ments were performed under continuous DC current with devices mounted on TO-5 headers. No heat sinking was employed. Fig. 2 shows the power as a function of current emitted from samples (a), (b) and (c). The detuned samples can be seen to give more power than the tuned sample over all currents. For the detuned devices, the peak external quantum efficiency is just over 6% for currents below lmA, giving around O.lmW, whilst at 5mA, the efficiency is still a modest 5.4%, giving OSmW of power. Even

more »

... at 13mA, where the device emits 1mW of power, the efficiency is still 4%. The peak efficiency for the tuned device is 4.5% at currents below I d , giving around 0.lmW of power. To deliver OSmW of power, we find that a current of 8mA is required, and the efficiency is 3.2%. External wall-plug efficiencies are similar in value to those stated for external quantum efficiencies. Fig. 2 also shows the output power for the 4nm detuned sample under pulsed current injection, with a l o p long current pulse and lms period. This indicates the increase in power that would be available if the device were to be adequately heat sunk. 60 50 4 4 0 -E pi' $ 30-c 20 10