Ultramicroscopy 42-44 (1992) 1596-1601 North-Holland

AFM and STM activities at Advanced Technologies Center

Yu.N. Moiseev, V.I. Panov, S.V. Savinov, S.I. Vasi l 'ev and I.V. Yaminsky Advanced Technologies Center, Moscow University, 119899 Moscow, Russia

Received 12 August 1991

A scanning tunneling microscope combined with an atomic force microscope as an option for surface studies in air at room temperature is described. The device characteristics (AFM and STM) are: normal resolution < 0.01 nm, thermal drift < 1 n m / h , overall measurement time per one pixel > 0.3 ms. The application of graphite structures for subnanometer metrology is discussed. A friction mode is implemented in the AEM for contrasting atomically resolved images. Different materials (quartz substrates, diamond-like thin films, Langmuir-Blodgett films) with unique atomic flatness appropriate for nanotechnological purposes are investigated.

I. Introduction 2. Instrumentation

Biological studies, thin-film investigations, na- noelectronic device development using atomic force microscopy (AFM) [1] and scanning tunnel- ing microscopy (STM) [2] lead to the necessity of conductive and dielectric thin films and substrate production and metrological control. It is impor- tant for these applications to obtain thin films and substrates, which are uniquely smooth down to nanometer scale. The corrugation height of particular surface features should be no more than a few nanometers. The flatness of thin films and substrates simplify the controlable deposition of materials and objects and make the interpreta- tion of experimental results very often easier. At present we apply AFM and STM for investigation of substrates proper for nanotechnological pur- poses and sample preparation. The tungsten dop- ing was used to vary the conductivity of dia- mond-like thin films. No cluster formations were revealed even with a high rate of doping. Quartz dielectric substrates prepared by fusion reveal the surface corrugation height to be less than a few ~tngstr/Sms.

The mechanical head of the STM is shown in fig. 1. Either a piezoelectric tripod or a tube can be used as precise actuator. A sample is installed in vertical position on a quartz plate. The plate can be moved manually or by a piezoelectic bi-

% t ~

Fig. 1. STM mechanical head: (1) sample holder, (2) tip, (3) piezoactuator (tripod-variant), (4) microscrew, (5) plate, (6)

b i m o ~ h plate, (7) mechanical transformer.

0304-3991/92/$05.00 © 1992 - Elsevier Science Publishers B.V. All rights reserved

Yu.N. Moiseev et al. / A F M and STM activities 1597

'v'z ;' h I 'r~u 200 ~V

0 0 2kHz

Fig. 2. The noise spectrum for gold sample. Tunnel current

I t = 0.1 nA, voltage U t = 20 mV.

morph in horizontal plane; it stand on three point supporters. The preliminary t ip-sample adjust- ment up to 0.4 mm is provided using a micro- screw and a mechanical transformer with 2:1 arms ratio. The precise pushing of the sample towards the tip to the tunnel region is done in automatic regime using a bimorph with the dy- namic range 0 -500 /zm. The mechanical head is installed on a compact vibration-isolation plat- form, having mechanical frequencies of the order of several Hz. The microscope stands on an ordi- nary laboratory plate without any additional pre- cautions for eliminating the essential seismic noise. The typical noise spectrum (the worst case of the tripod actuator) gives no evidence for the resonances in the mechanical head (fig. 2). The lowest resonant frequency of mechanical head oscillations equals 4 kHz. The high rigidity of the tripod was achieved due to the proper choice of actuator dimensions (electrode length and cross section are 35 mm and 3.5 x 3.5 mm correspond- ingly) and its rigid fixation to a quartz foundation. The distance/voltage transform efficiency is 2.3 nm / V .

For dielectric surface investigations the STM sample plate and tip holder are replaced by tun- nel displacement sensor with AFM tip attached to a lever and a sample holder respectively. The tunnel sensor tip is mounted on a piezobar with a

dynamic range of 0-1 .5 /xm. The opposite to the AFM tip surface of the lever is brought into the vicinity of the tunnel tip. The lever is usually made from a gold wire (30 ~m in diameter, 2-5 mm length), or plate with dimensions 20 x 100 ~m x (3-5) mm.

AFM was implemented in two regimes: for measuring only normal forces or for measuring both normal and lateral (or frictional) forces. In the first case the lever bending only in one nor- mal direction is used; in the second case usually a round wire lever is used with the same rigidity in normal and lateral directions.

STM and AFM parameters are: normal resolu- tion 0.01 nm, thermal drift 1 n m / h , measurement time per one pixel > 0.3 ms.

The STM calibration can be performed using H O P G as a metrological standard. The STM investigations in air atmosphere can lead to quite different visible images of graphite (figs. 3a and 3b). Additional modification of images can be seen during the repeatable scans. But it is of importance for metrology that the structural pa- rameters of the atomic lattice remain constant. It makes it possible to calibrate the actuator in two directions, X and Y, parallel to the graphite surface. The actuator calibration in normal direc- tion using H O P G is complicated: in constant-cur- rent mode the corrugation height visible in STM depends drastically upon tunnel current and volt- age, and also upon the presence of adsorbate layers. In constant-height mode one must obtain a precise current-distance dependence. These obstacles do not permit one to calibrate accu- rately the actuator in Z direction using H O P G in STM experiments. On the contrary, the graphite images, obtained using AFM in the normal force regime, are reproducible. Surface corrugation height corresponds to 0.02 nm. Meanwhile, in a friction mode the topographs reveal enlarged cor- rugation height of graphite up to several fractions of nm (fig. 3c). The lateral displacement of the lever due to friction forces can lead to the varia- tion of the tunnel-t ip-lever distances in the range of fractions of nm. Alternative calibration of the piezoactuator using precise capacitance dilatome- ter is in accurate coincidence with the previous measurements.

