Open Access
Add to BibSonomyAbstract
Stimulated Brillouin scattering in optical fibers can be used to measure strain or temperature in a distributed manner. Brillouin optical time domain analysis (BOTDA) is the most common sensor system based on the Brillouin scattering. This paper presents the experimental analysis of the characteristics of Brillouin gain spectrum (BGS) influenced by the width of launched pulse. Brillouin strain coefficient is also examined for the different pulse widths, which is important to apply a Brillouin scattering-based sensor to a structural health monitoring. Experimental results showed that not only the Brillouin linewidth and gain but also the Brillouin frequency were dependent on the pulse widths.
©2005 Optical Society of America
1. Introduction
Stimulated Brillouin scattering (SBS) has been extensively studied in recent years in optical fiber communications and in distributed fiber optic sensors [1–3]. Brillouin scattering refers to the scattering of a lightwave by an acoustic phonon [4,5]. When this process occurs in an optical fiber, the backscattered light suffers a frequency shift known as a Brillouin frequency shift, which is dependent on the temperature and strain environment of the fiber. This process has been proven useful as a sensing mechanism for distributed optical fiber sensors [3]. Brillouin optical time domain analysis (BOTDA) was first proposed as a nondestructive attenuation measurement technique for optical fibers [6]. BOTDA uses Brillouin gain spectroscopy, in which a pulsed light and a counter-propagating continuous-wave (CW) light are launched into an optical fiber. Brillouin scattering-based distributed optical fiber sensing allows for determination of either strain or temperature changes through measurement of the Brillouin spectrum along the fiber. The position at which the measured strain and temperature are located along the fiber is determined by the time of flight for the pulse to propagate down and back through the fiber. In a Brillouin gain spectrum (BGS), the Brillouin gain power and the linewidth are dependent on the launched pulse width; the power increases and the linewidth become narrow as the pulse width increases [7,8].
In this paper, the Brillouin frequency dependence on the pulse width was examined experimentally. Experimental results showed that not only the Brillouin linewidth but also the Brillouin frequency was dependent on the pulse width for shorter pulses than about 50 ns. The Brillouin strain and temperature coefficients were measured with varying the pulse widths to see whether the coefficients were also dependent on the pulse width or not. Additional experiments were carried out to examine the dependence of BGS characteristics on the excited time of a fiber by a pulse light.
2. Theory
2.1. Stimulated Brillouin scattering
When electromagnetic radiation of optical frequencies travels through matter various spontaneous scattering processes can occur such as Rayleigh scattering, Brillouin scattering and Raman scattering. When light is scattered by acoustic phonons we speak of Brillouin scattering. Acoustic phonons (thermally excited lattice vibration with acoustic mode) produce a periodic modulation of the refractive index. Brillouin scattering occurs when light is diffracted backward on this moving grating, giving rise to frequency shifted Stokes and anti-Stokes waves. The Brillouin frequency (frequency difference between pump light and scattered light) νB depends on the acoustic wave velocity and is given by
where Va is the acoustic wave velocity within the fiber, n is the refractive index and λp is the wavelength of the incident pump lightwave. The spectral width ΔνB is very small and is related to the damping time of acoustic waves or the phonon lifetime TB. In fact, if the acoustic waves are assumed to decay as exp(-t/TB), the Brillouin gain has a Lorentzian spectral profile given by
where νB is the full width at half maximum (FWHM). The Brillouin gain spectrum (BGS) peaks at the Brillouin frequency νB , and the peak value is given by the Brillouin gain coefficient g 0
where p 12 is the longitudinal elasto-optic coefficient, ρ 0 is the material density, c is the vacuum velocity of light and λp is the pump light wavelength [5,9].
2.2. Brillouin strain and temperature coefficients
The Brillouin frequency νB (Δε, ΔT), under the change of strain and temperature, is given by
Where, ν B0 represents the Brillouin frequency of the unperturbed fiber, that is, for a fixed strain and temperature; Δε and ΔT are the changes in strain and temperature, Cε and CT are the Brillouin strain and temperature coefficients. Typical values of the coefficients, at a wavelength λp =1550 nm, are Cε ≈ 500 MHz/% and CT ≈ 1 MHz/°C [10]. However, the values of the coefficients must be obtained experimentally for reliable measurements because the values change depending on the wavelength and fiber properties. In Eq. (4), actually, the Brillouin temperature coefficient has two terms of temperature dependence: the first is the temperature dependence of the Brillouin frequency shift for a bare fiber and the second is the Brillouin frequency shift caused by the thermal strain in the fiber [11]. In this study, the temperature term was not considered because experiments were carried out at fixed temperature allowing no Brillouin frequency shift due to temperature change.
