticles in a suspension were sprayed into a trapping chamber in front of the end face of the HC-PCF, where they were trapped by two counter-propagating laser beams and then propelled into the hollow core by adjusting the power ratio of the trapping beams. The trapping of single (as opposed to multiple) particles in the hollow core was verified by monitoring the scattered light from the side of the fiber using an infrared viewer.
We first tested the achievable precision of position measurement for silica microparticles with different sizes at the maximum sweep rate. In each configuration, a series of 100 repeat measurements was conducted for which the measured standard deviation is plotted in Fig. 5(a) as a function of sweep bandwidth. It can be seen that the measurement data fits excellently with the inverse proportionality suggested in Eq. (4). Since the sweep bandwidth is directly linked to the measurement duration, it is possible to balance position precision against measurement rate (see upper axis). Furthermore, the achievable precision is related to the signal-to-noise ratio (SNR) of the target reflection peak, which is higher for larger particles and greater sweep bandwidths. In practice, the maximum measurement rate will be reached once the SNR of the target peak is too low to be detected reliably. For highly reflective targets, it is limited by the minimum sweep duration of the tunable laser source. With the laser used, measurement rates up to approximately 250 Hz could be achieved. At the cost of reduced precision, measurement rates in the kHz range are feasible using commercially available sources.
Figure 5(b) shows the tracking of a moving silica particle with 2 µm diameter that is propelled along the fiber for around 38 s (see Table 1 for detailed COFDR parameters). We tentatively attribute the small-scale speed variations to fluctuations in trapping laser power (see Appendix A.4) that could be reduced using active power stabilization. The speed fluctuations may be also partly caused by fiber non-uniformities or static charges on the core wall, while the overall exponential speed reduction is due to fiber attenuation. The accuracy of the speed measurement (a few tens of µm/s) was assessed using a mirror on a linear stage moving at a known speed (see Appendix A.5). Note that both positive (forward) and negative (backward) particle speeds can be measured by the system. The maximum target speed that can be measured is limited by the Nyquist sampling theorem for highly reflective targets and by the SNR for weakly reflecting targets.
4. Remote temperature sensing
Next, we used flying particles in HC-PCF in combination with Doppler COFDR to build a distributed temperature sensor. In the experiment, silica particles of diameter 2 µm were once again propelled by adjusting the power ratio of the counter-propagating trapping beams. The terminal velocity of the particle is determined by the balance between the scattering force Fopt and the Stokes drag force Fdrag [17]:
where µ(T) is the viscosity of air, K is a wall correction factor accounting for the turbulence in the narrow core [19], ap the particle radius and v is the particle speed. For constant optical force Fopt the temperature-dependent particle velocity is given by: where the temperature dependence of the viscosity can be modelled using the Sutherland formula [20]:A 2-meter-long single-ring HC-PCF was used in the measurement. As shown in Fig. 4, a section of 2.5 cm was fed through an aluminum block and heated to temperatures up to 200 °C while the rest of the fiber was kept at room temperature. The temperature of the metal block was measured using an embedded temperature sensor with an accuracy of 2.5 K (IEC 584, type K, class 2). The particle was propelled through a ∼50 cm long fiber section containing the heater while its absolute position and speed were measured. The COFDR parameters are summarized in Table 1.
So as to mitigate the effects of any overall speed drift with time, all the experimental measurements of particle speed were normalized to the average speed in a reference fiber section held at room temperature (see Fig. 6(a)). The estimated spatial resolution for the COFDR configuration was 120 µm, corresponding to the ultimate theoretical spatial resolution, while the mean particle speed in the reference section was measured to be 29.1 mm/s, yielding position and speed data at intervals of 0.47 mm on average. Since this corresponds to the spatial sampling distance, it sets the upper bound for the spatial resolution of the temperature sensor. By further adjusting the particle speed it is possible to balance the spatial resolution and the time required to propel the particle along the fiber length. Note that the steep temperature gradients at the edges of the heater cause thermal creep flow of air within the hollow core (related to Knudsen pumps [22]), leading to visible undershoots and fluctuations in speed in the vicinity of these points. Based on Eq. (13) and (14), a speed reduction of ∼20% for a temperature increase of 100 K starting from room temperature is expected. This in good agreement with the experimental results. The observed decrease in speed over distance within the heated section we attribute tentatively to the thermal creep effect, pending a full investigation.
To calibrate the sensor, the average particle speed in the heater section was first measured at seven different metal block temperatures (Fig. 6). The error bars in Fig. 6(b) are given by the standard deviation of the speed fluctuations over distance within the heater section (see Fig. 6(a)). Combining Eq. (13) and (14), the temperature dependence of particle speed can be written as:
5. Conclusion
These proof-of-principle experiments demonstrate that a Doppler OFDR system can track the instantaneous absolute position and speed of moving objects with high accuracy. Such a system can be used to probe the motion dynamics of macroscopic objects with micrometer spatial resolution. It can also be used to probe the viscoelasticity of the gases or liquids in the hollow-core of a HC-PCF. The integration of Doppler OFDR with particle guidance in HC-PCF opens up novel possibilities for flying particle sensors capable of measuring multiple physical quantities with high-spatial resolution. Since the COFDR system can also track the positions of multiple particles [13], it may be exploited to investigate the dynamics of levitated particles or bound particle arrays in HC-PCF [18].
