Cryostat Beamsplitters
4 Experimental results .1 Thermal tuning
4.4 Time-resolved measurements
Figure 36(a) shows the spectrum acquired for different FP heating powers. As predicted the uncoupled FP cavity shifts about 0.3nm. This shift causes one of the FP modes to shift to the same position of the Noda mode causing them to go in resonance. The barrier between the Noda and the FP cavity was 6 periods and the barrier at the end of the FP was 10 periods. However the Q-factor measured for the uncoupled FP mode was around 4000, similar to the previous cases. As can be seen from Figure 36(b) this causes the Q-factor of the Noda mode to drop, while the resonant wavelength of the Noda mode does not change. Apart from that, the maximum intensity of the Noda cavity also drops, which could be an indication of a different Purcell factor however the change is much larger than can only be explained by a difference in Purcell factor. Both are evidence of weak coupling of the FP mode and the Noda mode. The change in Q-factor is about 20%, however the FP is not brought completely in and out of resonance. Therefore the maximum Q-factor of the Noda cavity could be much higher. The minimum Q-factor is already close to halfway the Q-factor of the FP cavity, indicating that the FP is indeed brought into resonance completely.
Figure 36: Results of the two beam experiment on InGaAsP where an FP mode is brought into resonance with the Noda mode. (a) Spectrum at different powers, showing the FP mode shift into the Noda mode for higher pump powers. The orange lines indicate the shift of the peaks of the FP modes. (b) Q-factor and intensity of the Noda mode as a function of the FP resonant wavelength, showing the drop in Q-factor when the FP mode shifts into resonance with the Noda mode.
This graph also shows a drop in intensity as the FP is brought into resonance. This could be an indication for a change in Purcell effect.
4.4 Time-resolved measurements
To see if the change in cavity lifetime would also change the light-matter interactions the time-resolved emission was measured. Since it was not possible to see single dot emission from InP samples these measurements were performed on an ensemble of dots at 70K. The temperature was lowered to ensure that the linewidth of the quantum dots was smaller than the linewidth of the mode.
To check for enhancement or inhabitation of emission, first the reference decay has to be measured for quantum dots outside any photonic crystal. The quantum dots measured are emitting at 1540nm, which was also the regime where the cavity resonances are at low power. The result of this measurement is shown in Figure 37. The decay time obtained from fitting this curve is 2.36ns.
1562 1564 1566 1568 1570
45
For the time-resolved measurements a high-power pulsed laser was used. This laser was attenuated using filters. If the power of the laser was too high a clear blue shift behaviour due to the carrier injection was observed. This is shown in Figure 38 where a colourmap of the decay versus wavelength is drawn for both high and low power excitation measured in a single Noda cavity. The long tail on the blue side for the high power measurements is caused by the cavity blueshifting several nanometer and then slowly returning to the relaxed position when the carriers relax again.
0 2 4 6 8 10 12 14 16
1 10
Counts (-)
Time (ns)
Figure 37: Decay curve of InAs quantum dots in the InGaAsP sample. The decay time obtained from this curve using the shown fit is 2.36ns
Figure 38: Colourmaps of the decay for different wavelength for single Noda cavities. (a) Decays taken at high excitation power, causing the mode to blue shift as indicated by the long tail at the lower wavelengths. (b) Decays taken at lower excitation power, the blue shift has almost disappeared here.
To be able to get the signal from the entire cavity, the slit at the end of the monochromator was opened 10 times as much as previously, reducing the resolution by a factor 10, causing the emission
1545 1546 1547 1548
46 with a 0.8nm linewidth to enter the detector. Since this is much higher than the linewidth of the mode it is expected that the entire mode will be collected in this way. From the decay-curve acquired in this way (shown in Figure 39) a lifetime of 0.48ns has been determined. This would correspond to a Purcell enhancement in the Noda cavity of Fp=5.
0 2 4 6 8 10 12 14 16
Counts (a.u.)
Time (ns)
Data Fit
Figure 39: Decay curve and fit for the single Noda cavity. The fit takes deconvolution into account but not the slow rise time due to the small blue shift. The noise in the fit is caused by the noise in the IRF. The small discrepancy in the fitting and the real data is believed to be caused by the effect of the blue shift.
Finally the decay curve of an uncoupled FP mode was also measured (Figure 40). This was done for reference purposes. The decay was found to be almost identical to that of the bulk quantum dot with a decay rate of 2.34ns. This is because of the large volume of the FP causes a small Purcell effect. The spectral position of the FP cavity could however still be distinguished from the increased background when the CW laser was pumping the FP cavity.
0 2 4 6 8 10 12 14 16
Counts (a.u.)
Time (ns)
Figure 40: Decay curve of an uncoupled FP mode with the fit. The decay rate found was found to be almost the same as the decay rate for bulk quantum dots.
47
Next the decay of a combined structure was measured as function of wavelength. The combined structure had a similar design as the previous measurements, since these appeared to be close to the ideal coupling regime. However this time the Noda and the FP mode had to overlap at liquid nitrogen temperature. Therefore a different cavity with similar design was used. The measurements were first performed with high excitation power. Therefore the cavity still blueshifts, causing the fast decay at the lower wavelengths. The results of these measurements are shown in Figure 41.
