Open Access
Add to BibSonomy
Abstract
Thermal impedance is an important material property which can be used in the design, optimisation and operation of semiconductor lasers. In this study, a new method is proposed for measurement of thermal impedance for all-active semiconductor lasers, using the athermalisation condition. This method is capable of measuring thermal impedance of several device sections at the same time for simple devices, while being capable of measuring the thermal impedance of the device as a whole for more complex structures. Three DBR devices of the same material properties were evaluated and the average impedance length product (ZthL) was determined to be 29.3±2.1 oC μm/mW and 39.33±2.8 oC μm/mW for the gain and grating sections respectively. Additionally, thermal impedance length product of the entire widely tunable Vernier device based on the same material was also determined to be 31.3±0.5oC μm/mW.
Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.
1. Introduction
Semiconductor lasers have been key components in a variety of applications, ranging from gas sensing to newly emerging time and wavelength division multiplexing (TWDM) systems. In particular, all-active devices such as distributed feedback (DFB) lasers or distributed Bragg reflector (DBR) lasers with surface etched slots are critical devices for such applications [1]. A significant difference between all-active and partly passive laser diodes is that carrier density clamps at threshold for the former, only varying slightly with changing injection current, which results in refractive index changes being dominated by temperature changes [2]. A key parameter which is affected by this is diode lasing wavelength, which is linearly dependant on refractive index and hence diode temperature. Diode temperature is given by the sum of the ambient temperature and current induced, self-heating (Joule heating) of the device. While ambient temperature is trivial to measure, temperature rise due to Joule heating is much more difficult to extract. A useful device parameter to calculate the expected temperature rise of the diode is thermal impedance, which is defined as the temperature rise due to the applied electrical power into a particular component, which can be given by,
where $\Delta T$ is the temperature rise of the component in question, and $P_{dis}$ is the power dissipated as heat. Laser diodes above threshold emit light, therefore this can be taken into account when calculated dissipated power, which is then given by where P is the total power input into the laser diode and $P_0$ is the optical output power. However, for typical semiconductor laser diodes, the optical output power is much lower than the dissipated power. In this work, the wall plug efficiency is somewhere on the order of 1 to 5 $\%$ [3,4] and therefore, $P_0$ has been omitted from our analysis. Due to this, $P_{dis}$ only requires the measure of input current and voltage across the junction which are both trivial to measure, therefore we are only required to measure $\Delta T$.Several methods to measure device temperature have been developed, such as one method by Paoli which exploits the temperature dependence of refractive index [5], making use of the lasing wavelength as a temperature probe. However, this technique is only capable of measuring thermal impedance for single section devices and is thereby unsuitable for any distributed Bragg reflector (DBR) or more complex lasers. Thermal imaging provides another approach, with infrared (IR) [6,7] and more recently CCD-Thermoreflectance (CCD-TR) [8–11] imaging yielding accurate, spatially distributed temperature profiles which could then be used to calculate thermal impedance. However, both these methods provide temperature profiles for the surface of the laser which may not lead to an accurate value of thermal impedance and in particular for CCD-TR, it requires a layer of material with high thermoreflectance coefficient on top of the surface of the laser to acquire an image, which may not be possible for some device structures.
Therefore, in this work we present an alternative method to calculate thermal impedance for all-active laser diodes. This technique is capable of measuring thermal impedance of several lasing sections at once for more simple DBR structures, while being capable of extracting the thermal impedance of the entire device for more complicated Vernier laser structures. Using this method, thermal impedance of our lasers has been calculated, which then can be used to calculate average section temperature. Therefore, this could allow significant optimisation of device modelling efforts and produce improved devices for optical networks or other purposes, such as device burn in.
2. Device structure and athermalisation theory
Two types of devices were considered for this study, one being a standard high order grating DBR and another being a widely tunable Vernier design, which also uses a high order slotted grating. These devices were monolithic, all-active lasers with high order surface gratings. Each of the lasers consist of a 2 $\mu$m wide, 1.85 $\mu$m tall ridge. Optical gain is provided by a 5 AlInGaAs quantum well structure with the emission peak centred near 1545 nm at room temperature. The lasers were fabricated on a commercially purchased wafer, which consists of 5 AlGaInAs QWs, above which there is a 1.6 $\mu m$ p-doped InP layer (referred to as cladding here), 50-nm thick, p-doped InGaAsP layer, and a 200 nm InGaAs contact layer. Below the quantum wells there is a 120 $\mu m$ layer of InP. The slots and the ridge of the laser are fabricated using two inductively coupled plasma (ICP) etch steps using Cl$_2$ and N$_2$ gas. Subsequently, the laser is contacted, cleaved and the facets are coated using high reflection (HR) and anti-reflection (AR) coatings as indicated in Figs. 1 and 2. Finally,the laser is bonded to the AlN carrier. This process removes the need for regrowth steps and also leaves the gratings material composition exactly the same as that of the gain sections.