Appendix
A.1. Sweep linearization of COFDR
Even for highly optimized sweeping lasers, there always remains a residual deviation from a perfect frequency ramp, resulting from environmental influences or the internal sweep feedback mechanisms. It is intuitively clear that any deviation from an ideal sweep will lead to a time-varying beat signal and therefore distort the reflection profile. As a result, the localization precision will be reduced or, in the case of oscillating deviations, side-lobes will occur and corrupt the reflection profile.
To analyze this effect, the ideal laser sweep signal is extended with a phase error function ϕe(t):
where ϕe(t) consists of a deterministic (but unknown) error term ϕnl(t) caused by the sweep nonlinearity and a stochastic process ϕpn(t) used to model the laser phase noise:This yields the beat signal in complex notation after down-mixing, low-pass filtering and scaling (assuming an interferometer delay τ and neglecting any losses):
To correct the effect of a deterministic phase error term ϕnl(t), an auxiliary interferometer was used to interpolate the actual laser frequency for each sample. An absolute wavelength reference (an absorption cell) is available from trigger signals provided by the laser source at predefined wavelengths. The measurement interferometer signal was then resampled at equidistant source frequencies.
Figure 8(a) plots a section of the auxiliary interferometer signal (after subtracting a slowly varying DC component). First, the exact zero crossings (red markers) were determined by linear interpolation between data-points on either side of the sign changes. In a second step, the mean distance between zero crossings with rising and falling slopes was calculated (blue markers) and used for interpolation. Figure 8(b) shows auxiliary signal period length (resembling equidistant frequency intervals) determined from the markers for a laser sweep from 1535 nm to 1545 nm in ∼5 ms sampled at 20.8 MS/s. The reference delay τref was chosen small enough to allow direct monitoring of the laser frequency. Even though a longer delay leads to a smaller free spectral range in the auxiliary interferometer, and thus a finer “sampling” of the laser frequency, the limit for τref is the maximum sampling frequency since oversampling is required to precisely determine the positions of the zero crossings. Table 2 lists the parameters of the auxiliary interferometer in the COFDR system. The delay fiber in the auxiliary interferometer was placed in a styrodur box so as to shield it from environmental influences. Furthermore, polarization maintaining fibers were used in the auxiliary interferometer to avoid polarization-related drift.
The correction term extracted from the auxiliary interferometer with delay τref is given by:
Obviously, a perfect correction is only achieved when τref = τmeas. In most cases, however, the error can still be significantly reduced for differing delays Δτ = τmeas – τref. Assuming a harmonic phase error function ϕnl(t) = m sin(2πfnlt) with frequency fnl and modulation index m yields a remaining phase error of the form:
A.2. Derivation of Eq. (10) and (11)
Starting from the approximation in Eq. (7), the beat frequencies resulting from an up and down frequency ramp are:
Equation (24) and (25) can be combined to determine the position and speed of the target yielding Eq. (10) and (11).
A.3. Effect of intermodal beating on particle speed
Figure 10 plots the measured particle speed close to the input face of the HC-PCF (at 3.568 m). The periodic fluctuations in speed due to intermodal beating are clearly visible. The period of ∼2.3 mm matches the beat period between the LP01- and LP02-like core modes. The beat pattern decays after ∼20 cm propagation due to walk-off between the different modal pulses.
A.4. Measured particle speed and trapping power fluctuation
Figure 11 depicts the normalized particle speed deviation and the trapping power fluctuation for a ∼10 s measurement interval. The trapping power propagating towards the PCF was probed with a photodiode placed after a beam splitter while the particle was propelled in the backward direction with an average speed of 32.7 mm/s. It can be seen that the speed variation mostly follows the trapping power fluctuation.
A.5. Characterization of the speed measurement accuracy
The accuracy of the position and speed measurements was estimated using a gold mirror mounted on a linear stage (Zaber X-LDQ0600C). The stage was set to a speed of +/–10 mm/s. The linear stage outputs a trigger signal every 500 µm as it travels. This was used as a reference.
Aligning the COFDR speed measurements (sweep rate: 2000 nm/s, sweep range: 1518 nm to 1562 nm) with the speed calculated from the trigger signal (see Fig. 12(a)) allowed the deviations to be calculated by linearly interpolating the reference speed at the times when the COFDR measurements were made (see Fig. 12(b)). The standard deviation (excluding outliers during acceleration and deceleration) was estimated as 49.5 µm/s. Note that due to limited internal sampling frequency (10 kHz) of the stage, the trigger signal uncertainty is 0.1 ms, causing a speed uncertainty of ∼20 µm/s in the reference trace.
Funding
Deutsche Forschungsgemeinschaft (418737652); Max-Planck-Gesellschaft; Erlangen Graduate School of Advanced Optical Technologies.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented here may be obtained from the authors upon reasonable request.
