High Reliability Evaluation and Lifetime Prediction of 50 GHz Athermal AWG Module

: A 96-channel (50 GHz-spacing) athermal AWG has been developed. It has a wide operating range due to reduced temperature dependence than conventional AWG. The temperature dependence of the center wavelength of the developed module satisﬁed the ± 0.05 nm range in all channels in the temperature range of − 40 ◦ C to 85 ◦ C, and the insertion loss variation was also less than ± 0.5 dB. As a result of validating its reliability through tests based on Telcordia-GR-1209 and GR-1221, the temperature dependence of the center wavelength satisﬁed the ± 0.022 nm range, and the insertion loss variation was also less than ± 0.2 dB. Accelerated life testing showed an expected service life of over 36.7 years, ensuring long-term safety of communication quality in harsh indoor and outdoor environments. test results of ﬁve items including vibration, temperature cycling, and high-temperature storage tests prove that the 96-channel (50 GHz-spacing) athermal AWG module is highly reliable without failure and ensures its reliability for long-term use in optical communication networks.


Introduction
Mobile data traffic continues to increase. Mobile data speed increased from 200 kbps to several Gbps with the transition from 3G over LTE and LTE-A to 5G, and the number of users, antennas, and base stations is increasing exponentially [1]. Low-power multi-channel photonic integrated devices are promising candidates to address the increasing energy consumption and bandwidth demands faced by network operators, and enable efficient deployment of wavelength division multiplexing (WDM) in next-generation passive optical networks (PON) [2,3].
DWDM (dense wavelength division multiplexing) [4][5][6][7] has expanded from longdistance transmission systems to large cities and access networks for many years to meet the rapid advancement, diversification, acceleration, and the large-capacity of communication services based on Internet technologies. In particular, arrayed waveguide grating (AWG) [8] fabricated using polymeric optical waveguides is playing an increasingly significant role in DWDM systems.
Compared to temperature control circuits adopting typical AWGs, the athermal AWGs do not require a power supply. Therefore, thermal AWGs are suitable for use in WDM-PON, VMUX, ROADM, and OXC. DWDM systems are used because of their support for multiple channels, low insertion loss [9], small device size, high stability, and integration with other polymer devices. A wide variety of studies have been conducted on fabricating AWGs in Korea and other countries, and most of them have focused on temperature compensation techniques using the thermal expansion coefficient of metals [10][11][12][13][14][15]. The center wavelength of the output end of an AWG wavelength division multiplexing device fabricated with conventional silica-based PLC (planar lightwave circuits) technologies has a temperature dependence of about 0.011 nm/ • C [10,16], which causes crosstalk between channels, so the temperature of the device needs to be stabilized when applied to DWDM Figure 1 shows the waveguide circuit structure of an arrayed waveguide grating (AWG), in which input waveguides, input slab waveguides, arrayed waveguides, output slab waveguides, and output waveguides are arranged on a substrate. The optical demultiplexer operates as follows. When a WDM signal containing multiple optical signals with wavelengths λ 1~λn enters the input waveguides, it is diffracted and diffused by the input slab waveguide and transmitted to the arrayed waveguide. This signal consists of waveguides placed side by side in each waveguide. The signal emitted from the input slab waveguide propagates, and adjacent waveguides are placed at a fixed optical path length difference of ∆L. Therefore, the signals propagating from each waveguide show a phase difference. Signals passing through the arrayed waveguide enter the output slab waveguide and are diffracted and diffused, but the signals passing through each waveguide interfere with each other and are virtually all diffracted in the direction where the wavefronts are aligned. The phase condition in which the wavefronts are aligned is as follows. n s D sin ∅ + n c ∆L = mλ (1) where n c is the effective index of arrayed waveguides and n s is the effective index of refraction of slab waveguides; ∅ is the diffraction angle; D is the distance between the arrayed waveguides; λ is the wavelength; and m is an arbitrary integer and the order of diffraction.
Appl. Sci. 2021, 11, x FOR PEER REVIEW 2 of 16 crosstalk between channels, so the temperature of the device needs to be stabilized when applied to DWDM systems. However, for active temperature stabilization, the use of cooling devices such as heaters or Peltier Coolers requires power consumption and electronic control circuits, and issues such as price, device size, and reliability limit the widespread use of DWDM systems [17,18]. This study performed reliability and accelerated life tests on a 96-channel (50 GHzspacing) athermal AWG module fabricated by installing a temperature compensation board on the back of a silicon substrate. We tried to verify the reliability for stabilizing the communication quality in the network environment and predict the life by checking the life of more than 10 years [19] required for optical communication parts. Figure 1 shows the waveguide circuit structure of an arrayed waveguide grating (AWG), in which input waveguides, input slab waveguides, arrayed waveguides, output slab waveguides, and output waveguides are arranged on a substrate. The optical demultiplexer operates as follows. When a WDM signal containing multiple optical signals with wavelengths λ1~λn enters the input waveguides, it is diffracted and diffused by the input slab waveguide and transmitted to the arrayed waveguide. This signal consists of waveguides placed side by side in each waveguide. The signal emitted from the input slab waveguide propagates, and adjacent waveguides are placed at a fixed optical path length difference of ΔL. Therefore, the signals propagating from each waveguide show a phase difference. Signals passing through the arrayed waveguide enter the output slab waveguide and are diffracted and diffused, but the signals passing through each waveguide interfere with each other and are virtually all diffracted in the direction where the wavefronts are aligned. The phase condition in which the wavefronts are aligned is as follows.