Fig. 1. An example surface grating laser. It is a multi-section device with an active, high order grating section. Note the curved SOA.
Fig. 2. An example of a vernier surface grating laser. It is a multi-section device with two active grating sections. Note the curved SOA.
As these devices are all active, the carrier density clamps at threshold, only varying slightly with changing injection current. Therefore, above threshold, thermal tuning effects dominate and changes in wavelength are primarily from laser self-heating due to variation in injection current. If ambient temperature changes, to maintain its original lasing wavelength, and thereby athermalise the laser diode, the tuning sections (gain and grating) will have to be accordingly retuned, by altering the self-heating through adjustment of the injection current. Therefore, assuming this retuning has been done continuously (mode-hop free) then the gain and grating temperatures should remain relatively constant. Hence, from this we can obtain the change in input power required to offset a change in ambient temperature. This is directly related to thermal impedance. In particular, this is useful as thermal impedance is temperature dependant [12], and as section temperature remains constant during continuous athermalisation, this gives thermal impedance for a given temperature, however it should be noted that this variance was not measured in this study. The following subsections will consider the structural features of the devices under study.
2.1 Distributed Bragg reflector laser
The structure of this laser is given in the Fig. 1. These devices were tested using three different lengths, with the only difference between these devices being the length of their gain section.
For these single grating devices, their reflector section is also split into three different periods to suppress the adjacent reflective peaks of the grating, more detail on which is given in Ref. [13]. The specific grating parameters are given in Table 1. The etch depth of each slot for these devices is 1.35 $\mu m$.
Table 1. Parameters of the triple period grating.
Table 2. Parameters of the two gratings. They are listed in order of which they appear from the photodiode section side to SOA side.
2.2 Widely tunable vernier devices
The structure of this laser is given in Fig. 2.
As shown, the device consists of two grating sections, a gain section, an unused photodiode section (PD) and an 600 $\mu m$ SOA section. The gain section is 500 $\mu m$ long, and the grating parameters are provided in the table below. Each grating contains 9 slots and each slot has a depth of 1.85 $\mu m$. The specific grating parameters are listed in Table 2. This device is operated via the multiplicative Vernier effect, and the tuning is performed via adjusting currents to the gain and both of the grating sections. This device has been previously shown to be capable of covering the entire C-band, with good SMSR and output power [14,15].
3. Measurement and results
To measure thermal impedance, the setup in Fig. 3 has been used.
The lasers were placed onto a copper heatsink which is attached to a thermoelectric cooler (TEC), which is used to monitor and vary the blocks temperature, effectively simulating ambient temperature changes. The lasers are then probed and output light is coupled using a lensed fibre, with the output being split 90$\%$ going to the optical spectrum analyser and 10$\%$ going to a PIN photodiode which is used for coupling optimisation purposes.
3.1 Distributed Bragg reflector lasers
To measure the thermal impedance of the DBR lasers, three laser diodes with the same material properties, of differing cavity lengths 400, 700 and 1000 $\mu$m were continuously athermalised over various temperature ranges. This was done to demonstrate a consistency in the measurement between differing devices and to obtain an average for the wafer material. The athermalisation was achieved by reducing the gain and grating currents with increasing temperature and vice versa) to ensure that the Bragg peak and the cavity modes stay aligned at the right wavelength when temperature is varied. The achieved wavelength stability in each athermalisation was $\pm$0.003 nm / $\pm$0.4 GHz, which is likely much higher precision than required for this measurement. The electrical input power versus ambient temperature for these athermalisations is given in Fig. 4. As expected, a linear relationship for each section (and hence, the whole device) is observed which has been fitted. The thermal impedance can then be calculated from the following relation,
where m is the slope for each respective section. For this case, the thermal impedance for each respective device is listed in Table 3.Fig. 4. Change of input power versus temperature while maintaining the athermalisation condition. A linear regression is used to fit the relationships for each section. (a) 1000 $\mu$m (b) 700 $\mu$m and (c) 400 $\mu$m devices respectively.
Table 3. Thermal Impedance ($Z_{th}$) of the respective lasers and their sections.
Multiplying by the length of each section and converting to milliwatts gives us the impedance length product ($Z_{th}L$), which is useful for comparing thermal impedances for sections of different lengths. This impedance length product in each respective device is given in Table 4.
Table 4. Thermal Impedance ($Z_{th}$) times length (L) of the respective lasers and their sections.
