Open Access
Add to BibSonomy
Abstract
Silicon core fibers are a promising candidate for optoelectronic and photonic applications. Fabrication and post-processing of those fibers is thus far done without precise knowledge of the processing temperatures. Here, a simple technique is presented that allows for in-situ temperature monitoring during thermal processing of silicon core fibers. The temperature was probed across the silicon melting point and cooling rates above 3500 °C s−1 were measured. The silicon core was found to be molten at a temperature of 1281 °C, more than 100 °C below the bulk silicon melting point. This is attributed to stress inbuilt to silicon core fibers during the fabrication process.
© 2021 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Silicon core fibers offer unique possibilities for THz-transmission, mid-IR sensing and nonlinear optical devices [1]. Optical transmission losses of as-drawn silicon core fibers can be below 0.2 dB cm−1 [2], yet are typically exceeding 1 dB cm−1 making suitable post-processing a necessary fabrication step [3,4]. One method is to anneal the fiber above the melting point of the silicon core, with the silica cladding acting as a crucible. Under suitable cooling conditions, the core recrystallizes into a single crystal with a subsequent decrease in optical transmission losses [5]. For this process the temperature gradient across the solidification interface is critical to the growth rate and final quality of the crystal [6]. Thermal dynamics during laser scanning has previously been estimated by monitoring the black-body radiation from the silicon core or using finite element modeling [5,7]. Here, we present a simple in-situ technique to measure the temperature dynamics of the fiber during laser scanning and the recrystallization process.
2. Experimental details
Thermal dynamics of a 127.5 µm diameter silica-clad silicon core fiber (SCF) were investigated during CO2 laser processing. The examined SCF was suspended in air between two clamps and had a core diameter of 20 µm. It was fabricated inhouse using the molten core technique [8] in a novel carbon monoxide laser based drawing tower described in detail in Ref. [9]. The experimental setup shown in Fig. 1 consisted of two parts: a fiber heating system and a temperature measurement system;
Fig. 1. (a) Top view and (b) side view of the measurement setup. The SCF was heated by a CO2 laser beam that was scanned along the fiber by moving the translation stage. The HeNe laser beam probed the SCF at a central position between two clamps at an incident angle of 46.8°. The reflected beam from the fiber surface as well as the internally refracted and reflected beam overlap and are incident on the sensors S1 and S2.
The fiber heating system was comprised of a 100 W CO2 laser (Synrad Firestar ti100HS) operating at a wavelength of $\lambda = {10.6}\,\mathrm{\mu}\textrm{m}$, a translation stage (Aerotech ALS130H-50) and a cylindrical lens ($f= {101.6}\,\textrm{mm}$) that focused the laser beam horizontally ensuring uniform heating across the entire fiber diameter. The chosen CO2 laser provided ample power to efficiently heat the SCF and melt the silicon core. The CO2 laser beam was directed onto the fiber by a golden mirror (M1) that was mounted on top of the translation stage. By moving the translation stage parallel to the fiber, different sections of the SCF were exposed to the CO2 laser beam. The desired annealing speed was set by the velocity of the translation stage.
The temperature measurement system consisted of a 5 W HeNe laser (Thorlabs HNL050LB) operating at a wavelength of $\lambda = {632.8}\,\textrm{nm}$, a spherical lens ($f= {50}\,\textrm{mm}$) and two sensors S1 and S2. The SCF was probed with the HeNe laser targeting an incident angle of 46.8°. In this case, given the refractive index of the silica cladding ($n={1.457}$), the first reflection from the fiber surface and the internally refracted and reflected beams overlapped and formed in the far field the interference pattern shown in Fig. 2. The temperature of the fiber could then be determined by monitoring the interference pattern, since the exact intensity profile of the interference pattern depends on the phase shift between the probe beam and the reference beam. The phase shift itself depends on the optical path length of the probe beam and thus the temperature, coupled through the thermal-expansion and the thermo-optic coefficient. The interference pattern can be described by the function
where $A=0.048I_0$, $B=-0.019I_0$, and $I_0$ is the initial laser beam intensity as has been shown in Ref. [10]. The phase shift, and thereby the temperature, was determined by quadrature phase shift detection as explained in Ref. [11]. This temperature measurement technique and the necessary data processing was conducted as detailed in Ref. [10] and Ref. [12], respectively.Additionally, a CMOS camera (Thorlabs DCC3240C) was placed perpendicular to the laser beam above the SCF (not shown in Fig. 1) to investigate the exposed fiber section visually (see Fig. 3(a) for a micrograph of a heated SCF). Real-time monitoring of the core of the exposed fiber section is possible since the transparent silica cladding is far less emissive than the silicon core. Further, since the emissivity of solid and liquid silicon around the melting point differ by approximately a factor two [13], the two phases can be easily distinguished as shown in Fig. 3(b).