Principle of AWG
where nc is the effective index of arrayed waveguides and ns is the effective index of refraction of slab waveguides; ∅ is the diffraction angle; D is the distance between the arrayed waveguides; λ is the wavelength; and m is an arbitrary integer and the order of diffraction. ∆ Therefore, signals with different wavelengths converge at different points at the output port of the output slab waveguide, and by placing an output waveguide at each of these points, signals with different wavelengths can be sent to different waveguides to extract signals with wavelengths λ1~λn. Therefore, signals with different wavelengths converge at different points at the output port of the output slab waveguide, and by placing an output waveguide at each of these points, signals with different wavelengths can be sent to different waveguides to extract signals with wavelengths λ 1~λn .

Temperature Dependence of the Center Wavelength
As described above, an AWG is used to multiplex and demultiplex signals of a set wavelength, but this wavelength (center wavelength) changes according to temperature. The refractive index of the waveguide material changes as the temperature changes, and the substrate and waveguide expand or contract, changing the length of the optical path. As a result, the focal position shifts in the input of the input slab waveguide, and the wavelength of the signal entering the input waveguide changes. In Equation (1), setting φ to zero and differentiating the differential equation with respect to λ by temperature T expresses the center wavelength's magnitude of temperature dependence. where dT is the substrate's coefficient of expansion for silicon. Since the coefficient of thermal expansion of the substrate is much greater than that of the waveguide, the coefficient of thermal expansion of the substrate was applied.
The first term to the right of the equation represents the temperature dependence of the refractive index, and the second term is the change in refractive index caused by the stress applied to the waveguide due to changes in the waveguide path's length as the material shrinks or expands. The temperature dependence of the silica glass refractive index is 8 × 10 −6 / • C, and the substrate's coefficient of expansion is 3 × 10 −6 / • C (for silicon, α s ), so the change in center wavelength is about 0.011 nm/ • C.

Principle of Temperature Compensation
This study focused on the linear dispersion of AWG to compensate for this temperature dependence. Figure 2 shows the input slab waveguide of AWG.
As described above, an AWG is used to multiplex and demultiplex signals of a set wavelength, but this wavelength (center wavelength) changes according to temperature. The refractive index of the waveguide material changes as the temperature changes, and the substrate and waveguide expand or contract, changing the length of the optical path. As a result, the focal position shifts in the input of the input slab waveguide, and the wavelength of the signal entering the input waveguide changes. In Equation (1), setting ϕ to zero and differentiating the differential equation with respect to λ by temperature T expresses the center wavelength's magnitude of temperature dependence.
Since the coefficient of thermal expansion of the substrate is much greater than that of the waveguide, the coefficient of thermal expansion of the substrate was applied.
The first term to the right of the equation represents the temperature dependence of the refractive index, and the second term is the change in refractive index caused by the stress applied to the waveguide due to changes in the waveguide path's length as the material shrinks or expands. The temperature dependence of the silica glass refractive index is 8 × 10 −6 /°C, and the substrate's coefficient of expansion is 3 × 10 −6 /°C (for silicon, ), so the change in center wavelength is about 0.011 nm/°C.

