Measuring SPM Piezo Displacement Responses Heinz Sturm*, Markus Heyde** and Klaus Radernann** * Federal institute of Materials Research (BAM), Berlin ** Institute of Physical Chemistry, Humboldt University, Berlin Introduction Scanning Probe Microscopy (SPU) is a versatile tool for the investiga- tions of surfaces. Additionally to the three-dimensional examination of sur- face topography, local mechanical or electrical properties can be meas- ured 1 . In this work we present a short contribution on the performance and reliability of piezoelectric transducers in a SPM, which take care of the fine positioning of the tip relative to the sample. Some of the calibration proce- dures known in literature [cited in 2] are; laser interferometry, scanning a slightly tilted surface, a grid, a crystallographic or artificial step of a known height or by using a reference piezo. To measure slow and fast piezo re- sponse simultaneously, we use a non-contact calibration procedure with a high dynamic range (Angstrom to several hundred micrometers) and high frequency range (D.C. to 200 kHz(-3dB)). Hysteresis of the piezo displace- ment can reach up to 20% of total elongation. So, for the z piezo which has to follow ascending and descending topographical steps and the sample tilt, hysteresis should be regarded as the most important source of errors. Fur- thermore, the displacement depends on the voltage pre-history, so often SPM data are collected in one lateral direction to minimize the divergence between the lateral commanded position and the true position, After all, the piezo response may suffer from creep and aging with time or temperature. The aim of this work is to measure the static and dynamic movement of a z piezo, The collected data can be used to characterize separately the non-linearity and hysteresis for slow and fast z-displacements as well, Addi- tionally, the problematic situation of a sample tilt (requiring slow but large z- displacement) superimposed by small topographic features (requiring fast but small z-displacement) is analyzed, time (mini 80 10O 100 » 0 a) •b) f c) d) w i—' I , ' ' _^ 1 (W) t f t 6 al o & 4 ; 80 j £ • I •a J" 3 T 1 4,55 f S 20 40 60 time (mil'] 80 100 Figure 1: a) Stepped piezo voltage excitation superimposed with 1V at 1 kHz (TRI piezo), b) Absolute piezo displacement, c) Relative piezo displacement, d) dynamic response at 1 kHz. Experiments and Results In the further experiments discussed in this article, the z-direction of a tri- pod scanner (TRI) and tube scanner (TUB) have been examined, The TRI 3 , is made from three discrete piezos for x, y and z, mounted orthogonal. The TUB is made from a hollow cylinder of piezoelectric ceramic which is commercially. available 4 . The outside of the TUB is supplied with four separate electrodes, the inner surface of the cylinder forms a continuous counter-electrode. By applying a voltage to the outer electrodes, the z-movement of the scanner is performed, The experimental set-up is described in detail elsewhere 2 . The displace- ment sensor 5 utilizes two bundles of glass fibers to illuminate a surface and to receive and measure the reflected light intensity from a surface. At the sensor tip, non-collimated light rays diverge outward from each fiber in a conical light path. The intensity of the reflected light leads to an output voltage with a sensi- tivity of 109.14 mV/pm. To determine the dynamic response of the piezos, an A. C. signal of 1 V at 1 kHz is superimposed to the D.C, voltage. These conditions are chosen to represent small topographic features, i.e., 1000 particles on a 1000 nm scan line with 1000 nm/s scan velocity. The dynamic displacement of the piezos is about 4.5 nm/V for the TRI and 0,65 nm/V for the TUB, respec- tively. The A.C, voltages are measured with a lock-in amplifier, At first, the step response of the TRI has been evaluated measuring the absolute (Figure 1b) and the relative (Figure 1c) D.C, displacement response and the simultaneously performed dynamic response (Figure 1d), At 50 V (marked with arrows) a strong hysteresis can be observed, Dividing the piezo displacement by the used voltage, more effects are visible, At 50 V the history dependent deviation is about 12.7 nm/V (i.e. 635 nm) even after 15 mm. Creep can be observed after a voltage change. If the voltage sign remains unchanged, the creep can be regarded as zero after 15 minutes settling time. Accepting an error of some percent, the required settling time may be regarded as 2 to 4 min- utes, With increasing voltage, saturation takes place, After a voltage step back to 50 V, a strong deviation is observed, caused by hysteresis as a dead band phenomenon, The amplitude response at 1 kHz (Figure 1d) shows several im- portant effects: /, The influence of the D.C. voltage is clearly visible and will lead to topography errors in a SEM image if an instantaneous change of the voltage takes place at topography steps. At 100 V a saturation effect can be noticed. //. A time dependent arm is found, which might hinder the interpretation of kinetic investigations. Hi. At 75 V, an increased drift as well as an unexpected noise can be observed. The sample tilt analysis is carried out using 0.3 V/s (Figures 24), The vir- gin curves are not given, up to 10 cycles are necessary for drift-free, i.e. closed 6 E B S I en a. a. CD o 30 40 50 60 z piezo voltage (V) 70 80 Figure 2: Piezo response or a triangular wave voltage cycle (25 nm/s steady state) and the first derivation of both branches, TRI piezo. -24- Downloaded from https://academic.oup.com/mt/article/7/4/24/6823306 by guest on 10 January 2023