Fig. 3. (a) Micrograph of a SCF being heated by a CO2 laser beam. (b) Micrograph of the same SCF heated at a higher laser power. The barely emissive silica cladding is not visible any more. The silicon core can clearly be seen, and molten silicon (in the center) can be distinguished from solid silicon (around the center), due to their different emissivities. (c) Intensity line scan along the fiber. Molten silicon has in contrast to solid silicon a clear dip in emissivity at the phase transition.
3. Results
3.1 Temperature measurements
The SCF was annealed by the CO2 laser beam, both stationary and with annealing speeds ranging from 0.1 mm s−1 to 10 mm s−1. Four different laser powers ranging between $P= {5.8}\,\textrm{W}$ and $P= {35.9}\,\textrm{W}$ were used. For each measurement, the temperature was monitored at the same fixed temperature measurement point on the fiber. Two types of measurements have been performed and are illustrated in Fig. 4.
Fig. 4. Illustration of the fixed measurement point (dotted line) for (a) the stationary case with a fixed CO2 laser exposure point and instantaneous power change (laser on/off) and (b) the moving case with a scanning CO2 laser.
The first measurement (see Fig. 4(a)) was stationary and the fiber was exposed at the temperature measurement point for a duration of 4 s without moving the translation stage. The CO2 laser was carefully aligned to heat the fiber exactly at the temperature measurement point of the HeNe laser beam. The result for a laser power of $P={23.0}\,\textrm{W}$ is shown in Fig. 5(a). The measured temperature rose from room temperature up to 1173 °C after the shutter was opened at $t={1}\,\textrm{s}$. With continuous exposure the temperature was stable at the maximum temperature. When the shutter was closed at $t= {5}\,\textrm{s}$, the measured temperature decreased rapidly back to room temperature.
For the second measurement, the SCF was annealed along 2 cm across the temperature measurement point as illustrated in Fig. 4(b). The measured temperature behavior is shown in Figs. 5(b)-(d) for annealing speeds of 1.0 mm s−1, 3.0 mm s−1 and 10.0 mm s−1 at a fixed laser power of $P={23.0}\,\textrm{W}$. For the lowest annealing speed $v= {1.0}\,\textrm{mm s}^{-1}$, the measured temperature profile was approximately symmetrical on the heating and the cooling side. The maximum temperature of 1153 °C was comparable to the stationary case (1173 °C). For higher annealing speeds, the maximum measured temperatures decreased to 758.0 °C for $v={10}\,\textrm{mm s}^{-1}$. In addition to the lower temperatures, the shape of the temperature profiles was the less symmetrical the faster the annealing speed was.
Fig. 5. Comparison of the measured temperature profiles for a SCF when annealed with different speeds by a CO2 laser at a power of $P= {23.0}\,\textrm{W}$.
The effect of different laser powers between $P={5.8}\,\textrm{W}$ and $P={35.9}\,\textrm{W}$ on the temperature profile when annealing with a fixed speed of $v={3}\,\textrm{mm}$ is shown in Fig. 6. The measured temperature rose rapidly as the laser beam approached the measuring point. Cooling of the SCF is slower than heating. The main difference between the measured temperature profiles is the different maximum temperatures caused by different laser powers. The shape of the temperature profiles was similar for all power settings.
Fig. 6. Temperature profile for a SCF showing the effect of different laser powers on temperature dynamics and peak temperature at an annealing speed of $v= {3}\,\textrm{mm s}^{-1}$.
