nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

How to use JMP to estimate activation energy and current coefficient according to the Arrnuius model

Valerie_Nedbal

Hi there,

a customer would like to have some guidance on how to use JMP to estimate activation energy

and current coefficient according to the Arrenius model. This customer was previously using Reliasoft ALTA for this purpose. It would be great to have an answer until end of this wee

Thanks in advance,

Cheers,

Valerie

Accepted answers

Phil_Kay

Hi Valerie,

Search for "arrhenius". Jmp can fit that model in the destructive degradation and fit life by x platforms, I think. Otherwise nonlinear platform might be a solution.

Regards,
Phil

All comments

Phil_Kay

Hi Valerie,

Search for "arrhenius". Jmp can fit that model in the destructive degradation and fit life by x platforms, I think. Otherwise nonlinear platform might be a solution.

Regards,
Phil

Valerie_Nedbal

Hi all,

Thanks - another question:

The kind of plots we use when reporting wear-out life-times for semiconductor lasers is the

distribution plot at use condition.

Question 1: From the “Fit Life by X” - can I use the activation energy from the model to recalculate all time-to-failures

to typical use conditions (need to generate a new column with data) for generating a distribution plot (see example below).

Is that easily accessible in JMP?

That would typically look like this in Reliasoft ALTA. A nice function in ALTA was that you could choose unit (hours/days/Years) on

the presented time-scale. The accelerated life-times are normally measured in hours, but life-time at use condition is

more interesting in years.

The accellerated life-test would typically consist of several temperatures and currents.

Time-to-failrue ~ I^N*exp (Ea/(Kb*T))

Where

I is the current

N is the current exponent

T is the temperature (K)

Reported current coefficients (N) in the literature are in the range of 2-4 for semiconductor lasers.

Question 2: Can JMP estimate N?

In ALTA you could choose between “Arrenius” model or “Temperature-Non-Thermal” model,

Which then would give me a table including: Ea, N, sigma, µ including confidence limits.

These four parameters would typically be included in the reliability report presented to the customers.

Practically you often choose to “lock” N=2 to get more reliable prediction on TTF in the final analysis, but I always want to

run a full model to confirm that the new lasers quantum wells and material design would show N~2.

An example, below are median time-to-failures for different stress groups (data is from an old product, that is no longer on the market)

It would be a nice time-saver if you could get the distribution plot graphs below the table without having to calculate the “corrected TTF”

at typical current (7mA) and temperature (371K). In a real analysis, the distribution plot includes all lasers in test, not the median TTF for each group.

Fit Life by X.jpg Screen Shot 05-30-17 at 09.30 AM.PNG

malcolm_moore1

Hi Valerie, It looks like your customer wants to replicate the following model in JMP: http://reliawiki.com/index.php/Temperature-NonThermal_Relationship Unfortunately Fit Life by X constrains us to fitting models involving one acceleration factor. I think to fit the above model would need use of non-linear model and definition of the appropriate model (T-NT Exponential, T-NT Weibull or T-NT Lognormal) as a formula to be used in the non-linear modelling. From the fitted model the summary information being requested could then be obtained. Perhaps you can check with your customer that this would be an acceptable solution? Would be good to also understand a little more about the motivational drivers for potentially switching from Reliasoft to JMP. Best, Malcolm