Pulse width dependence of the Brillouin gain and linewidth has been reported in several literatures [7], but few literatures have been reported for the pulse width dependence of the Brillouin frequency. In this study, we observed that the Brillouin frequency changed with the pulse width launched into a fiber. If the Brillouin strain and temperature coefficients also change depending on the pulse width, every strain and temperature coefficient corresponding to the launched pulse width should be obtained for each strain and temperature measurement, which is very troublesome but necessary.
Hence, in this paper, the relationship between the coefficients and the pulse width was examined. In addition, the BGS dependence on the fiber excitation time, associated with acoustic phonons causing the Brillouin scattering, was examined, as will be shown in Section 4.
3. Experimental setup
The BOTDA system shown in Fig. 1 was used to measure the Brillouin gain spectrum. The output of DFB LD at 1553 nm is first split by a coupler. One output of the coupler is amplitude-modulated by a 2.5 Gb/s electro-optic modulator to generate pulses. This pump light is amplified by Erbium-doped fiber amplifier (EDFA) and is launched into a test fiber. The other output of the coupler is frequency-modulated by a 10 Gb/s electro-optic modulator and is launched into opposite end of the test fiber. This configuration offers determining advantages such as no dependence on the laser frequency drift and no need of a tunable laser source [12].
The BGS was measured by sweeping the modulation frequency fm in the vicinity of the Brillouin frequency νB and by detecting the total intensity of the probe light using photo receiver. The probe light signal detected by a photo detector is classified into two power signals such as the power detected before the launch of the pulse and the power detected after the launch of the pulse. Hence, the net Brillouin gain amplified by a contribution of the pulse was obtained through subtracting the power detected before the launch of the pulse from the power detected after the launch of the pulse. Because the power of the frequency-modulated probe light showed a little variation for the modulation frequency fm , the net Brillouin gain obtained through the interaction with pulse was normalized through dividing by the probe light power at the given modulation frequency.
Fig. 1. Experimental setup for Brillouin gain spectrum measurements. (ISO = isolator, PC = polarization controller, PS = polarization scrambler, DET = detector, ATT = attenuator, CIR = circulator, FUT = fiber under test.)
4. Experimental results
Brillouin gain spectra were measured with varying widths of launched pulses for different power values of the pulses such as 280 mW, 550 mW and 1160 mW. The dispersion shifted single mode fiber (DSF) whose length was about 300-m-long was used in the measurement; the DSF was set free to have no strain changes and room temperature was fixed at 20 ☐. The sensing location was 100 m from the pulse injection end. BGS’s of all pulses, except for 200 ns pulse, were averaged to provide information on the same section of the fiber. The BGS of 200 ns pulse, which was the longest pulse among the launched pulses, provided information on the 20 m section of the fiber before the sensing location. So, for example, in case of a 50 ns pulse, the BGS’s were averaged from 85 m to 100 m to identify the fiber section under comparison. Thus every BGS obtained with different pulse widths provided the Brillouin frequency information on the same fiber section ranging from 80 to 100 m. Figure 2 shows the typical characteristics of the pulses injected for this study. The rise and fall times were set as 2 ns by the pulse generator for all the pulses injected for this research, and the rise and fall times of the output pulses were highly consistent with the set value as shown in Fig. 2.
Figure 3 and Fig. 4 show the BGS characteristics observed in the measurement, and, as already known, the Brillouin gain decrease and the linewidth broadening can be easily seen in Fig. 3 and 4 as the pulse width is reduced. What is noteworthy is that the Brillouin frequency is dependent on the widths of pulses especially for the short pulses less than about 50 ns.
From the viewpoint of the variation of the Brillouin frequencies with increase of the power and the width of pulses, the Brillouin frequency became higher as the power and the width increased, and it eventually converged on a specific value (~10.5075 GHz) for longer pulses than about 60 ns regardless of the pulse powers. For each pulse power, it required longer pulses than about 50 ns to have a smaller deviation than 1 MHz from the convergent Brillouin frequency. Even with different pulse powers, the Brillouin frequency also converged on the specific value for the pulses longer than about 50 ns with showing a deviation less than 1 MHz. Here, the 1 MHz error in frequency indicates the error values of about 20 με and 1 ☐ in strain and temperature respectively.