An overview of all measurements and the obtained maximum temperatures is given in Fig. 7. It can be summarized that slower annealing speeds and higher powers lead to higher peak temperatures while faster annealing speeds and lower powers lead to lower peak temperatures. The missing data points for the lowest speeds at the highest powers were caused by deformation of the fiber at these high temperatures. As the glass is softened, the fiber deformed and Eq. (1) is no longer valid. Additionally, convection made the fiber vibrate and consequently altered the interference pattern. However, a temperature of 1367 °C could still be measured for a laser power of 35.9 W and an annealing speed of 3 mm s−1. Note that at the highest power setting the images of the CMOS camera clearly showed a molten silicon core of the fiber for two annealing speeds: 3 mm s−1 and 5 mm s−1; The latter corresponded to a measured maximum temperature of 1281 °C, well below the ambient pressure melting point of crystalline silicon at 1414 °C. For the second highest power setting $P= {23.0}\,\textrm{W}$ the maximum temperature of 1173 °C was insufficient to melt the silicon core at any speed.
Fig. 7. Maximum measured temperatures for a SCF depending on annealing speed and laser power. The measurement points highlighted by dashed circles correspond to those with a molten silicon core. The grey dotted line indicates the ambient melting point of silicon at $T={1414}\,^{\circ}\textrm{C}$.
3.2 Cooling rate
The cooling rate of the SCF was calculated from the measured temperature profiles. An exponential was fitted to the cooling side of the temperature profile. Here, the cooling rate $dT/dt$ was defined as
where $T_\textrm {MAX}$ is the maximum temperature, $T_{1/e} = T_\textrm {MAX}/e$ and $\tau$ is the time constant obtained from the exponential fit. The obtained cooling rates for the SCF are shown in Fig. 8. For the stationary case, the highest measured cooling rate was 3504 °C s−1. The plot shows how the cooling rate increased proportionally with peak temperature, $T_\textrm {MAX}$, and with faster annealing speeds. The highest cooling rate measured during annealing was 2003 °C s−1 in the case of 5 mm s−1 annealing. It can be expected that for annealing at 10 mm s−1 the cooling rate would be even higher, if the same maximum temperatures were to be reached without fiber deformation.Fig. 8. Cooling rates of the SCF for the stationary case (dashed line), with an instantaneous power change (laser on/off) and the moving case (solid lines).
3.3 Raman spectroscopy
The investigated fiber was further characterized by Raman spectroscopy at room temperature. The spectrum from the drawn fiber was compared to the spectrum of a crystalline silicon wafer. The samples were exposed to a HeNe laser beam and the spectra shown in Fig. 9 were taken by a Horiba iHR 550 spectrometer. The spectra were fitted by a Voigt function and the obtained Lorentzian linewidth was 2.6 cm−1 for the silicon wafer and the SCF. This indicates a high degree of crystallinity of the silicon core of the SCF. The downshift of the Raman spectrum of the fiber compared to the silicon wafer was measured to be $\Delta \omega = {2.8}\,\textrm{cm}^{-1}$. The measured Raman shift corresponds to a tensile stress of approximately 1.1 GPa [14].
4. Discussion
The shape of the measured temperature profiles during annealing (see Fig. 5 and Fig. 6) can be understood by considering the heating and cooling mechanisms acting on the SCF. Heat is provided through the CO2 laser beam. The amount of deposited heat is set by the laser power. Additionally, varying the annealing speed changes the heating time of each fiber section and thereby the deposited heat. Heat is removed from the fiber only by natural cooling mechanisms: convection, conduction and radiation.
In the stationary case, shown in Fig. 5(a) the heat source is added and removed instantaneously by opening and closing the shutter. It can be seen that the natural cooling of the SCF is slower than the heating by the CO2 laser beam. For the slowest annealing speed $\mbox {v}= {1.0}\,\textrm{mm s}^{-1}$ (see Fig. 5(b)), the measured temperature profile was approximately symmetrical on the heating and the cooling side. Moreover, the location of the maximum temperature in the SCF followed the position at which the SCF was exposed to the CO2 laser beam closely. The long heating time of each fiber section ensured full thermalization of the SCF. This also slowed the cooling of the SCF due to residual heating from the CO2 laser beam. For faster annealing speeds (see Fig. 5(c)-(d)) the laser beam was scanned too fast for full thermalization of the SCF. As a result, the maximum measured temperatures decreased and the exact position of peak temperature in the fiber was delayed relative to the center of the CO2 laser beam. Additionally, the cooling of the SCF was less limited by residual heating of the CO2 laser beam and resembled more the natural cooling of the stationary case. Further, it was observed, that the cooling rate increased with peak temperature respectively with deposited heat (see Fig. 8). The deposited heat depended on the laser power and the annealing speed. A faster annealing speed led to an increase in cooling rate, since residual heating is no longer limiting the cooling of the fiber. It also led to a decrease of the maximum temperature (see Fig. 7). This can be counteracted by higher laser powers at the risk of unwanted fiber deformation. For the fastest annealing speed investigated, the measured cooling rates were slower than the experimentally determined natural cooling rates of the stationary case. This indicates that the cooling rate during annealing can be further increased by either increasing laser power or increasing annealing speed simultaneously with laser power. This may be vital information as higher cooling rates are expected to decrease optical transmission losses in SCF [6].
