Junction temperature

The junction temperature $T_j$ is the temperature of the light-emitting active region of a semiconductor laser or LED. It sets the threshold, wavelength, efficiency and lifetime, and it exceeds the case or heatsink temperature a controller displays, because the electrical power not emitted as light is released as heat in and near the active region. For a telecom laser on a cooled mount the difference is typically a few kelvin to about 10 K; for high-power chips it can be considerably larger.

Temperature rise from dissipated power

The dissipated power is the electrical input minus the optical output, and the rise above the heatsink is that power times the thermal resistance between the active region and the heatsink:

$$P_\text{diss} = I V - P_\text{opt},$$ $$T_j = T_\text{hs} + R_\text{th}\,P_\text{diss}.$$

For a 1550 nm DFB laser driven at 100 mA and 1.4 V, emitting 20 mW, the input is 140 mW and the dissipated power 120 mW. With $R_\text{th} = 60$ K/W, the upper end of the 30–60 K/W range typical of a DFB on a submount, the junction runs 7.2 K above the heatsink. On a heatsink held at 70 °C it sits at 77.2 °C. Using the electrical input instead of the dissipated power overstates the rise, by 17% in this example and by more for efficient high-power lasers.

A thermoelectric cooler controls the temperature at its sensor, usually a thermistor on the submount, so a stabilized laser still has its junction above the set point by $R_\text{th} P_\text{diss}$, and that offset changes whenever the drive current changes.

Effects on the device

Wavelength. A DFB laser tunes at about 0.1 nm/K and a Fabry-Perot laser's gain peak at 0.4–0.5 nm/K near 1550 nm (wavelength temperature coefficient). The 7.2 K rise above therefore moves the DFB line by about 0.7 nm relative to its isothermal value, and would move a Fabry-Perot gain peak by roughly 3–3.6 nm.

Threshold and efficiency. The threshold current grows as $\exp(\Delta T/T_0)$ with the characteristic temperature $T_0$; for $T_0 = 60$ K the 7.2 K rise raises it by 13%. The slope efficiency falls at the same time. At high current the heating feeds back on itself until the light output peaks and falls in thermal rollover.

Lifetime. Most wear-out mechanisms are thermally activated and follow an Arrhenius law in the absolute junction temperature, so a lifetime figure applies only at the junction temperature it was quoted for, and screening procedures such as burn-in are specified by junction temperature and current.

LEDs behave the same way: output falls and the emission shifts to longer wavelength as the junction heats.

Measurement

The junction is not accessible to a thermocouple, so its temperature is inferred from a temperature-sensitive parameter of the device itself, calibrated under conditions where self-heating is negligible. With pulses of about 1 µs or shorter at 1% duty cycle, the average temperature rise in the example above falls a hundredfold, to about 0.07 K; the rise within each pulse is larger, because the active region heats on a sub-microsecond to microsecond timescale, so shorter pulses are used where this matters.

• Wavelength-shift method: the lasing wavelength of a DFB is recorded against heatsink temperature under pulsed drive, then measured under CW drive at the same current. The red shift divided by the calibrated coefficient gives $\Delta T_j$.
• Forward-voltage method: at a small fixed sense current the forward voltage falls by about 1–2 mV/K. After CW operation the current is switched to the sense value within microseconds, before the junction cools, and the voltage is converted to temperature with the calibration. At −1.5 mV/K the 7.2 K rise is 10.8 mV, and a voltage resolution of 0.1 mV corresponds to about 0.07 K.
• Pulsed versus CW LIV: comparing the two curves shows how much of the threshold rise and slope loss is thermal.

Both main methods yield $R_\text{th}$ as a by-product. The procedures and their corrections are given in Measuring laser diode junction temperature, and the junction temperature calculator applies the formula above.

Common questions

What is the difference between junction temperature and case temperature?

The case temperature is measured on the package or heatsink; the junction temperature is that of the active region. They differ by $R_\text{th} P_\text{diss}$, which is zero only when no power is dissipated.

How hot does a laser diode junction get?

It depends on the thermal resistance and the drive. A telecom DFB typically runs a few kelvin to about 10 K above its submount; poorly mounted chips and high-power devices can run tens of kelvin hotter, which is where rollover and accelerated aging begin.

Why do datasheets specify pulsed measurements?

Short, low-duty pulses keep the junction at the heatsink temperature, so the measured threshold and wavelength describe the device at a known temperature. Under CW drive the same parameters include the self-heating.

References: L. A. Coldren, S. W. Corzine and M. L. Mašanović, Diode Lasers and Photonic Integrated Circuits, 2nd ed. (Wiley, 2012); G. P. Agrawal and N. K. Dutta, Semiconductor Lasers, 2nd ed. (Van Nostrand Reinhold, 1993); M. Fukuda, Reliability and Degradation of Semiconductor Lasers and LEDs (Artech House, 1991).