Note, all of the errors in this work have been calculated using standard Gaussian error propagation. It is interesting to note that this product is larger for the grating section in all devices. This is due to the fact that there are slots in the waveguide, resulting in larger current density passing into the grating per milliamp of current. In particular, the value for the 1000 $\mu m$ long laser seems to be significantly higher. This is likely due to the longer gain sections output power varying more significantly with changing current, for example, when the current is reduced to compensate for increasing ambient temperature, the output power is reduced, resulting in a lower photon density entering the grating section which will keep the carrier density higher in the respective section, altering the gratings refractive index and thus increasing the apparent value of the thermal impedance. Thus, there is a limitation if output power varies significantly (assuming any amplification remains constant) over the measurement range. This was not the case for the 400 and 700 $\mu m$ devices (where the power remained relatively constant over the measurement range), however it was the case for the 1000 $\mu m$ laser (where the power dropped by 3 dB over the measurement range). This results in the measurement only being valid for the whole device and not the individual sections for the 1000 $\mu m$ device. Additionally, as one of the fundamental assumptions of the method is that the wavelength change is only a result of temperature change, we expect there to be a low current limitation, as wavelength change in the low current regime is more significantly impacted by carrier effects.
Previously, our group conducted thermal impedance measurements using CCD-TR imaging [11] and the average values obtained in this paper (29.3$\pm$2.1 $^{\circ}$C $\mu$m/mW for the gain section and 39.33$\pm$2.8 $^{\circ}$C $\mu$m/mW for the grating section) seem to be lower than the ones acquired previously using CCD-TR. This could be due to heating at the gold to semiconductor junction, which would have a higher effect on the surface temperature than the waveguide temperature. As CCD-TR is a surface imaging technique, this would explain the higher thermal impedance values shown there. The values acquired here however are in line with other group research outlined in Ref. [16].
Finally, a necessary thing to investigate is how many temperature steps (in this case the steps are 2 $^{\circ}$C) it takes to achieve convergence towards these values. One reason is that a single step might not produce an accurate value, because it is not necessary to align the cavity modes and the reflective peak at the same positions every time to produce the same wavelength, hence this might introduce significant error, however over many steps, this error will average out. Additionally, if too few steps are taken, then it may be plausible to just use the gain section to athermalise or just the grating section, however this single section tuning is not possible without mode hops for more than a few temperature steps. The ambient temperature versus thermal impedance is plotted in Fig. 5. It can be seen that after approximately 10 steps (20 $^{\circ}$C), thermal impedance converges to its steady state value for all sections. This suggests that for this method to be of significant accuracy, an athermalisation range of approximately 20 $^{\circ}$C is required. This level of athermalisation has been achieved by many devices in literature and hence is not a significant limitation to the technique.
Fig. 5. Ambient Temperature versus thermal impedance, measured from the starting temperature of 20 $^{\circ}$C. As can be seen, convergence is achieved over a relatively low temperature range. (a) 1000 $\mu$m (b) 700 $\mu$m and (c) 400 $\mu$m devices respectively.
3.2 Widely tunable vernier devices
As with single grating devices, electrical input power versus temperature was recorded, and is shown in Fig. 6. These Vernier lasers are generally used to provide a very broad tuning range from a single device. However, they can be athermalised keeping their wavelength unchanged even as the ambient temperature changes.
For this device, each section significantly influences the operation of the other two section, and therefore thermal impedance of each individual section is not feasible to measure. Athermalisation of this device is carried out via control of 3 tuning sections rather than the 2 of the DBR lasers.
Fig. 6. Input Power versus Temperature for the Vernier devices. As with the single grating devices a linear relationship is observed.
Thus, there is a much larger set of potential athermal paths that can be taken for a single target wavelength. Overtuning in one section can be compensated by undertuning in the other 2 sections, making it more difficult to isolate the thermal impedance values of any individual section. Additionally, to reduce error in the calculation, several athermal paths were taken and the average thermal impedance was obtained. This thermal impedance for each of the paths is given in Table 5 and as seen, the average (31.3$\pm$0.52$^{\circ}$C $\mu$m/mW) is in line with the values shown in section 3.1. This demonstrates that this method is still valid for widely tunable devices, albeit with increased limitations.
Table 5. Thermal Impedance (TI) of each athermal path for the Vernier laser.
4. Conclusion
A new method using the athermalisation condition to determine thermal impedance is presented. This method allows measurement of thermal impedance for several types of laser devices, with single grating DBRs having the capability of measuring thermal impedance of several laser sections, here being measured as 29.3$\pm$2.1 $^{\circ}$C $\mu$m/mW for the gain section and 39.33$\pm$2.8 $^{\circ}$C $\mu$m/mW for the grating section. For the Vernier devices, only whole device measurements are possible with the resulting average impedance length product achieved being 31.3$\pm$0.5$^{\circ}$C $\mu$m/mW. By extrapolating, this technique allows measurement of device/section average temperature. The results from this technique can be used for design and optimisation of laser diodes. Furthermore, it can be used to monitor device burn in, where thermal impedance is expected to vary significantly, particularly for high-power usage cases. Finally, usage of the technique is non-invasive and does not require special equipment, which is desirable.