It might seem surprising that the silicon core of the SCF melts well below the expected ambient pressure melting point of bulk silicon at 1414 °C as highlighted in Fig. 7. To assess this, first the errors of the temperature measurement need to be considered. Reference [10] states a relative error of $\pm {2.6}\,\%$ for the temperature measurement of a typical optical fiber. For the lowest measured temperature of molten silicon at 1281 °C, this error corresponds to a temperature uncertainty of $\pm {33}\,^{\circ}\textrm{C}$.
Additionally, inbuilt stress of the fiber affects the melting point. This stress arises during the fabrication of the fiber explained in Ref. [9]. A preform is heated above the silicon melting point. The glass cladding is softened and acts as a crucible. When the fiber leaves the draw tower furnace it cools down and the core is solidifying. Now, two effects are competing and cause the final stress distribution in the SCF. One is, that silicon expands by $\approx {10}\,\%$ when solidifying. Axially the melt zone can compensate for this rapid expansion while radially the glass cladding can only accommodate a certain amount leaving residual radial compressive stress. The second are the different thermal expansion coefficients of the silicon core ($\approx {2.6}\,\textrm{K}^{-1}$) and the silica cladding ($\approx {0.6}\,\textrm{K}^{-1}$) [15,16]. When the fiber cools down to room temperature, the radial compressive stress will decrease and axial tensile stress will rise. This tensile stress at room temperature can be seen in the downshift of the silicon peak shown in Fig. 9. However, the experiments presented in this study are conducted at higher temperatures. Increased temperatures will lower the axial tensile stress which is expected to disappear at the melting point. The radial compressive stress however increases with increasing temperature. This compressive stress is believed to cause the lowered melting point of the silicon core at 1281 °C. Following the T-P phase diagram of bulk silicon given in Ref. [17] the measured melting point indicates a compressive stress of $\approx {2}\,\textrm{GPa}$.
5. Conclusion
In this study, a direct temperature measurement of a silicon core fiber during CO2 laser annealing has been demonstrated for the first time. The highest measured temperature was $T= {1367}\,^{\circ}\textrm{C}$. At this temperature and even at 1281 °C the silicon core of the fiber was already molten, which was attributed to the compressive stress in the fiber. Furthermore, the cooling rates were determined from the measured temperature profiles and were as high as 3504 °C s−1. This study proves the presented technique to be a viable means of in-situ temperature monitoring during semiconductor core fiber fabrication and post-processing and will help in understanding how optical transmission losses in those fibers can be further reduced.
Funding
Stiftelsen för Strategisk Forskning (RMA15-0135).
