Wavelength Dependence

The powder shaped up-converters were bound in a transparent binder to solid samples and attached to the rear of a bifacial silicon solar cell using a refractive oil to enhance the optical contact. In this way it was possible to attach the different up-converters in all measurements to the same solar cell⁵. The measurements were performed using a setup as described in Appendix A.1. A comparison of the spectra of all BaCl₂ based up-converter samples is shown in Figure 4.3. The peak external quantum efficiency ranges from 0.2×10⁻⁵ to 35.6×10⁻⁵% for the different samples and the details are listed in Table 4.2 and given in terms of a scaling factor for each spectrum in Figure 4.3. In addition to the maximum efficiency, the integrated efficiencies also provide information about the width of the excitation range and are therefore a better measure for the performance of the up-converter. The integrated efficiencies are listed in Table 4.2.

J5 shows the lowest efficiency and could, therefore, be a base reference. Unfortunately the photoluminescence signal from J5 was not detectable. Therefore, for a better comparison all relative numbers are given in relation to J11, for which photoluminescence measure-ments were successful. An illustration of the figures given in Table 4.2 is given in Figure 4.4, where the integrated and the peak external quantum efficiency is shown relative to J11 and as a function of the erbium content.

As can be seen in Figure 4.4 the samples with erbium contents close to the predicted optimum of 28 mol% (J4 and J10) show the highest efficiencies. Due to the unexpectedly low response of J5 and J7 the dependency of the up-conversion efficiency on the erbium concentration as expected from previous reported experiments is not clearly evident (see Section 2.1.5).

An obvious explanation of the lower response of these two samples could be the

uncer-5In the infrared spectral response measurements the cell 7ac, processed as described in Section 3.1, was used. An exception is the measurement of up-converter sample J5, which was attached to cell 7ab.

Both cells differ in efficiency by about 2% absolute, what equals 15% relative. As explained later the response found for J5 is about 2 orders of magnitude lower than expected. So this can not be explained by the different solar cells.

1530 1540 1550 1530 1540 1550

J11 (x5)

J10 (x1)

J7 (x2)

J5 (x145) J4 (x1)

J3 (x4)

J2 (x4)

J1 (x2)

EQE [a.u.]

Wavelength [nm]

Figure 4.3: Spectral response under excitation in the infrared for all barium chloride based up-converters. The absolute efficiencies differ between the samples and the scaling factor is indicated for each spectrum. Also the shape of the spectra shows differences. Two groups of spectra can be found, assigned as SR1 (J1, J3 and J11) and SR2 (J2, J4 and J5). J10 shows features of both kinds of spectra.

tainty of the quality of the optical contact between the up-converter and the solar cell in the measurement setup. Therefore, as discussed in Appendix A.1, the numbers extracted from these measurements give the lower limit of the possible efficiency. This explanation is unlikely, however, since, as will be shown in Section 4.4, the same relation of emission intensityη_J7 < η_J4 < η_J10 is found in photoluminescence measurements, while the pho-toluminescence signal from J5 was not detectable at all. Therefore this low response of J5 and J7 is more likely a sample property than an artefact of experimental uncertainties.

Beside differences in efficiency also differences in the shapes of the spectra are found.

Two groups (assigned as SR1 and SR2) can be identified, the grouping is given in the

Sample Peak EQE Integrated EQE Peak EQE Group

J7 14.5 5.19 2.39 not clear

J10 35.6 10.71 5.86 not clear

J11 6.1 1 1 SR1

Table 4.2: Results of the infrared spectral response measurements. In the first column the maximum external quantum efficiencies are listed, in the other columns the results are given relative to sample J11. The samples are affiliated into two groups (SR1 and SR2) concerning the shape of the excitation spectra as shown in the last column.

last column of Table 4.2. Group SR1 shows nearly no response below 1525 nm, but a rising response between 1525 and 1533 nm. Three significant peaks occur at 1545, 1548.5 and 1555.5 nm. Contrary to group SR1, group SR2 exhibits a broad peak between 1515 and 1525 nm, less efficiency around 1530 nm, a double peak around 1540 nm and only a rudimentarily developed peak at 1555.5 nm, while the peaks at 1545 and 1548.5 nm seem to be absent.A special case is sample J10, which does not match either group. In fact

10 15 20 25 30 35 40

Figure 4.4: Integrated and peak external quantum efficiency relative to results of the sample J11, which showed the lowest response, dependent on the erbium concentration.

it contains features of both groups of spectra and an additional peak between 1558 and 1565 nm.

Different excitation spectra can be caused by a different degree of Stark splitting. As explained in Section 2.1, the exact location of the energy levels and therefore the dis-tances between certain energy levels are influenced by the electrical field caused by the surrounding ions of the host material. A change in crystal symmetry around the erbium ion therefore might change the distance between energy levels, which is reflected in the excitation wavelengths. The crystal structure of the samples is investigated in Section 4.6 and it will be shown, that the samples assigned to group SR2 show features in the X-ray spectrum, which do not occur in samples assigned to group SR1. Therefore it is very likely that the different shapes of the excitation spectra are caused by different Stark splitting.