• 沒有找到結果。

Chapter 4 Chirped-Multilayer Quantum Dot Laser

4.3 Results and Discussion

4.3.5 Spectral Characteristics

The main issue in this thesis focuses on whether our CMQD LD could exhibit broad spectral width and centered near 1.3 μm to meet the demand of low-coherence applications. The RT lasing spectrum with varying injection levels and cavity lengths for CMQD LD with 3 μm ridge width are shown in Fig. 4.11.

It is obvious that the spectral ripple is severe for short-cavity device in low injection current. This phenomenon can be attributed to the longitudinal mode existed in a Fabry-Perot cavity. The longitudinal mode

Fig. 4.11(a)~(h) The RT lasing spectrum operating from lasing 53

threshold to well-above threshold current for CMQD LD with 3μm

1150 1160 1170 1180 1190 1200 1210 1220

2 Ith DO 1560_Spectrum

L = 0.5mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1220 1230 1240 1250 1260 1270 1280 1290

DO 1560_Spectrum L = 0.75mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1230 1240 1250 1260 1270 1280 1290

DO 1560_Spectrum L = 1mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1220 1230 1240 1250 1260 1270 1280 1290

DO 1560_Spectrum L = 1.5mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1230 1240 1250 1260 1270 1280 1290

DO 1560_Spectrum L = 2mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1230 1240 1250 1260 1270 1280 1290

DO 1560_Spectrum L = 3mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1230 1240 1250 1260 1270 1280 1290 1300

DO 1560_Spectrum L = 4mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

1230 1240 1250 1260 1270 1280 1290 1300

DO 1560_Spectrum L = 5mm, W = 3μm 10us / 1ms pulse operation resolution = 0.1nm

separation measured in the output spectrum can be expressed by 2nL

2/ λ λ ≈

Δ , where n and L are the group effective index of active region and the cavity length, respectively. The longer the cavity length is, the closer the longitudinal mode separation will be. In other words, the longitudinal mode spacing cannot be specifically defined for devices with longer cavity length. In our case, the longitudinal mode spacing can only be defined as 0.37 nm, 0.28 nm and 0.21 nm for devices with L of 0.5 μm, 0.75 μm and 1 μm, respectively. The corresponding group effective index of 3.73 is obtained. It is believed to be a rational value in our InAs/InGaAs active region. In fact, whether the longitudinal mode is clear or not strongly depends on the resolution, an adjustable parameter when measuring power spectrum with OSA. However, due to the Fourier-transformation relation between power spectrum and interferogram, it is the narrow peak and random-like behavior of longitudinal mode in spectrum that severely affects the profile of interferogram. That is why the resolution must be mentioned when the spectral width of a broadband laser is quantitatively determined.

Table 4.2 lists the corresponding peak wavelength and FWHM as increasing the injection level for these devices. Fig. 4.11 shows the plotof peak wavelength versus current injection level for each device. The broadest spectrum equals to 14.2 nm occurs in the 3-mm LD at 10*I . th As stated before, the lasing wavelength is expected to be coming from the GS of QDL for all devices except for the one with cavity length of 0.5 mm.

Around the threshold current, a slight blue-shift occurs when cavity length decreases. It could be attributed to the moving of maximum gain spectrum curve towards higher photon energies as the injection current increases below the lasing threshold current, which is proposed by Maximov et al [45]. Due to the mirror loss is inversely proportional to the cavity length, the threshold current density of short-cavity device is

higher than that of long-cavity one. As a result, the degree of blue-shift for short-cavity diode laser will be much more apparent. This is consistent with our data.

As shown in Fig. 4.12, ES of QDL emitted by device with 0.5-mm-length was red-shifted severely after the threshold current due to thermal effect which is caused by high injection current density. However, this red-shift is compensated by the blue-shift caused by smaller dots in