Acknowledgments
We thank Prof. Mats Götelid for providing access to the Raman spectrometer.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. A. C. Peacock and N. Healy, “Semiconductor optical fibres for infrared applications: A review,” Semicond. Sci. Technol. 31(10), 103004 (2016). [CrossRef]
2. M. Kudinova, G. Bouwmans, O. Vanvincq, R. Habert, S. Plus, R. Bernard, K. Baudelle, A. Cassez, B. Chazallon, M. Marinova, N. Nuns, and L. Bigot, “Two-step manufacturing of hundreds of meter-long silicon micrometer-size core optical fibers with less than 0.2 dB/cm background losses,” APL Photonics 6(2), 026101 (2021). [CrossRef]
3. A. C. Peacock, U. J. Gibson, and J. Ballato, “Silicon optical fibres–past, present, and future,” Adv. Phys.: X 1(1), 114–127 (2016). [CrossRef]
4. L. Shen, H. Ren, M. Huang, D. Wu, and A. C. Peacock, “A review of nonlinear applications in silicon optical fibers from telecom wavelengths into the mid-infrared spectral region,” Opt. Commun. 463, 125437 (2020). [CrossRef]
5. X. Ji, S. Lei, S. Y. Yu, H. Y. Cheng, W. Liu, N. Poilvert, Y. Xiong, I. Dabo, S. E. Mohney, J. V. Badding, and V. Gopalan, “Single-crystal silicon optical fiber by direct laser crystallization,” ACS Photonics 4(1), 85–92 (2017). [CrossRef]
6. N. Healy, M. Fokine, Y. Franz, T. Hawkins, M. Jones, J. Ballato, A. C. Peacock, and U. J. Gibson, “CO2 Laser-Induced Directional Recrystallization to Produce Single Crystal Silicon-Core Optical Fibers with Low Loss,” Adv. Opt. Mater. 4(7), 1004–1008 (2016). [CrossRef]
7. D. A. Coucheron, M. Fokine, N. Patil, D. W. Breiby, O. T. Buset, N. Healy, A. C. Peacock, T. Hawkins, M. Jones, J. Ballato, and U. J. Gibson, “Laser recrystallization and inscription of compositional microstructures in crystalline SiGe-core fibres,” Nat. Commun. 7(1), 13265 (2016). [CrossRef]
8. J. Ballato and A. C. Peacock, “Perspective: Molten core optical fiber fabrication - A route to new materials and applications,” APL Photonics 3(12), 120903 (2018). [CrossRef]
9. C. M. Harvey, K. Mühlberger, T. Oriekhov, P. Maniewski, and M. Fokine, “Specialty optical fibre fabrication: fibre draw tower based on a CO laser furnace,” J. Opt. Soc. Am. B 38 (2021).
10. C. M. Harvey, K. Mühlberger, and M. Fokine, “Mach-Zehnder interferometer for in-situ non-contact temperature monitoring during thermal processing of an optical fibre,” J. Lightwave Technol. 39(22), 7223–7230 (2021). [CrossRef]
11. J. Wang and J. L. Pressesky, “Quadrature phase-shift interferometer (QPSI) decoding algorithms and error analysis,” Proc. SPIE 5188, 71–79 (2003). [CrossRef]
12. K. Mühlberger, C. M. Harvey, and M. Fokine, “High-performance arduino-based interferometric quadrature phase-shift detection system with 1 nm resolution,” AIP Adv. 11(10), 105304 (2021). [CrossRef]
13. E. Takasuka, E. Tokizaki, K. Terashima, and S. Kimura, “Emissivity of liquid silicon in visible and infrared regions,” J. Appl. Phys. 81(9), 6384–6389 (1997). [CrossRef]
14. F. Bianco, K. Fedus, F. Enrichi, R. Pierobon, M. Cazzanelli, M. Ghulinyan, G. Pucker, and L. Pavesi, “Two-dimensional micro-Raman mapping of stress and strain distributions in strained silicon waveguides,” Semicond. Sci. Technol. 27(8), 085009 (2012). [CrossRef]
15. P. J. A. Sazio, A. Amezcua-Correa, C. E. Finlayson, J. R. Hayes, T. J. Scheidemantel, N. F. Baril, B. R. Jackson, D.-J. Won, F. Zhang, E. R. Margine, V. Gopalan, V. H. Crespi, and J. V. Badding, “Microstructured optical fibers as high-pressure microfluidic reactors,” Science 311(5767), 1583–1586 (2006). [CrossRef]
16. M. Fokine, A. Theodosiou, S. Song, T. Hawkins, J. Ballato, K. Kalli, and U. J. Gibson, “Laser structuring, stress modification and Bragg grating inscription in silicon-core glass fibers,” Opt. Mater. Express 7(5), 1589–1597 (2017). [CrossRef]
17. C. C. Yang, J. C. Li, and Q. Jiang, “Temperature-pressure phase diagram of silicon determined by Clapeyron equation,” Solid State Commun. 129(7), 437–441 (2004). [CrossRef]