The major cause of the dependence of the Brillouin frequency on the pulse widths is thought to be the excitation time, which is varied with pulse widths, of the fiber by the pulse lights, as will be shown later in this paper.
Fig. 4. Brillouin linewidth (left) and Brillouin frequency (right) observed with varying pulse widths for different pulse powers.
In view of the pulse width dependence of the Brillouin frequency, it should be carefully examined whether the Brillouin strain coefficient also depends on the pulse width or not. If the Brillouin strain coefficient depends on the pulse width, the relationship between the coefficient and pulse widths must be considered in each strain measurement, as previously mentioned, and thus each value of the coefficient corresponding to the pulse width should be measured for reliable measurements.
To observe the relationship between the Brillouin coefficients and pulse widths, experiments were carried out as follows. The 12-m-long section of the DSF was elongated with varying the strain of the section; Brillouin frequencies were measured at each strain state for pulses of 30, 50 and 100 ns respectively; the launched pulses were 550 mW in their power. As similar to the above BGS measurement, the Brillouin gain signals of the 30 and 50 ns pulses were averaged, for the 10-m-section which was correspondent to the location of the Brillouin gain signal of the 100 ns pulse, to identity the fiber section to be considered.
The measured Brillouin strain coefficients are shown in Fig. 5. Fortunately, as shown in Fig. 5, we can say that the Brillouin strain coefficient is nearly not dependent on the width of launched pulses with consideration of the whole results presented in this paper.
For the DSF used in this measurements, the Brillouin strain coefficient was 467 MHz/% (= 46.7 kHz/με) without regard to the pulse width. The reason of showing an error of about 0.2 kHz/με for the 30 ns pulse, when compared with other pulses, is thought because the signal to noise ratio (SNR) of the BGS obtained with the 30 pulse was relatively low compared with those obtained with other pulses. However, the error value is small enough to be neglected.
Generally, the Brillouin scattering is known as a scattered light caused by acoustic phonons and the acoustic phonons occur due to the excitation of a fiber by the light propagating through the fiber [4]. Hence, in an optical fiber, the Brillouin scattering can be said to be originated from the excitation of the fiber by the pulse launched into the fiber. Then, the dominant parameter of the pulse to the Brillouin scattering is the fiber excitation time determined by the width of the pulse.
In a BOTDA system, pulse width determines the Brillouin interaction length, which is the spatial resolution, between the pulse and probe lights as well as the fiber excitation time. The spatial resolution (S) equals the half of the pulse length and is given by
Here, vg is the group velocity of light within the fiber and W is the pulse width. For the pulse whose width is W , when the pulse leading edge falls in the position of L in an optical fiber, the corresponding Brillouin gain signal, detected at the detector located at the pulse injection end, has the relationship between the Brillouin interaction part and its excited time as shown in Fig. 6. As shown in Fig. 6, for a longer pulse, not only the Brillouin interaction length become longer but also the part excited for more longer time become larger, when compared with a short pulse.
To observe the effect of the fiber excitation time on the BGS, experiments were carried out as follows; a 1-m-long DSF (νB ≈ 10.508 GHz) was inserted into a 400-m-long standard single mode fiber (SMF, νB ≈ 10.873 GHz) using a fusion splicer at the location ranging from 308 m to 309 m. Actually, the location of the inserted DSF was not consistent with what we wanted by ~ 10 cm. Figure 7 shows the configuration of the fiber under test.
If the probe light power is observed by sweeping the modulation frequency near the Brillouin frequency of the inserted DSF, only the Brillouin gain amplified within the inserted DSF could be detected; this gives the fixed Brillouin interaction length of 1 meter, regardless of the pulse widths for the pulses longer than 10 ns. A 200 ns pulse was injected into the fiber under test, and BGS’s were measured with sweeping the modulation frequency near the Brillouin frequency of the inserted DSF with a 200 MHz sampling rate. Figure 8 shows the obtained BGS with respect to the pulse leading edge locations and the excited time of the inserted DSF with respect to those. Figure 9 shows the measured Brillouin linewidths and frequencies with respect to the fiber excited time.
Fig. 8. BGS’s (left) and corresponding excited time (right) with respect to the pulse leading edge locations.
