Sistema de control basado en lógica borrosa

To determine whether the LIFE features were, or remained, charged following exposure to hydrofluoric acid, the samples were examined with an atomic force

microscope (AFM) which had been modified to detect surface charge as well as topography. This technique has been successfully applied to ferroelectrics (Luthi et al., 1993) and lithium niobate in particular; Tsunekawa et al. (1999) have reported the use of electrostatic force microscopy (EFM) to observe microdomains. They found that the−_{zcrystal face has a positive charge and the}+z_{microdomains on it}

were negatively charged in air. Soergel et al. (1998) have also used EFM to study charge distributions on Bi12SiO20(BSO) crystal surfaces.

The experiments discussed here were carried out using equipment at the University of Bonn, Germany, with the assistance of Prof. Dr. Karsten Buse, Dr. Elisabeth Soergel and Nils Benter. Electrostatic force microscopy measures the force between surface charges and a periodic charge induced on the conductive (Al-coated) AFM tip, by a voltageV0cosω2t. A lock-in amplifier was used to measure the electrostatic

force; the phase signal from the lock-in also allowed discrimination between areas of positive and negative charge.

To detect charges the AFM has to be operated in non-contact mode. The amplitude of the voltage used was 10 V, with a frequency of 2 kHz. To reduce the effect of vibrations, the AFM was mounted on a floating table, and was covered with a shield. Nitrogen gas was used to purge the air surrounding the tip and sample, to reduce the water content which was known to affect charge detection.

The best results were obtained when an initial AFM scan was performed in contact mode (non-charge detecting), to remove any surplus surface charges generated by the pyroelectric and photorefractive effects. The pyroelectric effect causes charges to accumulate at the crystal faces, which compensate for the change in spontaneous polarisation that occurs with change in temperature. This results in the+zface to become more positively charged on cooling. After the initial contact mode scan, the measurement was repeated in non-contact mode to detect the static charge distribution.

Two periodically poled lithium niobate samples were used to test the operation of the instrument. The first (Fig. 3.23) had been etched before examination by AFM, so had topographic features, while the second (Fig. 3.24) was unetched so had no topography. In both sets of results charge was detected and in the case of Fig. 3.23 the charge image matched that of the topography. The resolution of the charge image from the unetched PPLN does not appear to be as good as that from the etched sample, but the scan is of a smaller size (20µm compared to 40µm), and it does show that charge can be detected independent of topography.

(a) Topography (b) Charge

Figure 3.23: Topographic and charge images of an etched PPLN sample recorded by AFM. Area shown is 40µm by 40µm.

(a) Topography (b) Charge

Figure 3.24: Topographic and charge images of an unetched PPLN sample recorded by AFM. Area shown is 20µm by 20µm.

After the operation of the AFM had been confirmed in topographic and charge- detecting modes, a 0.1 % Fe:LiNbO3 LIFE sample was examined. A good quality

topography image was obtained, as shown in Figure 3.25. However, a charge image could not be obtained, due to the unusually high number of surface charges which caused the AFM tip to be strongly attracted to the surface. The attraction was such that the AFM could not even operate in non-contact mode, even after several scans in contact-mode which should have removed the surface charges. The excess charge was thought to be a consequence of the relatively high iron-dopant concentration. A sample with 0.01 % iron dopant concentration gave better results, as shown in Figure 3.26. The topographic image shows a single etch-frustrated feature, which

Figure 3.25:AFM topographic image of a 0.1% Fe:LiNbO3LIFE sample. Area shown is 20µm by 20µm.

is higher than the surrounding areas. The stucture is considerably larger than those shown in Fig. 3.25, which is to be expected due to the lower iron dopant concentration (as discussed in Section 3.3.3). The charge image strongly resembles an inverted topography image; the feature is darker than the surrounding areas, which shows it is more postively charged than the surroundings.

(a) Topography (b) Charge

Figure 3.26: AFM topography and charge images of a feature on a 0.01 % Fe:LiNbO3LIFE sample. Area shown is 15µm by 15µm.

A small area of the crystal around the LIFE site was then repoled to become a+z

domain. The area of domain inversion is visible in the polarised optical microscope picture shown in Figure 3.27, and includes the LIFE structures, visible at the top of the image. The crystal surface appears quite rough as it had been etched during the LIFE process.

LIFE site

repoled area unpoled area

Figure 3.27: Polarised optical microscope image showing the repoled LIFE site and surrounding area.

Following repoling, further AFM scans were taken of the LIFE structures. The results in Figure 3.28 show that while the topography is unchanged, the charge of the feature is almost indistinguishable from the surrounding (+z) area.

(a) Topography (b) Charge

Figure 3.28: AFM topography and charge image of arepoledLIFE site. Area shown is 20µm by 20µm.

The charge image also reveals that the surroundings are darker than those in Fig. 3.26, which confirms that domain inversion has taken place and that both the feature and surrounding crystal surface are positively charged.

The results are consistent with the etching behaviour that was observed. As discussed previously, the−_{zface of lithium niobate is susceptible to etching by HF}

acid, while the+zface is known to resist etching. The opposite sign of the charge of the etch frustrated feature compared to the normally etched−_{zsurroundings is}

hence likely to be the cause of the etch-frustration. Further evidence of this was obtained by poling the entire region, after which the charge of the new+zdomain surrounding area changed sign to become the same as that of the etch frustrated feature. The actual polarity of the unmodified−_{zsurface was found to be negative,}

while the +z surface was positive. This is contrary to the results of Tsunekawa et al. (1999), who reported that−_{zplanes have a positive charge. However, lithium}

niobate surfaces are known to attract compensating charges (Weis and Gaylord, 1985) due to the dipoles which exist in the bulk material, so it is possible that Tsunekawa et al. were detecting these, while in the present experiments surface charges were removed by the contact-mode scan that was carried out before the non-contact, charge-detecting scan. However, it is the difference in charge between the etch-resistant features and the surrounding −_{z surface that is of particular}

importance.

It was not possible to obtain charge images for 0.1% iron dopant concentration samples, however future experiments could be conducted using a different gas surrounding the sample (instead of N2), which may help to reduce the surface

charges. Mirza et al. (1978) studied light emission from the surface of lithium niobate as a crystal was heated and cooled, and found that the luminescence was primarily a surface controlled effect. Indeed, while strong emission was observed in a vacuum, the presence of CO2 or O2 almost totally quenched the light. The

authors believe that light is seen on heating as the pyroelectric effect creates a large electric field and subsequently electrical breakdown may occur, during which some energy is liberated as luminescence.