Principle of Temperature Compensation
This study focused on the linear dispersion of AWG to compensate for this temperature dependence. Figure 2 shows the input slab waveguide of AWG.  The symbol S corresponds to the position when Ø = 0 and wavelength λ 0 in Equation (1), and is expressed as follows under these conditions.
If the focus point with diffraction angle Øe is defined as E and the distance between S and E as x, then the wavelengths λ and x are expressed as: where L f is the focal length of the slab waveguide; D is the pitch of adjacent arrayed waveguide; n s is the effective refractive index of the slab waveguide; and n g is the group refractive index of the arrayed waveguide. Equation (8) shows the linear dispersion of the AWG, showing that when the input waveguide is at a distance dx from S, it can input light with a different wavelength from λ 0 by dλ. The input waveguide moves a distance corresponding to the wavelength change dλ to compensate for the temperature change and the temperature dependence of the center wavelength. To implement this principle, this study applied a new concept that separates one of the slab waveguides. Figure 3 shows the structure of this new athermal AWG. The AWG's circuit cuts one slab waveguide and separates it into a large part and a small part. These two parts are attached to the temperature compensation board (Invar), which acts as a compensation plate, and the temperature is compensated by the thermal expansion coefficient of SUS 304 bolts. These bolts are connected to both ends of the temperature compensation board.
x + ∆L = m (5) x If the focus point with diffraction angle Øe is defined as E and the distance between S and E as x, then the wavelengths λ and x are expressed as: where Lf is the focal length of the slab waveguide; D is the pitch of adjacent arrayed waveguide; ns is the effective refractive index of the slab waveguide; and ng is the group refractive index of the arrayed waveguide. Equation (8) shows the linear dispersion of the AWG, showing that when the input waveguide is at a distance dx from S, it can input light with a different wavelength from λ0 by dλ. The input waveguide moves a distance corresponding to the wavelength change dλ to compensate for the temperature change and the temperature dependence of the center wavelength. To implement this principle, this study applied a new concept that separates one of the slab waveguides. Figure 3 shows the structure of this new athermal AWG. The AWG's circuit cuts one slab waveguide and separates it into a large part and a small part. These two parts are attached to the temperature compensation board (Invar), which acts as a compensation plate, and the temperature is compensated by the thermal expansion coefficient of SUS 304 bolts. These bolts are connected to both ends of the temperature compensation board.  Figure 4 shows the temperature compensation mechanism. In conventional AWGs, the focus point shifts according to temperature change, and the center wavelength of AWG varies with temperature.   However, in an athermal AWG, the focus point also shifts according to temperature change, but the center wavelength remains constant because the output waveguide moves to the shifted focus point by the thermal contraction and expansion of SUS 304 bolts. This temperature compensation mechanism also applies when light is input from the output port. Athermal AWG uses a technology to control the focus in slab waveguides, so it is not temperature sensitive and has an insignificant effect on optical properties. This implements an athermal AWG with many channels. The compensation is also very stable because the thermal expansion coefficient of the SUS304 bolts is constant and very stable. However, in an athermal AWG, the focus point also shifts according to temperature change, but the center wavelength remains constant because the output waveguide moves to the shifted focus point by the thermal contraction and expansion of SUS 304 bolts. This temperature compensation mechanism also applies when light is input from the output Appl. Sci. 2021, 11, 11107 5 of 15 port. Athermal AWG uses a technology to control the focus in slab waveguides, so it is not temperature sensitive and has an insignificant effect on optical properties. This implements an athermal AWG with many channels. The compensation is also very stable because the thermal expansion coefficient of the SUS304 bolts is constant and very stable.
To accurately compensate for the temperature dependence of the athermal AWG's center wavelength, the distance of the temperature compensation board, determined as follows, should be adjusted by referring to the parameters in Table 1. The change in position compensation dx can be expressed as a function of temperature change.
Equation (11) below is derived by using the parameters in Table 1.
That is, when the temperature changes by 1 • C, the focus point must change by 0.280 µm/deg in the input end of the input slab waveguide.
Therefore, the temperature compensation board needs to compensate for this value. The thermal expansion coefficient of the SUS304 bolts is 1.73 × 10 −5 / • C, so the distance of the temperature compensation board becomes 48.55 mm, according to the design of AWG.

Athermal AWG Fabrication Results
Based on the principle of temperature compensation and patent described above, a compact athermal AWG module Gaussian-type (130 mm × 90 mm × 11 mm) was fabricated using 96-channel (50 GHz-spacing) athermal AWG chip technology and PLC manufacturing technology, as shown in Figure 5 below.