Funding
Science Foundation Ireland (15/IA/2854).
Acknowledgements
The authors would like to thank Gaurav Jain with assistance in production of some of the schematics within this manuscript.
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. Abdullaev, Q. Lu, W. Guo, M. J. Wallace, M. Nawrocka, F. Bello, A. Benson, J. O’Callaghan, and J. F. Donegan, “Improved performance of tunable single-mode laser array based on high-order slotted surface grating,” Opt. Express 23(9), 12072–12078 (2015). [CrossRef]
2. L. A. Coldren, S. W. Corzine, and M. L. Mašanović, Diode Lasers and Photonic Integrated Circuits (2012).
3. M. J. Wallace, R. O. Meehan, R. Enright, F. Bello, D. McCloskey, B. Barabadi, E. N. Wang, and J. F. Donegan, “Athermal operation of multi-section slotted tunable lasers,” Opt. Express 25(13), 14414–14426 (2017). [CrossRef]
4. D. Mickus, R. Mckenna, C. Murphy, and J. F. Donegan, “Wide range thermal and athermal operation of slotted surface grating lasers,” Opt. Express 29(11), 16893 (2021). [CrossRef]
5. T. Paoli, “A new technique for measuring the thermal impedance of junction lasers,” IEEE J. Quantum Electron. 11(7), 498–503 (1975). [CrossRef]
6. J. Brown, P. K. Footner, and B. P. Richards, “Failure analysis of plastic encapsulated components–the advantages of IR microscopy*,” J. Microsc. 148(2), 179–194 (1987). [CrossRef]
7. J. Christofferson, K. Maize, Y. Ezzahri, J. Shabani, X. Wang, and A. Shakouri, “Microscale and Nanoscale Thermal Characterization Techniques,” J. Electron. Packag. 130(4), 1 (2008). [CrossRef]
8. S. Wang, C. Xu, F. Duan, B. Wen, S. M. S. Rassel, M. C. Tam, Z. Wasilewski, L. Wei, and D. Ban, “Thermal dynamic imaging of mid-infrared quantum cascade lasers with high temporal–spatial resolution,” J. Appl. Phys. 128(8), 083106 (2020). [CrossRef]
9. D. Pierscinska, K. Pierscinski, M. Morawiec, P. Karbownik, P. Gutowski, and M. Bugajski, “CCD thermoreflectance spectroscopy as a tool for thermal characterization of quantum cascade lasers,” Semicond. Sci. Technol. 31(11), 115006 (2016). [CrossRef]
10. D. Pierscinska, “Thermoreflectance spectroscopy–Analysis of thermal processes in semiconductor lasers,” J. Phys. D: Appl. Phys. 51(1), 013001 (2017). [CrossRef]
11. R. McKenna, D. Mickus, S. Naimi, C. Murphy, M. McDermott, S. Corbett, D. McCloskey, and J. F. Donegan, “Spatially resolved self-heating and thermal impedance of laser diodes using CCD-TR imaging,” OSA Continuum 4(4), 1271–1281 (2021). [CrossRef]
12. J. Paasschens, S. Harmsma, and R. van der Toorn, “Dependence of thermal resistance on ambient and actual temperature,” in Bipolar/BiCMOS Circuits and Technology, 2004. Proceedings of the 2004 Meeting, (2004), pp. 96–99.
13. G. Jain, M. J. Wallace, R. McKenna, K. Brazel, F. Bello, Q. Lu, W. Guo, and J. F. Donegan, “Design Optimization for Semiconductor Lasers With High-Order Surface Gratings Having Multiple Periods,” J. Lightwave Technol. 36(22), 5121–5129 (2018). [CrossRef]
14. M. Nawrocka, Q. Lu, W.-H. Guo, A. Abdullaev, F. Bello, J. O’Callaghan, T. Cathcart, and J. F. Donegan, “Widely tunable six-section semiconductor laser based on etched slots,” Opt. Express 22(16), 18949–18957 (2014). [CrossRef]
15. M. J. Wallace, G. Jain, R. Mckenna, F. Bello, and J. F. Donegan, “Tuning behaviour of slotted vernier widely tunable lasers,” Opt. Express 27(12), 17122–17137 (2019). [CrossRef]
16. I. Mathews, A. Abdullaev, S. Lei, R. Enright, M. J. Wallace, and J. F. Donegan, “Reducing thermal crosstalk in ten-channel tunable slotted-laser arrays,” Opt. Express 23(18), 23380–23393 (2015). [CrossRef]
