show, how the sample quality reflects in the results of ferroelectric measurements. By comparing the results we have concluded that quality of our sample, in terms of dielectric behavior is similar to the sample used by Schrettle for their measurements.
Figure 5.5: (a) Temperature dependent of the heat capacity of magnetite with TV = 122.8 K (b) 0(T) of magnetite measured at different frequencies and plotted together with the reported data, taken from ref [97], which confirmed high quality of the sample.
5.3 Experiment
After confirming the quality of the crystal a thin slab from the sample was cut along the (001) direction, polished down to a thickness of∼150µm using sand paper and sputtered with gold of 100 nm thickness on both sides, by masking the edges of the crystal to avoid any short circuit during the experiment. The choice of gold over silver electrode was because gold has excellent corrosion resistance. But silver can tarnish very easily when exposed to the atmosphere which can be an effective insulator [151]. Moreover the work function of gold is higher than silver. Time resolved experiment was performed at the hard x-ray beam line P09, PETRA III. The experimental set up is shown in the figure 5.6.
We began our experiment by mounting the crystal on a alumina sample holder for electrical insulation. Although alumina is a good thermal conductor, during the experiment we observed that the sample temperature was much higher than we expected.
With the available standard closed-cycle cryostat one could reach the base temperature of
∼4 K, but during our measurement the sample thermometer was displaying ∼7−8 K
Figure 5.6: (a) Magnetite crystal sputtered with gold and mounted on the copper sample holder.
(b) Flow chart of the novel data acquisition system, taken from reference [150]. High voltage (HV) electronics (left upper corner) generates the HV along with the digital pulses synchronized with the change in the HV polarities. The analog HV is directly delivered to the crystal. The digital signals, the current movement status of the diffractometer axes and the signals from the scintillating counter are processed by the field programmable gate array board(FGPA). (c) one of the many shapes of voltage pulses tried during the experiment( green color puses are voltage and yellow color is current).
without the electric field and more than 10 K with the electric field. Along with the sample holder, there was also contribution from copper wiring made inside the chamber to apply the electric field however, the effect of which was unavoidable. Considering the results of macroscopic polarization [figure 5.3], attaining as low temperature as possible was very crucial for the experiment to observe the intrinsic ferroelectric behavior. Hence, to overcome with the temperature issue, eventually we mounted the sample directly on the cold finger and made the electrical connections as shown in the figure 5.6 (a). One cable was directly connected on the top of the crystal and the other cable was grounded.
And these two cables from the sample were connected to the periodic high voltage supply.
The flow chart of the novel experimental setup developed by our collaborators at the university of Siegen at a different beamline for studies of piezoelectricity is shown in Fig 5.6. The crystal is subjected to x-rays and at the same time the periodic high voltage
5.3 Experiment
is applied to the crystal. The diffracted photons coming out of the point detector were redistributed between 10000 time channels synchronized with the above state of the applied electric field. In this way each time channel probes the state of a crystal at a specific time delay after the beginning of a high voltage cycle. After adapting the setup to the P09, we had the time-resolution of at least several 100 Hz that is necessary for switching to occur, according to the macroscopic measurement (figures 5.3 and 5.4)
The experiment was challenging with a lot of technical issues. One of the necessary condition for the measurement was the presence of low temperature. As can be seen in figure 5.3 only the curve at 5.6 K shows indications of saturation at high field that is expected for intrinsic polarization switching [figure 5.3]. However, the presence residual conductivity at low temperature phase was hindering us to apply long voltage pulses by developing the heat load on the sample. Not only the voltage, but also the effect of beam heating, inductive current component and the switching current effecting the system to stay at low temperature. As the temperature increases sample becomes more conducting and leaky, as a result the extra charges from the sample coming back to the circuit and high voltage supply use to trip off. After several attempts we could optimize the system with the long dead time between the two voltage pulses, as shown in the figure 5.6(c).
Apart from these technical issues another challenge was to choose a right reflection for the measurement. Though our preliminary structure factor calculation gave significant contrast between the Friedel pairs, not all the reflections could be reached at the Fe K-edge because of the rotation limit of the diffractometer. The rotation was severely restricted by the presence of the cryostat. The calculations showed that the relative difference between the intensities were high primarily for weak reflections. Considering the time constrains, due to the limited period of beam time, it was rather difficult to choose very weak reflection because it takes relatively longer time to collect the data with sufficient good statistics. With the above mentioned issues, and after testing the several structural reflections finally we have chosen the reflection (2 -2 10) for the continuation of the measurement.
As we had problem in attaining the low temperature (below 5 K) using a Displex closed cycle cryostat in the first beam time, to resolve the temperature issue we have used a cryostat with a lower base temperature of 1.7 K employing Joule-Thomson cooling in the second beam time. However the decision to use this cryostat turned out to be a mistake, because the thermal decoupling associated with the Joule-Thomson stage implied that the heat-load from the resistive heating of the sample could not be overcome, and
consequently the reachable temperatures were even higher than in the first beam time.