Athermal AWG Fabrication Results
Based on the principle of temperature compensation and patent described above, a compact athermal AWG module Gaussian-type (130 mm × 90 mm × 11 mm) was fabricated using 96-channel (50 GHz-spacing) athermal AWG chip technology and PLC manufacturing technology, as shown in Figure 5 below.
Based on Telcordia GR-1221-CORE [19] and Telcordia-GR-1209-CORE [20] for passive optical components, reliability tests were performed on athermal AWG modules under the conditions in Table 2 to measure the performance characteristics before and after the test at room temperature.
Five samples were used for each reliability test to compare the center wavelength and insertion loss values before and after the tests at room temperature. Figure 6 shows the spectra over all channels for the 96-channel (50 GHz-spacing) athermal AWG modules and the spectrum of one of the samples before the test.  Based on Telcordia GR-1221-CORE [19] and Telcordia-GR-1209-CORE [20] for passive optical components, reliability tests were performed on athermal AWG modules under the conditions in Table 2 to measure the performance characteristics before and after the test at room temperature. Five samples were used for each reliability test to compare the center wavelength and insertion loss values before and after the tests at room temperature. Figure 6 shows the spectra over all channels for the 96-channel (50 GHz-spacing) athermal AWG modules and the spectrum of one of the samples before the test.

Athermal AWG Fabrication Results
Based on the principle of temperature compensation and patent described above, a compact athermal AWG module Gaussian-type (130 mm × 90 mm × 11 mm) was fabricated using 96-channel (50 GHz-spacing) athermal AWG chip technology and PLC manufacturing technology, as shown in Figure 5 below.
Based on Telcordia GR-1221-CORE [19] and Telcordia-GR-1209-CORE [20] for passive optical components, reliability tests were performed on athermal AWG modules under the conditions in Table 2 to measure the performance characteristics before and after the test at room temperature.
Five samples were used for each reliability test to compare the center wavelength and insertion loss values before and after the tests at room temperature. Figure 6 shows the spectra over all channels for the 96-channel (50 GHz-spacing) athermal AWG modules and the spectrum of one of the samples before the test.     Figure 7 shows the wavelength and the loss variation measured and monitored during the temperature cycling test. The center wavelength shift and insertion loss values of channels 1, 48, and 96 were measured in the temperature cycling test (−40 • C~85 • C). Figure 7a shows the temperature dependence of the center wavelength for three wavelengths (1 ch., 48 ch., 96 ch.) of the 96-channel (50 GHz-spacing) athermal AWG module, confirming that the temperature was compensated. Figure 7b shows the temperature dependence of spectrum 48 ch., in which the center wavelength shift at 85 • C was ±0.05 nm. Figure 7c shows the temperature dependence of the center wavelength for the AWG module (1 ch., 48 ch., 96 ch.) between −40 • C~85 • C, in which the temperature compensation was less than ±0.05 nm. Figure 7d shows that the insertion loss variation was less than ±0.5 dB in the range of −40 • C~85 • C.
(1 ch., 48 ch., 96 ch.) of the 96-channel (50 GHz-spacing) athermal AWG module, confirming that the temperature was compensated. Figure 7b shows the temperature dependence of spectrum 48 ch., in which the center wavelength shift at 85 °C was ±0.05 nm. Figure 7c shows the temperature dependence of the center wavelength for the AWG module (1 ch., 48 ch., 96 ch.) between −40 °C~85 °C, in which the temperature compensation was less than ±0.05 nm. Figure 7d shows that the insertion loss variation was less than ±0.5 dB in the range of −40 °C~85 °C.

Reliability Test Result for Outdoor Environments Application
Based on Telcordia-GR-1221-CORE [19] and Telcordia-GR-1209-CORE [20] for passive optical components, reliability tests were performed on the 96-channel (50 GHz-spacing) athermal AWG modules under the conditions in Table 3 to obtain the results. Five samples were used for each reliability test item to measure and compare the center wavelength, insertion loss values, and insertion loss variation before and after the tests at room temperature.
According to the reliability test results in Table 3, the maximum insertion loss change was within ±0.19 dB, which satisfies the 0.5 dB or 10% requirements of Telcordia-GR-1221-CORE. The average insertion loss is the average value of the 1 to 96 channels of five samples tested separately and was within a maximum of −3.72 dB, which is less than the Telcordia-GR-1209-CORE reference of 6.14 dB.
Among the five reliability tests, the insertion loss average of the low-temperature storage test showed the greatest loss of −3.54 dB. The average insertion loss changes of the five reliability items were between −0.08 dB and 0.03 dB, indicating that the performance was stable even after the reliability tests without significant differences.  Figure 8 shows the 96-channel average data of five AWG modules after the temperature cycling test. Figure 8a shows the center wavelength shift, and Figure 8b the insertion loss variation. After testing 500 cycles (−40 • C~85 • C), the center wavelength shift was within ±0.022 nm and the insertion loss variation within ±0.22 dB, which is stable against temperature changes.  Figure 9 shows the 96-channel average data of five AWG modules after the hightemperature storage reliability test at 85 • C for 2000 h. The center wavelength shift was within ±0.060 nm, and the insertion loss variation was within ±0.15 dB, which is stable for a long time under a high-temperature environment.  Figure 10 shows the 96-channel average data of five AWG modules after the lowtemperature storage reliability test at −40 • C for 2000 h. The center wavelength shift was within ±0.030 nm, and the insertion loss variation was within ±0.35 dB, which is stable for a long time under a low-temperature environment.
The reliability test results of five items including vibration, temperature cycling, and high-temperature storage tests prove that the 96-channel (50 GHz-spacing) athermal AWG module is highly reliable without failure and ensures its reliability for long-term use in optical communication networks. The reliability test results of five items including vibration, temperature cycling, and high-temperature storage tests prove that the 96-channel (50 GHz-spacing) athermal AWG module is highly reliable without failure and ensures its reliability for long-term use in optical communication networks.