1598 Yu.N. Moiseet~ et al. / A F M and STM actit~ities

0,I nm/div

(al

o ,1 nm/di'~

Fig. 3. (a,b) Different STM images of the same region of HOPG surface. (c) AFM image of HOPG in the friction mode.

(c) nm/~ i

Fig. 3. (continued).

3. Diamond-like films

Diamond-like films (DLF) were prepared us- ing a novel method described in ref. [3]. The surface conductivity of films can be varied in the range of more than 10 orders (from typical insula- tor to metallic) by implementing a different dop- ing by a number of different metals (Cu, Hf, W, Cs and others).

Local electronic structure and sample topogra- phy of DLF, doped by tungsten, is investigated using STM and AFM, as described above. The experimental data are indicative of high flatness and uniformity of the DLF. There is no experi-

n r n / d i v C

11 I n /

Fig. 4. (a) Small-scale STM and (b) AFM images of DLF.

Yu.N. Moiseeu et al. / A F M and STM activities 1599

(

Fig. 5. Large-scale STM images of DLF on (a) alumina ceramics and (b) Si substrates.

mental evidence for any periodic structure for either small (1 nm) or large (100 nm) dimensions. The STM (fig. 4a) and A F M (film conductivity 700 k 1 2 / [ ] ) (fig. 4b) topographs are very similar.

The microprofile peculiarities are not higher than 0.5 nm on the scan area 5 × 5 nm. The surface structure on the greater areas depends upon the substrate, on which the DLF is deposited. The

O.1

1 v ~ n 1 (b) n~/dl i m/dl • iv

Fig. 6. AFM images of cadmium stearate LB film on (a,c) CdS and (b,d) GaAs substrates.

1600 Yu.N. Moiseev et aL / AFM and STM activities

comparison of the DLF (film thickness 300 nm, conductivity 9 11 / [ ] ) on alumina ceramics (fig. 5a) and DLF (film thickness 400 nm, conductivity 200 l - l / [ ] ) Si substrate (fig. 5b) is done. The film on Si substrate reveals the characteristic features 20-30 nm in length and 2 nm in amplitude. For alumina ceramics substrates the dimensions are rather greater: 40-60 mm in length and 10-20 nm in amplitude. The topographic structures de- pend upon the type of substrates despite of the big thickness of the film.

The microprofile character of the DLF proofs, that there is no tungsten cluster formation on the surface even with the high concentration of dop- ing (which can exceed 50%). We assume that tungsten atoms penetrate into the holes (about 0.3 nm in diameter), which appear during film formation. Nonuniform peculiarities of micropro- file can be interpreted as such holes less than 0.5 nm in diameter, as it was sometimes seen in STM topographs on films with 200 12/[] conductivity.

stearate reveal its high uniformity without any damage, dislocations and gaps (figs. 6c and 6c).

5. Quartz substrates

Quartz substrates display unique optical, elec- trical and mechanical properties and are widely used in microelectronic device fabrication. The experimental data about local structure of the substrate at near-atomic scale is necessary for nanotechnological applications.

The microstructure of dielectric quartz sub- strates is investigated using AFM as described above. The samples were prepared in the form of spheres, 2 -8 mm in diameter, by melting quartz in an oxygen gas burner. After essential cooling at air a tmosphere the substrates were kept at normal conditions (room temperature and air at- mosphere with ordinary humidity) for several weeks. Surface topography is studied in the re- pulsive-force (10 -8 N) regime. The only type of

4. Cadmium stearate LB films

Cadmium stearate may be used for thin film dielectric coatings on different substrates. Pre- sent investigations deal with films deposited on GaAs and CdS substrates. Thin films can be exposed to intensive electric fields (105-106 V / c m ) without the risk of damage.

CH 3 groups can be clearly resolved in AFM images of cadmium stearate on CdS (fig. 6a). The characteristic dimensions are d~ = 0.49 nm and d 2 = 0.54 nm. Dimension d I corresponds to the distance between alkine tails (d = 0.5 nm). A small discrepancy between dl, d 2 and d may be explained using a dense-packed model.

The topograph of cadmium stearate on GaAs substrate is shown in fig. 6b. The dimension of characteristic features is d = 0.63 nm. The differ- ence between two images lies in the lower com- pactness of the layer on the GaAs substrate. This is proved by the comparison of the efficiency of layer transportation on the substrate surface. The efficiency of LB layer transportation on CdS sub- strate is about 100%, while for GaAs substrate it is about 95%. The topographs of cadmium Fig. 7. Fused silica substrates.

Yu.N. Moiseev et al. / A F M and STM activities 1601

surface irregularity observed at the profile images (fig. 7) is the smooth "wave", no more than a few ~ingstr6m in height; above this molecular level the surfaces are practically flat and uniform.

The indirect method [4] based on the hydrody- namic measurements of viscous forces at small distances (nanometric range) between two quartz substrates of spherical form indicates also the unique flatness of the quartz substrates.

6. Summary

Both dielectric and conductive substrates with unique molecular-scale flatness are investigated.

The reproducible flatness of the substrates on nanometer scale is among other important prop- erties which make them advantageous for nano- technological applications and sample prepara- tion.

References

[1] G. Binnig, C.F. Quate and Ch. Gerber, Phys. Rev. Lett. 56 (1986) 930.

[2] G. Binnig and H. Rohrer, Helv. Phys. Acta 55 (1983) 726. [3] V.F. Dorfman et al., to be published. [4] V.N. Steblin, E.D. Shchukin, V.V. Yaminsky and I.V.

Yaminsky, Hydrodynamic interaction of surfaces in elec- trolyte solution, J. Colloid Interface Sci., in press.