• PDF / 359,144 Bytes
• 6 Pages / 414.72 x 648 pts Page_size
• 19 Downloads / 287 Views

DOWNLOAD

REPORT

---

PHOTOLUMINESCENCE EXCITATION In general for PLE measurements one has to ensure to work with a homogeneously light emitting sample due to the decreasing penetration depth of light with increasing excitation energy. For our samples a quite uniform PL emission is observed from a cleaved edge as well for aged as for RTO material [8]. In addition we observe a narrowing and a redshift of the luminescence line with increasing penetration depth of the light (see Fig. 1). This is typical for the selective excitation of larger particles within a size distribution. The contrary is expected for an inhomogeneous sample since in this case a low energy excitation averages over a larger size distribution.

104 3

10 Eo

102

101 1.0

1.2

a. 0.5

0.8

1.0 r

0.6 0.0

1.5

2.0

2.5

0.4

"

0.2

i"a.

-1 0.0 3.0

Energy (eV) Figure 1: Upper part: Comparison of absorption coefficients obtained by PLE and PDS (data taken from Ref. [5]). Lower part: PL for two excitation energies and square root of PLE intensity times photon energy for different detection energies. 1.59 eV represents the maximum of the PL-line, the other two energies are below and above the maximum respectively. The lines indicate the extrapolation to the bandgap.

448

For PLE measurements in high excitation energies a saturation effect is observed due to the fact that the sample becomes optically thick. This saturation together with the sample thickness can be used to evaluate the absorption coefficient. Fig. 1 shows in its upper part the absorption coefficient obtained this way with PLE detection on the PL maximum. In agreement with literature [9, 10] we find that the PLE-absorption is considerably below the values obtained by other methods like absorption measurements or photothermal deflection spectroscopy (PDS). PDS data shown for reference in Fig. 1 are measured with the same type of sample [5]. In the low energy region the PLE absorption does not represent a straight line on the log-scale of the figure. In contrast good linearity is obtained by plotting the data as square root of the PLE intensity (being proportional to absorption in the optically thin limit) times photon energy. This is the behaviour expected for an indirect semiconductor and allows to determine a bandgap E.. This gap lies about 0.2 eV above the detection energy for detection on the PL peak. Due to the size distribution of crystallites in PS higher extrapolated gaps are obtained for higher detection energies (Fig. 1), however the energy difference between extrapolated gap and detection energy decreases for increasing detection energy. In a quantum confined system as well as for surface states it is in contrast expected that the distance between the lowest light emitting state and a region with high density of states leading to the extrapolation of the gap should rather increase than decrease with confinement energy. Regarding this distance in a first approximation as constant within one PL line we have to conclude that a quite large linewidth of the luminescence corresponding to one bandgap