Accelerated Life Test
Accelerated life tests were performed along with the reliability tests to predict the service life of the 96-channel (50 GHz-spacing) athermal AWG. The test estimated the lifestress relationship using the Arrhenius equation by elevating temperatures according to the Meeker and Hahn plan (optimized 4:2:1 allocation), which considers stress extrapolation and time extrapolation to increase the accuracy of lifetime prediction [21].

Preparation
Based on the accelerated life test conditions in Table 4, the samples were measured at 250-h intervals for 4000 h using three high-temperature chambers (ESPEC), an optical source (Amonics), and an Optical Spectrum Analyzer (Anritsu, MS9740A) at three different temperatures, 71 °C, 83 °C, and 95 °C. The failure criteria were based on accidental

Accelerated Life Test
Accelerated life tests were performed along with the reliability tests to predict the service life of the 96-channel (50 GHz-spacing) athermal AWG. The test estimated the lifestress relationship using the Arrhenius equation by elevating temperatures according to the Meeker and Hahn plan (optimized 4:2:1 allocation), which considers stress extrapolation and time extrapolation to increase the accuracy of lifetime prediction [21].

Preparation
Based on the accelerated life test conditions in Table 4, the samples were measured at 250-h intervals for 4000 h using three high-temperature chambers (ESPEC), an optical source (Amonics), and an Optical Spectrum Analyzer (Anritsu, MS9740A) at three different temperatures, 71 • C, 83 • C, and 95 • C. The failure criteria were based on accidental failure or a 50% change in the initial measurement. Data reliability was improved by measuring channels 1 and 48. A total of 28 samples (Gaussian type 21, Flat type 7) were distributed by temperature. As shown in Table 5, the likelihood function values of the Weibull distribution, lognormal distribution, and exponential distribution were compared using ALTA S/W to find a suitable distribution for the data measured for 4000 h during the accelerated life test. As a result, the lognormal distribution, which had the largest likelihood function value, was the most suitable lifetime distribution.  Figure 11 shows the failure data observed in the accelerated life test using ALTA S/W. Since the data of each test condition are close to a straight line and the lifetime distribution estimation line is parallel, the lognormal distribution is suitable, and acceleration is established between the test conditions. failure or a 50% change in the initial measurement. Data reliability was improved by measuring channels 1 and 48. A total of 28 samples (Gaussian type 21, Flat type 7) were distributed by temperature. As shown in Table 5, the likelihood function values of the Weibull distribution, lognormal distribution, and exponential distribution were compared using ALTA S/W to find a suitable distribution for the data measured for 4000 h during the accelerated life test. As a result, the lognormal distribution, which had the largest likelihood function value, was the most suitable lifetime distribution.  Figure 11 shows the failure data observed in the accelerated life test using ALTA S/W. Since the data of each test condition are close to a straight line and the lifetime distribution estimation line is parallel, the lognormal distribution is suitable, and acceleration is established between the test conditions.

Lognormal Distribution and Life-Stress Relationship
The life distribution model was based on a lognormal distribution, and the life-stress relationship uses the Arrhenius relationship, which is widely used in accelerated life tests by temperature, as shown in Equation (12) below. Figure 11. Lognormal probability plot.