0 500 1000 1500 2000
Counts (-)
FP heating 0W FP heating 729W FP heating 1175 W
1544 1545 1546 1547
0 1 2 3 4 5
Decay time (ns)
Wavelength (nm)
Figure 41: Results of time-resolved measurement on a combined structure for different FP heating powers. (a) shows the integrated counts measured during one decay curve and is thus a measure for the intensity at that wavelength. (b) shows the decay time at that particular wavelength. The black line indicates the resonant wavelength of the Noda cavity while the red line indicates the background decay time.
From Figure 41(a) it can clearly be seen that for higher power a second peak is red shifting out of the position of the Noda mode. When this mode is brought out of resonance, the decay time drops from near the background level to less than 1ns. This is a change in decay time of 61%, caused by bringing the FP mode out of resonance with the Noda mode. Apart from that on the blue side a large region with low counts but very fast decay is observed. This is similar to the tail in Figure 38 and is caused by the blue shifting of the cavity due to the high power of the pulsed laser. This can also be seen from Figure 42 where a 2D plot of the counts vs. time for different wavelengths for this measurement is plotted. Apart from that an increase in decay time is observed at the red side of the Noda mode. This is also observed for the single Noda cavity measurements and is not yet well understood but is also believed to be related to the blue shifting of the cavity mode.
Figure 42 also clearly shows the emerging of a second mode, which decays slower and the increase in decay rate for the mode at the Noda frequency (indicated by the red line). This is a clear evidence of the change in lifetime obtained.
48
To prevent the influences of the blue shift, the power of the laser was reduced and the same structure was measured again. However, due to the lower excitation power, the Noda cavity was heated less and as a result the resonance wavelength of the mode decreased. This brought the Noda mode in resonance with a different FP mode. The result of these measurements is shown in Figure 43 where the 2D plot of the decay curve for different wavelengths is plotted.
Figure 43: 2D plots of the decay curve for different wavelengths as measured with low excitation power. (a) measured with low FP heating. (b) Measured with high FP heating, showing that the FP mode with the characteristic high background appears at higher wavelength. The red line shows the wavelength of highest integrated intensity, which is believed to be the wavelength of the Noda mode
1545 1546 1547
Figure 42: 2D-plots of the decay curves for every wavelength as obtained by scanning the wavelengths and then interpolating all the decay curves obtained. (a) decay curves for no FP heating, where the FP and the Noda cavity are in resonance and one peak is observed. (b) decay curve for maximum FP heating, where a second peak is red shifted away from the wavelength of the Noda mode (indicated by the red line) and the decay rate of the Noda mode increases dramatically.
(a) (b)
49
The change in decay curves is made clearer in Figure 44, where the decay curve for all the FP heating powers at the indicated wavelength is plotted. In these graphs a decrease in both rise time and decay time is observed. The decrease in decay time can be explained from the increased Purcell effect when the modes are uncoupled. The increased rise time cannot be explained that easily and more investigation is needed to fully understand the phenomena taking place here.
0 2 4 6 8 10 12 14 16
0,0 0,2 0,4 0,6 0,8 1,0
Normalized Counts (a.u.)
Time (ns)
FP Heating 0W FP Heating 222W FP Heating 729W FP Heating 1175W
Figure 44: Decay curves at the wavelength corresponding to the red line in Figure 43, for different FP pumping powers. A decrease in rise time and decay time is observed when the FP is shifted out of resonance.
Figure 43 shows the FP mode shifting to higher wavelengths from the Noda mode for high FP heating power. The FP is characterized by the high background caused by the CW laser pumping it.
The decay rate of the Noda is also seen to increase significantly. This can also be seen in Figure 45 where the decay time for different wavelengths is plotted. This also shows that the large tail from the blue shifting has disappeared but still the effect of the FP tuning is visible. The decay time of the Noda quantum dots in the Noda cavity is changed by a factor of 2 for high FP heating powers. The fact that the Noda decay time is not yet the same as that of the uncoupled cavity seems to suggest that the FP mode is still somewhat coupled. This can also be seen from the spectrum where the two peaks are not completely separated yet. Because of this a change in Q-factor smaller than 50% is expected since this change is only obtained when tuning over the entire range.
50
1544,5 1545,0 1545,5 1546,0 1546,5
0 1 2 3
Decay time (ns)
Wavelength (nm)
0 500 1000 1500
Intensity (a.u.)
FP Heating 0W FP Heating 222W FP Heating 729W FP Heating 1175W
Figure 45: Decay rate and integrated intensity versus wavelength for different FP heating powers for the coupled system. The red line indicates the decay rate of the uncoupled quantum dots, while the black line indicates the wavelength of the Noda mode. For higher powers a second peak is seen to emerge at higher wavelengths than that of the Noda mode. This is accompanied by a decrease in decay time.
Due to the blue shifting and the complexity of the coupled modes the time-resolved measurements have not been completely understood. However a clear trend of decreasing decay time for decreased coupling is visible. The decrease in decay time seems to be larger than the decrease in Q-factor, suggesting that the Purcell effect is changing more. Since the Purcell factor is proportional to Q/V this would suggest a change in mode volume when the two cavities are coupled. However, more investigation is necessary to validate this hypothesis.
51