1.3-μm Laser Reliability Determination for Submarine Cable Systems

Abstract

To meet the stringent requirements of a submarine cable system, our 1.3-μϊα laser prequalification program has two objectives-first, to define the testing methodology that will accurately evaluate the potential reliability of the laser; and second, to obtain a preliminary indication of laser reliability on which the system configuration can be designed. Our testing methodology involves a combination of step-temperature, step-power, and isothermal test ing over the temperature interval between 10 and 80°C and power levels of 1 to 5 mW per facet. Our results show that the long-term degradation process is thermally accelerated, with a median activation energy of 1 eV and a standard deviation of 0.13 eV. By using these activation energies, in conjunc tion with our measurements of degradation rates, we can project laser perform ance to 10°C, i.e., system operating temperature. It is estimated that the median time to failure for "light bulb" operation at 10°C is over 2 X 10 7 hours; and with 98-percent probability it is greater than 5 X 10 6 hours. Hence, when viewed strictly in terms of light bulb sources of stimulated power, these 1.3-μιη lasers have adequate life. In addition, other potential operational malfunc tions are being investigated, and they do not seem to change our basic conclusion about the usefulness of these 1.3-μνα lasers for submarine cable application.

Methodology of Accelerated Aging

“…15 The number of electrons which escape from the active layer and become available for recombi nation as minority carriers in the confining layer has a Boltzmannfactor dependence upon the conduction-band energy step Ei and hence upon the composition x of the confining layer. 16 Because x can vary inadvertently from wafer to wafer, and even across a given wafer, apparently meaningful wafer-to-wafer 12,17 and device-to-device differ ences in the activation factor do not necessarily rule out the possibility of a single degradation mechanism.…”

“…We will not consider such generalizations any further here except to note that when a complete theoretical model in two (i and T) or more inde pendent variables appears to be quite complex, then the empirical onevariable reduced-observable method of eqs. (6), (12), and (54) becomes particularly appealing.…”

“…[22] over a useful temperature range and the approxi mate invariance of T 0 during aging have been reported for certain types of lasers. 6,12 ) Without specifying the time dependence of the isothermal aging law, we define a possibly-time-dependent b a -by the requirement that isothermal aging along the solid-line path 1-5 in Fig. 4 causes ii = i(t = 0, T a ·) to evolve into 4, i.e., b a -4 is defined as the isothermal increase in 4,…”

Effects of having two populations of defects growing in the cavity of a semiconductor laser

“…To simplify the problem, some models suggest that the long-term degradation progression varies in a linear fashion over time. 5,6 The lifetime of a device, under this assumption, can then be estimated linearly after bypassing the transient low-activation-energy modes or burn-in period. 3,6,7,8 Alterna-tively, Kondo et al 9 proposed that the long-term degradation should be described by a cosh function or an exponential function of aging time.…”

“…5,6 The lifetime of a device, under this assumption, can then be estimated linearly after bypassing the transient low-activation-energy modes or burn-in period. 3,6,7,8 Alterna-tively, Kondo et al 9 proposed that the long-term degradation should be described by a cosh function or an exponential function of aging time. Many attempts to describe the aging of devices are device-specific, or failure-mode specific, and do not provide an organized, broader explanation to the generic degradation behavior.…”