Lognormal Distribution and Life-Stress Relationship
The life distribution model was based on a lognormal distribution, and the life-stress relationship uses the Arrhenius relationship, which is widely used in accelerated life tests by temperature, as shown in Equation (12) below.
where ζ(T) is the lifetime distribution parameter; E is the activation energy (eV); k is the Boltzmann constant (8.617 × 10 5 eV/ • C); T is the absolute temperature (K, • C + 273.15); and A is a constant according to the material properties and test conditions.
The acceleration factor (AF) from the life-stress relationship is shown in Equation (13) where T d is the lifetime under operating conditions, and T a is the lifetime under accelerated conditions. The activation energy (Ea) estimated from the software was 0.482. Therefore, as a result of calculating the acceleration factors for each acceleration stress, as shown in Figure 12, the life-stress relationship for temperatures 71 • C, 83 • C, and 95 • C shows that the lifetime reduced as the temperature increased. where ζ(T) is the lifetime distribution parameter; E is the activation energy (eV); k is the Boltzmann constant (8.617 × 10 5 eV/°C); T is the absolute temperature (K, °C + 273.15); and A is a constant according to the material properties and test conditions. The acceleration factor (AF) from the life-stress relationship is shown in Equation (13) below. AF = ζ(Td)/ζ(Ta) = exp[(E/K)·(1/Td − 1/Ta)] (13) where Td is the lifetime under operating conditions, and Ta is the lifetime under accelerated conditions. The activation energy (Ea) estimated from the software was 0.482. Therefore, as a result of calculating the acceleration factors for each acceleration stress, as shown in Figure  12, the life-stress relationship for temperatures 71 °C, 83 °C, and 95 °C shows that the lifetime reduced as the temperature increased.  Table 6 shows the failures that occurred under the temperature conditions during the accelerated life test for 4000 h. The samples consisted of 28 96-channel (50 GHz-spacing) athermal AWGs (Gaussian type 21, Flat type 7). Four failures occurred only in the Gaussian type. After the accelerated life test, the insertion loss for each channel of each representative sample was measured, as shown in Figure 13. After the high-temperature acceleration test, the average insertion loss was −4.35 dB at 71 °C, −4.89 dB at 83 °C, and −4.68 dB at 95 °C, and the higher the temperature, the greater the loss tends to be.

Accelerated Life Test Results
Although the high-temperature accelerated life test lasted for 4000 h, it satisfied the Telcordia-GR-1209 standards and showed reliability in wavelength.  Table 6 shows the failures that occurred under the temperature conditions during the accelerated life test for 4000 h. The samples consisted of 28 96-channel (50 GHz-spacing) athermal AWGs (Gaussian type 21, Flat type 7). Four failures occurred only in the Gaussian type. After the accelerated life test, the insertion loss for each channel of each representative sample was measured, as shown in Figure 13. After the high-temperature acceleration test, the average insertion loss was −4.35 dB at 71 • C, −4.89 dB at 83 • C, and −4.68 dB at 95 • C, and the higher the temperature, the greater the loss tends to be. Although the high-temperature accelerated life test lasted for 4000 h, it satisfied the Telcordia-GR-1209 standards and showed reliability in wavelength.

Accelerated Life Test Results
The accelerated tests on the athermal AWG module under high-temperature stress followed a lognormal distribution. The activation energy calculated by ALTA S/W was 0.482 eV, and the predicted mean life under 40 • C operating conditions was up to 3.22 × 10 5 (about 36.7 years), which guarantees the 10-year service life required for optical communication components. Although an accelerated life test of 4000 h was performed at high temperature, the wavelength showed reliability. The accelerated tests on the athermal AWG module under high-temperature stress followed a lognormal distribution. The activation energy calculated by ALTA S/W was

Conclusions
The temperature dependence of the center wavelength in the 96-channel (50 GHzspacing) athermal AWG module developed based on a new temperature compensation board shape and patent satisfied the ±0.05 nm range in all channels in the temperature range of −40 • C~85 • C, and the insertion loss variation was also less than ±0.5 dB.
As a result of performing reliability tests based on Telcordia-GR-1209 and GR-1221, the temperature dependence of the center wavelength satisfied the ±0.022 nm range, and the insertion loss variation was also less than ±0.2 dB. According to the accelerated life test by high-temperature stress, the predicted mean life of the AWG module was 3.22 × 10 5 (about 36.7 years), which is more than enough to guarantee the 10-year service life required for optical communication components.
The developed athermal AWG module showed sufficient performance and reliability to be applied to 5G mobile networks such as reducing the power consumption and volume of DWDM network systems and enabling easy management.