In Fig. 8 and 9, the values of excited time correspond to the mid-points, at which each value corresponds to the mean excitation time within the Brillouin interaction section. The changes of linewidth and frequency in BGS characteristics are shown well in Fig. 8; The BGS rose toward right upside and eventually converged to a specific configuration for the excited time longer than about 35 ns. The Brillouin linewidths converged more rapidly to a specific value when compared with those obtained with varying the pulse width as shown in Fig. 4; this is because the BGS’s obtained with pulses are under the effect of an overlapping when the Brillouin interaction lengths are equal to the corresponding pulse widths. For example, the BGS of 40 ns pulse in Fig. 3 is similar to the result obtained through the overlapping the BGS’s of 309 m ~ 312 m in Fig 8. The same explanation is valid for the Brillouin frequency result.
More experiments with shorter insertion length than 1 meter are expected to clarify the BGS characteristics in more detail. Nevertheless above experimental results provide important information on the researches of Brillouin scattering-based optical fiber sensors and characteristics associated with the sensors.
5. Conclusions
In this paper, BGS characteristics were examined experimentally using a BOTDA sensor system. We observed that the Brillouin frequency changed depending on the width of the pulses launched into an optical fiber. To see the variation tendency of the Brillouin frequency, pulses were launched into an optical fiber with varying the width and power of the pulses, and each corresponding BGS was measured and examined. The results showed that the Brillouin frequency became higher with increase of the width and power and converged on a specific value for longer pulses than about 50 ns.
Brillouin strain coefficients were measured with different pulse widths to see whether the coefficient was dependent on the pulse width or not. From the result of the measurements, it was concluded that the coefficient was fortunately indifferent to the pulse width. Additional experiments were carried out for the fixed Brillouin interaction length of 1 meter to see the effect of the excitation time on the BGS characteristics.
The experimental results presented in this paper are expected to contribute largely to a Brillouin scattering-based sensor technology and its application to a structural health monitoring technique.
Acknowledgments
This study was supported by the Intelligent Robotics Development Program, one of the 21st Century Frontier R&D Programs funded by the Ministry of Commerce, Industry and Energy of Korea.
References and links
1 . B. Culshaw and J. P. Dakin , Eds., Optical Fiber Sensors ( Artech House, Boston, London , 1989 ), I and II.
2 . T. L. Tang , “ Saturation and spectral characteristics of the Stokes emission in the stimulated Brillouin process ,” J. Appl. Phys. 37 , 2945 – 2955 ( 1966 ). [CrossRef]
3 . T. Horiguchi and M. Tateda , “ Tensile strain dependence of Brillouin frequency shift in silica optical fibers ,” IEEE Photonics Technol. Lett. 1 , 107 – 108 ( 1989 ). [CrossRef]
4 . Y. R. Shen , The Principles of Nonlinear Optics , ( John Wiley & Sons, New York , 1984 )
5 . W. Kaiser and M. Maier , “ Stimulated Rayleigh, Brillouin and Raman Spectroscopy ,” in Laser Handbook , F. T. Arecchi and E. O. Schulz-Dubois , eds. ( North-Holland , 1972 ) 2 , 1077 – 1150 .
6 . T. Horiguchi and M. Tateda , “ BOTDA-Nondestructive measurement of single-mode optical fiber attenuation characteristics using Brillouin interaction: Theory ,” J. Lightwave Technol. 7 , 1170 – 1176 ( 1989 ). [CrossRef]
7 . J. Smith , A. Brown , M. DeMerchant , and X. Bao , “ Pulse width dependence of the Brillouin loss spectrum ,” Opt. Comm. 168 , 393 – 398 ( 1999 ). [CrossRef]
8 . A. Fellay , L. Thevenaz , M. Facchini , M. Nikles , and P. Robot , “ Distributed sensing using stimulated Brillouin scattering: owards ultimate resolution ,” Proceedings of OFS’97 16 , 324 – 327 ( 1997 ).
9 . G. P. Agrawal , Nonlinear Fiber Optics , ( Academic Press, Boston , 1989 ), chap. 9.
10 . S. B. Cho , J. J. Lee , and I. B. Kwon , “ Temperature compensation of a fiber optic strain sensor based on Brillouin scattering ,” J. Opt. Soc. Kor. 8 , 168 – 173 ( 2004 ). [CrossRef]
11 . T. Kurashima , T. Horiguchi , and M. Tateda , “ Thermal effects on the Brillouin frequency shift in jacketed optical silica fibers ,” Appl. Opt. 29 , 2219 – 2222 ( 1990 ). [CrossRef] [PubMed]
12 . M. Niklés , L. Thévenaz , and P. A. Robert , “ Brillouin gain spectrum characterization in single-mode optical fibers ,” J. Lightwave Technol. 15 , 1842 – 1851 ( 1997 ). [CrossRef]
