Estimation of lifetime distributions on 1550-nm DFB laser diodes using Monte-Carlo statistic computations

Abstract

High performances and high reliability are two of the most important goals driving the penetration of optical transmission into telecommunication systems ranging from 880 nm to 1550 nm. Lifetime prediction defined as the time at which a parameter reaches its maximum acceptable shift still stays the main result in terms of reliability estimation for a technology. For optoelectronic emissive components, selection tests and life testing are specifically used for reliability evaluation according to Telcordia GR-468 CORE requirements. This approach is based on extrapolation of degradation laws, based on physics of failure and electrical or optical parameters, allowing both strong test time reduction and longterm reliability prediction. Unfortunately, in the case of mature technology, there is a growing complexity to calculate average lifetime and failure rates (FITs) using ageing tests in particular due to extremely low failure rates. For present laser diode technologies, time to failure tend to be 106 hours aged under typical conditions (P01O mW and T80°C). These ageing tests must be performed on more than 100 components aged during 10000 hours mixing different temperatures and drive current conditions conducting to acceleration factors above 300-400. These conditions are highcost, time consuming and cannot give a complete distribution of times to failure. A new approach consists in use statistic computations to extrapolate lifetime distribution and failure rates in operating conditions from physical parameters of experimental degradation laws. In this paper, Distributed Feedback single mode laser diodes (DFB-LD) used for 1550 nm telecommunication network working at 2.5 Gbitls transfert rate are studied. Electrical and optical parameters have been measured before and after ageing tests, performed at constant current, according to Telcordia GR-468 requirements.Cumulative failure rates and lifetime distributions are computed using statistic calculations and equations of drift mechanisms versus time fitted from experimental measurements. CONTEXT AND OBJECTIVESHigh-rate optical system performances are strongly related to micro-optoelectronic device parameters inserted into the transmitter and receiver blocks used for telecommunication applications. These blocks contain InP photonic and GaAs or InP electron devices. It has been already demonstrated that hybrid or monolithic integration used to assemble these modules (OEICs) suffer from various intrinsic and process dependent parasitic effects. In this case, reliability estimation is traditionally based on life-testing and a current approach is to apply Telcordia requirements (468GR) for optoelectronic applications [1]. But actual levels of reliability regarding long time to failure and very low failure rates lead to a dramatic increase difficulty for experimental evaluation. Moreover the complexities of system using infrared optics, used for align component and fiber, increase the difficulty to localize degraded zones responsible to optical power loss. Actual performances...

Electroluminescence spectroscopy for reliability investigations of 1.55 μm bulk semiconductor optical amplifier

“…This last assumption is in good agreement with simulation and experimental results reported on 1550 nm InP SSC-SOA by H. Kamioka (NTT) and G. Talli (Corning) [6,14]. Cumulative failure function and first times of failure have been calculated using an approach, which has been already presented in a previous paper [15]. The most sensitive failure indicator, pseudo-threshold current for an optical power of 0.6 mW (I 0.6 ), was specifically monitored and variation versus time was fitted by a well-known power law [ fig.…”

A unified multiple stress reliability model for microelectronic devices — Application to 1.55 μm DFB laser diode module for space validation

“…Degradation (nature and localization) of DFB-LD technology has been well described for example in Ref. [31] were the following failure mechanisms were recalled: i. Intrinsic defects located between active and blocking layer (due to misfits or threading dislocations), induce an increase of threshold current I th . While an increase of misfit concentration in active zone implies an increase of internal number of carriers injected in the active zone increasing the rate of non-radiative recombination in active area.…”

Reliability investigation and failure analysis of 25Gb/s 850 nm oxide-confined VCSELs

“…Due to time constraints, we extrapolated the aging trend of the devices by fitting the decay curve and extrapolating to the time corresponding to the EOL criterion. In the case of semiconductor lasers, a power-law function is commonly chosen as the extrapolation function, as it effectively fits the experimental data [3,4]. The extrapolation function is represented by Equation 1:…”