Ith

I /

L 1.1 2 4 6 8 10

λp (nm) 1173.4 1173.9 1179.2 1183.9 1187.7 1193.8 0.5mm

λ

Δ (nm) 1.4 6.4 6.2 9.1 5.9 8.7

λp (nm) 1254.3 1255.7 1256.1 1257.9 1259.7 1261

0.75

mm λ

Δ (nm) 2.2 3.7 4.3 5.1 5.3 5.3

λp (nm) 1259.3 1259.5 1261.3 1262.3 1265.3 1268.3

1mm

Δ (nm)λ 2.2 2.9 4.9 4.3 4.0 4.3

λp (nm) 1259.1 1257.8 1257.1 1260.6 1260.6 1260.7

1.5mm

Δ (nm)λ 1.2 4.7 8.1 9.8 11.6 12.4

λp (nm) 1264.3 1264 1264.8 1265.1 1264.9 1265.2

2mm

Δ (nm)λ 1.5 3.9 8.4 10.2 12.5 12.8

λp (nm) 1268.3 1267.9 1268.2 1268.4 1268.3 1268.5

3mm

Δ (nm)λ 1.1 4.7 8.8 12.6 13.7 14.2

λp (nm) 1270.4 1270.4 1269.3 1270.2 1270.1 1267.9

4mm

Δ (nm)λ 1.2 3.9 6.5 9.3 12.3 14.0

λp (nm) 1273 1272.1 1271.7 1270.2 1270.6 1270.1

5mm

Δ (nm)λ 1.4 3.8 6.8 9.0 10.5 11.5

Table 4.2 The list of corresponding peak wavelength and FWHM as increasing injection levels for all devices.

Fig. 4.12 The plot of peak wavelength versus current injection level for each device.

0 2 4 6 8 10

1170 1180 1190 1260 1280

0.75 / 1.5 / 1 / 2 / 3 / 4 / 5 mm

Peak Wavelength (nm)

Current Injection ( I / Ith) 0.5mm

the devices with cavity length of 0.75 mm, 1 mm and 1.5 mm, and furthermore the blue-shift becomes dominant for longer devices. For our unpackaged and uncoated devices, the temperature-controllable heat sink plays an important role to prevent severe thermal effect. With larger contact area with the heat sink, thermal effect of long-cavity devices can be effectively suppressed.

A progressive broad bandwidth in the lasing spectrum is noticed with increased injection current because of the inhomogeneous broadening. Larger dots with lower energy would be filled at first as increasing current injection. A blue-shift in gain spectrum caused by filled smaller dots with higher energy occurred at high pumping. However, the spectral width is not as broad as expected due to non-uniform enhancement of gain spectrum as increasing pumping. Furthermore, the effect of 4*QDM and 3*QDS in the structure of CMQD are not apparent from our measurement at injection level up to ten times of threshold

Fig. 4.13 The evolution of spectrum at high injection level for CMQD.

1200 1220 1240 1260 1280 1300

-10 -8 -6 -4 -2 0

2.0 A 2.5 A 3.0 A

Normalized Intensity (dBm)

Wavelength (nm) (W,L)=(3μm,3mm)

(τ,T)=(5μs,1ms) Pulsed @ 293K resolution = 0.1nm

current. It means that the carrier population cannot be well-distributed and results in a non-equal contribution of threshold modal gain. On the other hand, Fig. 4.13 shows the measured emission spectrum at much higher injection level under pulse operation from a 3-mm-long device of CMQD LD. Table 4.3 lists the spectral dependence on current injection.

The spectral FWHM decreased at current level from 1.5 A to 2.3 A is due to the dip at 1270 nm. Simultaneous lasing from GS of QDL and a higher state occurred wherein incomplete gain clamping and retarded carrier relaxation process are the main attribution [35]. It is believed that this higher state was contributed by both the ES of QDL and GS of QDM. Besides, an extremely broad spectrum of 29 nm centered at 1270 nm was observed. It is worth noting that these measurements are all under a fine resolution of 0.1 nm. However, the energy spacing between GS and ES is too large ( > 60 meV for S-K type QDs generally [46]) to joint these two spectra to be a really broad one. To overcome these problems, it is advised to optimize the structure of active region in CMQD, such as capping thickness, number of stacking layers and growth condition for

I (A) 0.3 0.5 0.8 1 1.3 1.5 1.8 2 2.3 2.5 2.8 3 FWHM

(nm) 12.0 16.6 18.7 19.6 20.8 29.0 6.0 24.7 17.8 27.7 28.0 28.3

InAs QDs. By the way, simultaneous two GS lasing for CMQD with different structure at active region can be observed [47], which implies the potential for CMQD to be a suitable broadband light source.

相關文件