References
1. D. Uttam and B. Culshaw, “Precision time domain reflectometry in optical fiber systems using a frequency modulated continuous wave ranging technique,” J. Lightwave Technol. 3(5), 971–977 (1985). [CrossRef]
2. W. Eickhoff and R. Ulrich, “Optical frequency domain reflectometry in single-mode fiber,” Appl. Phys. Lett. 39(9), 693–695 (1981). [CrossRef]
3. R. Passy, N. Gisin, J. P. von der Weid, and H. H. Gilgen, “Experimental and theoretical investigations of coherent OFDR with semiconductor laser sources,” J. Lightwave Technol. 12(9), 1622–1630 (1994). [CrossRef]
4. U. Glombitza and E. Brinkmeyer, “Coherent frequency-domain reflectometry for characterization of single-mode integrated-optical waveguides,” J. Lightwave Technol. 11(8), 1377–1384 (1993). [CrossRef]
5. S. A. Kingsley and D. Davies, “OFDR diagnostics for fibre and integrated-optic systems,” Electron. Lett. 21(10), 434–435 (1985). [CrossRef]
6. B. J. Soller, D. K. Gifford, M. S. Wolfe, and M. E. Froggatt, “High resolution optical frequency domain reflectometry for characterization of components and assemblies,” Opt. Express 13(2), 666–674 (2005). [CrossRef]
7. J. P. von der Weid, R. Passy, G. Mussi, and N. Gisin, “On the characterization of optical fiber network components with optical frequency domain reflectometry,” J. Lightwave Technol. 15(7), 1131–1141 (1997). [CrossRef]
8. 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]
9. S. T. Kreger, D. K. Gifford, M. E. Froggatt, B. J. Soller, and M. S. Wolfe, “High Resolution Distributed Strain or Temperature Measurements in Single- and Multi-Mode Fiber Using Swept-Wavelength Interferometry,” in Optical Fiber Sensors. OSA Technical Digest (CD) (OSA, 2006), paper ThE42.
10. D. K. Gifford, “Distributed fiber-optic temperature sensing using Rayleigh backscatter,” in 31st European Conference on Optical Communication (ECOC 2005) (IEE, 2005), 511–512 vol. 3.
11. D. Arbel and A. Eyal, “Dynamic optical frequency domain reflectometry,” Opt. Express 22(8), 8823–8830 (2014). [CrossRef]
12. L. Shiloh and A. Eyal, “Sinusoidal frequency scan OFDR with fast processing algorithm for distributed acoustic sensing,” Opt. Express 25(16), 19205–19215 (2017). [CrossRef]
13. J. Podschus, M. Koeppel, B. Schmauss, A. Sharma, S. Sundaramahalingam, S. Xie, and P. Russell, “Position Measurement of Multiple Microparticles in Hollow-Core Photonic Crystal Fiber by Coherent Optical Frequency Domain Reflectometry,” 27th International Conference on Optical Fibre Sensors (OFS) Alexandria 2021 (2021); accepted for presentation.
14. P. H. Eichel, D. C. Ghiglia, and C. V. Jakowatz, “Speckle processing method for synthetic-aperture-radar phase correction,” Opt. Lett. 14(1), 1–3 (1989). [CrossRef]
15. D. E. Wahl, P. H. Eichel, D. C. Ghiglia, and C. V. Jakowatz, “Phase gradient autofocus-a robust tool for high resolution SAR phase correction,” IEEE Trans. Aerosp. Electron. Syst. 30(3), 827–835 (1994). [CrossRef]
16. M. K. Garbos, T. G. Euser, O. A. Schmidt, S. Unterkofler, and P. St. J. Russell, “Doppler velocimetry on microparticles trapped and propelled by laser light in liquid-filled photonic crystal fiber,” Opt. Lett. 36(11), 2020–2022 (2011). [CrossRef]
17. D. S. Bykov, O. A. Schmidt, T. G. Euser, and P. St. J. Russell, “Flying particle sensors in hollow-core photonic crystal fibre,” Nat. Photonics 9(7), 461–465 (2015). [CrossRef]
18. D. S. Bykov, S. Xie, R. Zeltner, A. Machnev, G. K. L. Wong, T. G. Euser, and P. St. J. Russell, “Long-range optical trapping and binding of microparticles in hollow-core photonic crystal fibre,” Light: Sci. Appl. 7(1), 22 (2018). [CrossRef]
19. N. Al Quddus, W. A. Moussa, and S. Bhattacharjee, “Motion of a spherical particle in a cylindrical channel using arbitrary Lagrangian-Eulerian method,” J. Colloid Interface Sci. 317(2), 620–630 (2008). [CrossRef]
20. W. Sutherland, “LII. The viscosity of gases and molecular force,” The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science 36(223), 507–531 (1893). [CrossRef]
21. F. M. White, Viscous fluid flow, 2. ed. (McGraw-Hill, 1991).
22. O. A. Schmidt, M. K. Garbos, T. G. Euser, and P. St. J. Russell, “Reconfigurable optothermal microparticle trap in air-filled hollow-core photonic crystal fiber,” Phys. Rev. Lett. 109(2), 024502 (2012). [CrossRef]