Boron-Oxygen defects

During the iron measurements we noticed a lifetime degradation after light soaking. This could be due to formation of boron oxygen defect centers. This formation is not reversible at room temperature, i.e lifetime remains low or gets even lower after additional illumination. Schmidt and Bothe [SB04] have investigated the generation and annihilation kinetics of these defects.

They found, that defect generation as well as annihilation rate follow the law of Arrhenius [Mor96, p. 222], which is given as

k=Aexp

k is the reaction rate, A a constant called frequency factor,R the universal gas constant, k_B the Boltzmann constant,T the absolute temperature andEa,mol andEa the activation energy per mole and per atom respectively.

50 100 150 200 250 Temperature in °C

10

10

10

10

10

10

rate in 1/s

defect generation defect annihilation

Figure 6.11: Effective generation or annihilation of boron- and oxygen-related defect centers. The critical temperature is 170^◦C.

The defect generation is characterized by an activation energy per atom of Ea,gen = 0.4 eV and the defect annihilation by an activation energy of Ea,ann= 1.3 eV. Unfortunately, they do not state the frequency factors, so that we have to take the data from their plots and perform the fits ourselves again. The frequency factor for defect generation isAgen = 1.84·10²s⁻¹ and for defect annihilation Aann = 2.29·10¹²s⁻¹ and the activation energies coincide. With this knowledge, and the assumption, that the frequency factors are material independent, a critical temperature can be determined, above which defect annealing prevails defect generation. For this purpose, we equalize the defect generation and annihilation rates, substitute by Arrhenius’

equation, solve for the temperature and get

Tcrit = E_a,ann−E_a,gen

kBln (Aann/Agen) ≈170^◦C. (6.14) To get rid of the boron oxygen defect centers, the ingot must be heated to more than 170^◦C.

A plot of the effective annihilation and generation rates is shown in Figure 6.11.

The first experiment was carried out with CR15 and a heating tape as shown in Figure6.12a.

The controller was set to a temperature of 180^◦C, but the temperature sensor, which was fixed between ingot and heating tape, showed only a maximum temperature of 140^◦C. After three hours, the heating was stopped and the ingot cooled down to room temperature over night.

Four positions were chosen, two below the tape on opposite sides and two about 9 cm away in axial direction. The results are shown in Figure 6.12b.

The lifetimes below the heating tape have significantly increased, while the lifetimes in the other part remained the same. This shows, that there was some kind of annealing. We are not sure about the actual temperature just below the tape. The temperature sensor of the controller showed up to 140^◦C, but the infrared thermometer showed even more in some parts of the exterior side of the heating tape. The temperature at positions 3 and 4, was below 70^◦C for the temperature sensor as well as for the infrared thermometer. From this experiment we cannot conclude, that the calculation of the critical temperature is correct, but the temperature was sufficient to obtain some annealing effects.

6.2 p-Type Czochralski ingots

(a)Experimental setup for heating of CR15. The num-bers mark the measurement positions

0 1 2 3 4 5

position 0

1 2 3 4 5 6

τ in µs /ρ inΩ cm

τpre

τ_post ρ_pre ρpost

(b) Lifetime and resistivity before and after heating

Figure 6.12: Experimental setup and results of the heating of CR15

As the temperatures that could be reached with the technique above are not satisfactory, a second experiment with an CR6, an UFO which fits into an oven was carried out. The ingot was treated in the same way as above. First it was illuminated, then heated for three hours with the heat tape and finally placed in the oven. The oven reaches a temperature of 220^◦C after 60 min of heating. This temperature was held for approximate three hours. Then oven and ingot cooled down to room temperature over night. To avoid thermal shocks, the oven was kept closed until room temperature was reached.

The lifetimes were measured in QSS mode at 15 different positions. On the body surface we did two series of three measurements with the long side of the coil perpendicular to the growth axis and two parallel. Additionally three measurements have been taken at the endcone and four at the shoulder. The results are shown in Figure 6.13.

The evaluation of the lifetime changes must be done with care, because as we have seen in section6.1.7, the lifetime is heavily influenced by the position of the instrument. We suppose, that changes in resistivity are mainly caused by slightly different positions. Therefore we only consider measurements as comparable, where the resistivity change is less than 20 %. Mea-surements, which are not comparable according to this definition, are marked with hatching in Figures6.13a and 6.13b.

The illumination treatment does not seem to cause significant variation of the lifetimes.

The lifetime changes can also be explained by the change of the resistivity. If any, one could obtain a slight tendency for a lifetime increase after illumination, because the lifetime ratios are in most cases greater than the resistance ratios. Effects of iron boron pair separation can be excluded, because the ingot was not treated during two days. At the moment, we cannot explain this behaviour. Perhaps it is only an artifact of our measurements.

For the heating tape treatment, the data is even worse. One could either compare the lifetimes to the initial lifetimes or to the lifetimes after illumination, but in both cases the difference could again be explained by the change of resistivity. This time, we do not see any annealing effect. Maybe it is hidden in the scattering of the data.

For the thermal treatment in the oven significant larger lifetimes are obtained. One can

1 2-1 2-2 2-3 3 4-1 4-2 4-3

not comparable to τpre

not comparable to precursor

not comparable to τpre

not comparable to precursor

(b) Lifetimes on endcone (5-X) and shoulder (6-X)

1 2-1 2-2 2-3 3 4-1 4-2 4-3

(d) Resistivities on endcone (5-X) and shoulder (6-X) Figure 6.13: Lifetime and resistivity after the different treatments.

also note, that the resistivities coincide very well with the ones taken before any treatment.

The best enhancement was noted at the endcone, where lifetime rose from below 5µs to above 20µs. The average improvement for all the comparable measurements is 8.5µs. Hence we can consider the annealing process successful. It remains the question, how many of the boron oxygen pairs have been separated.