1. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. Gain-Cavity Alignment in Efficient Visible (660nm) VCSELs Studied Using High Pressure Techniques T.E.Sale, S.J.Sweeney, G.Knowles, A.R.Adams, Department of Physics, University of Surrey, Guildford, GU2 7XH, U.K. Tel: +44 1483 876813 Fax: +44 1483 876781 e-mail: t.sale@surrey.ac.uk J.Woodhead & S.M.Pinches. Department of electronic & electrical engineering, University of Sheffield, Mappin Street, Sheffield, S1 3JD, U.K.

2. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. 660nm Vertical-Cavity Surface-Emitting Lasers for polymer fibre applications are very sensitive to temperature and structural errors. Two devices with a 3meV wavelength separation were studied and significant differences in the output performance can be explained by the relative offset of the cavity mode and the peak of the gain spectrum. It also explains the different behaviour as a function of hydrostatic pressure. The temperature and pressure dependence can be used to shift the offset but self-heating means that the c.w. variation is not always as expected. Electron leakage also changes with both pressure and temperature, and in both cases its effect is dominant over the gain-cavity offset in determining the threshold current variation. Abstract:

3. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15.  VCSELs operating close to 650nm are of use in low cost polymer optical fibre datacomms systems over short distance (<100m) at speeds up to 10Gb.s -1 per channel.  Here we investigate two ~660nm devices taken from different parts of a large section of fabricated wafer. They differ only in the energy of the cavity resonance.  We use a liquid (Essence F) filled piston and cylinder system to apply hydrostatic pressure to the devices to change the offset between the QW gain peak and cavity (lasing) energy. Leakage currents via the X-minimum also increase as a consequence of reducing the Г-X splitting with pressure.  Very significant differences in the c.w. and pulsed L-I’s as a function of pressure can be readily explained by the small difference in resonant wavelength alone. Introduction:

4. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. X  Energy, E Electrons leak from Γ in active region through X states in p-DBR causing leakage current and loss of quantum efficiency via free carrier absorption, activation energy approx. 250meV, decreases at - 9.1meV/kbar. Carrier densities in high 10 18 cm -3 readily obtained.

5. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. Substrate AlAs/Al 0.5 Ga 0.5 As InGaP/AlGaInP Contact Metalisation Outline device structure 1x8 Array used with POF Double sided 1x8 VCSEL chip

6. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. C.W. L-Is of devices A & B. in air, 0kbar 20°C.  Device A E cav =1877.4meV Device B E cav = 1874.8meV  Threshold current I thB <I thA due to better gain-cavity alignment.  Device A rolls over at a higher current I max.  Useful operating range (I th - I max ) is similar for both, but higher electron leakage in A due to higher I max, causes absorption in output mirror and reduced efficiency. I maxB I maxA I thB I thA P maxB P maxA

7. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15.  We need to consider the pressure and temperature dependence of the gain peak (E gmax ) and cavity energy (E cav ). Over a suitable range they are described by:  The partial derivatives are found by experiment:  The devices heat above ambient (T r =20°C) by Ohmic heating at a rate 7.7°C.mA -1. Pressure is applied in a piston cylinder system.

8. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. Device A. High pressure measurements in Essence F  E qw <E cav  I th (pulsed) <I th (c.w.) at 0 kbar since Ohmic heating worsens gain-cavity alignment.  Applied pressure improves initial alignment but I th (pulsed) increases with pressure as electron leakage dominates the threshold current.  I th (c.w.) increases even faster as leakage current heats device and further adds to leakage.  c.w. I th & I max in air.  pulsed I th in air

9. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15.  E qw >E cav  I th (c.w.) <I th (pulsed) at 0kbar since d.c. current improves gain-cavity alignment.  Applied pressure improves alignment but I th (pulsed) increases as leakage dominates the threshold.  At higher pressures I th (c.w.) >I th (pulsed) as gain- cavity crossover passed. Device B. High pressure measurements in Essence F  c.w. I th & I max in air.  pulsed I th in air

10. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15.  E cav (0, 20°C))=1877.4meV (660.5nm)  E gmax (0,20°C) =1875meV  E cav & E qw cross at ~0.6kbar  For c.w. operation threshold E gmax & E cav do not cross at any pressure due to current dependent heating which keeps the two energies at almost constant separation.  Energies are computed from experimental data using pressure, temperature and current coefficients described earlier. Device A. Pressure dependence of gain, cavity and roll-over energies.

11. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15.  E cav (0, 20°C))=1874.8meV (661.4nm)  Pulsed E cav & E qw diverge with applied pressure.  For c.w. injection, Ohmic heating causes initial convergence of E cav & E gmax  E cav & E gmax are better aligned for c.w. operation in this device due to the very significant temperature rise (~15 °C) at threshold. Device B. Pressure dependence of gain, cavity and roll-over energies.

12. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. Conclusions:  Very small differences (<3meV, or 1nm) in the gain-cavity offset in visible VCSELs have big effects on the output characteristics. So epitaxial growth control need to be very tight.  High pressure causes a blue shift in E cav & E gmax, whilst temperature rise from Ohmic heating cause a red shift. E gmax moves around 3 times as fast as E cav. Here we have used pressure to vary the offset between the two.  The relative alignment of E cav & E gmax is evident in determining whether the threshold is smaller under pulsed or c.w. operation, and the peak output power.  Whatever the polarity the initial alignment of E cav & E gmax, leakage dominates the temperature and pressure variation of threshold (see our poster 25P13 and talk 27D1 at this conference [1,2]). This indicates the problems associated with achieving shorter wavelength emission.  Further studies using higher power devices and wider variations in cavity energy are planned.

13. Uni S T.E.Sale et al., HPSP 9, Sapporo 2000, paper 27P15. References: