Ultramicroscopy 42-44 (1992) 1606-1609 North-Holland

Device for high resolution positioning

R o g e r Car r

Stanford Synchrotron Radiation Laboratory, P.O. Box 4349, bin 69, Stanford, CA 94309, USA

and

R u t h Ellen T h o m s o n

Department of Physics, Unicersity of California, and Materials Sciences DA, ision, Lawrence Berkeley Laboratory, Berkeley, CA 94720, USA

Received 12 August 1991

This paper is a description of a high resolution actuator consisting of a smooth round shaft and a mechanism which translates when the shaft is rotated. The motion is monotonic and controllable to less than 900 ~ngstr/Sms per degree of rotation. The mechanism is rigid, stable, and offers an unlimited range; it is vacuum compatible, and ideally suited to the approach function for a scanning tunneling microscope.

I. Introduction

In scanning tunneling microscopy, a probe tip is brought near the surface of the sample by means of combined coarse and fine approach mechanisms. The fine approach mechanism is usually a feedback-controlled piezoelectric actua- tor. Coarse positioning may require motion over several millimeters, and the final coarse approach requires a resolution of a few hundreds of ~ngstr6ms. The range requirement is dictated by the need to exchange samples, or by certain other analytic or preparation techniques which need clearance from the tunneling tip. The resolution requirement comes from the need to avoid tip- to-sample contact by allowing the fine positioning apparatus to retract when contact is imminent. It is very important to avoid contact, or "crashing", as this can harm both tip and sample, perhaps irreversibly. Thus, rapid forward jumps are risky. Coarse positioning does not require very uniform or repeatable motion, but it must be stable, hav- ing reached its target.

The history of scanning tunneling microscopy records many coarse positioning devices, which are generally the most trouble-prone parts of the STM. There are piezoelectric capacitive louses, magnetic and inertial walkers, and various me- chanical lever, screw, and spring motion demagni- tiers. In order to limit vibrational noise, and to maximize thermal equilibration speed, an actua- tor which firmly couples the sample or tip to the microscope body is preferred. Piezoelectric, iner- tial, and magnetic walkers have very weak me- chanical and thermal coupling between tip and sample. They often have reliability problems which require time consuming debugging.

As STM's have become simpler, we have found that simple differential screw mechanisms are adequately sensitive, rigid, and stable for the coarse approach. However, most of these devices have a limited range. Our differential screw mechanisms have a range of only about 300 mi- crons, and lever reduction and spring motion demagnifier mechanisms typically have a range less than 1 mm. But one may need more clear-

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

R. Carr, R.E. Thomson / Device for high resolution positioning 1607

ance space for sample transfer or sample treat- ment and characterization. A variation of the pocket STM has a tilt mechanism which allows this type of motion [1]. This design incorporates a " foot" mechanical contact to improve tip-to-sam- ple coupling. We present here another purely mechanical design, which converts rotation to lin- ear translation with potentially unlimited range.

We were motivated by the design of the Rho'lix ® drive [2]. This mechanism is a commer- cial motion actuator which works by rotations of a smooth shaft through a block. On each end of the block are three bearings, whose outer races contact the shaft. The bearings are canted in such a way that they do not rotate in a plane normal to the axis of the shaft. Instead, each one contacts the shaft at an angle. When the shaft is rotated, there is a component of force applied along the axis of the shaft, and this causes the shaft to translate. The smallest amount of translation for a commercial Rho'lix ® is about 0.6 m m / t u r n . When we contacted the manufacturer of the Rho'lix ® device and asked if smaller transla- t ions / tu rn were possible, they advised us that manufacturing such a device would be very diffi- cult.

We took the approach that a shaft moving through a Rho'lix ® device whose bearings were not canted at all would probably still translate, because of inaccuracies in fabrication of the de- vice, and we expected such translations to be very small. We fabricated such a device, and found that such translation did occur. However, a stan- dard Rho'lix ® executes reversed translation when

the direction of rotation of the shaft is reversed. Our device showed no reversal of direction. But when we inserted a very thin shim under one of the bearings, we broke the symmetry of the de- vice enough to give it reversibility.

2. Experiment and results

Our device is shown schematically in fig. 1. It was fabricated to standard 0.025 mm tolerances. The bearings were ABEC class-7 precision ball bearings [3]; we found that lower tolerance ABEC class-3 bearings gave more erratic results. The shaft is a standard precision ground and hard- ened steel shaft. We noticed that the mechanism moves at different rates depending on position along the shaft; this suggests small variations in the diameter or surface finish. The inner diame- ter of the bearings provides a slight 0.05 mm interference fit with the shaft; thus the shaft is gripped more tightly as one tightens the shoulder screws which hold the bearings. The resulting radial loads, especially at the low speeds we used, would not shorten the life of the bearings signifi- cantly. In a commercial Rho'lix ® device, the block is split, and the halves are held together with spring tension; one can regulate the force re- quired to slip the shaft with this spring tension. The original design of the Rho'lix ® was appar- ently motivated by a desire to have an actuator which could slip, to limit its force for safety.

The actuator was mounted on a 6.35 mm di- ameter shaft, and was prevented from rotating by

Fig. 1. The actuator assembly, showing a shim placed under the innerrace of one of the bearings. The main body is 25 mm in diameter and 25 mm long. The bearings are 9.5 mm in diameter, and the shaft has a 6.35 mm diameter. The ends of the shaft are

supported by the stationary part of the apparatus.

1608 R. Carr, R.E. Thomson / Device for high resolution positioning

attaching it to a linear ball bushing that rode on a parallel shaft. Both ends of the main shaft turned in lubricated holes which were line-bored in an aluminum block. Thus the motion recorded took into account any imperfections in this simple shaft-bearing scheme. All of our materials are

t

vacuum compatible, though in a vacuum one would prefer a bushing to support the shaft which would not gall with the shaft. A loaded linear bearing which permit ted both rotary and linear motion could also be used. One end of the pro- truding shaft was turned by a DC gear motor at a rate of 1 revolu t ion/min at 5 VDC or 3 revolu- t ions /min at 15 VDC. There was little difference in motion at these two speeds, but overrun upon stopping could be controlled simply by moving more slowly. We performed measurements using a Kaman KD2810 inductive measuring system, which is calibrated to yield 40 m V / / z m when its sensor is used on a target of aluminum, as we did [4].

With a piece of 0.05 mm shim stock placed under the inner race of one bearing, we observed an average translation of 40 / , m / t u r n in the forward direction, and 50 / . tm/ tu rn in the reverse direction. With a 0.025 mm thick shim, we ob- served 31 / zm/ tu rn forward, and 32 / zm / t u rn in reverse. This is less than 900 A per degree of shaft rotation. With still thinner shims, the mo- tion tends to become erratic. A typical graph of displacement versus shaft rotation for the actua- tor is shown in fig. 2.

The figure illustrates the monotonicity of mo- tion, and shows the behaviour when motion is reversed. The steps which do appear in the curve are not fast enough to cause crashing. Ordinarily, the feedback circuit controlling the t ip -sample separation is put into a fast response mode when the approach is done, and it is capable of retract- ing from coarse advances of a few hundred ~ngstr6ms quickly enough to avoid a t ip -sample contact. The motion of our actuator appears to be stable within the range of normal thermal drifts when the motor is deactivated. Since it is relatively easy to obtain stepping motors with several thousand s teps / turn , it should be possible to obtain discrete incremental motions on the order of 100 .A units with this actuator.

We also tested another type of smooth shaft actuator, a commercial unit known as the Rolling Ring Drive ® by Uhing [5]. It was a larger unit than the one described above, with a 15 mm shaft. It consists of a housing which contains three internally crowned ball bearings that sur- round the shaft and are canted against it with a spring mechanism. It comes with a direction se- lector and a tensioning device which determines the rate of translation. While it is well engi- neered, the standard version can only be adjusted down to a translation rate of about 350 / zm/ tu rn , which is bet ter than the commercial Rho'lix ®, but still within the range of f ine-threaded screws. There is also some non-monotonicity in the mo- tion. However, this concept could be of value if specially engineered as we did with the Rho'lix ~.

r ~ 1 2 0 z o ~a~ l O O

, 8 0

~.1 6 O

• < 4 0

~ 2 0

. . . . i . . . . i . . . . i . . . .

i L i L I i i i i I i b L * I i , , ,

0 1 2 3 4

S H A F T R O T A T I O N S

Fig. 2. G r a p h of typical d i sp l acemen t versus shaf t ro ta t ion, showing a reversa l of mot ion . The sh im was 0.025 m m thick,

and the mo to r was dr iven at 1 t u r n / r a i n .

3. Conclusion

In summary, we have described a very simple device which is capable of executing small mono- tonic translations. It consists of materials which could all be used in vacuum environments. It couples this motion very rigidly and stably to the stationary part of the apparatus, which also as- sures good thermal contact. Since it has a very large range of motion, it would be very suitable as an STM coarse positioning device, especially in those designs where access to the sample is de- sired.

R. Carr, R.E. Thomson /Device for high resolution positioning 1609

Acknowledgements References

We wish to thank John Scott of the S L A C shops for his expert fabrication of the actuator. R. Carr is suppor ted by the Office of Basic En- ergy Sciences, Division of Materials Science of the US Depa r tmen t of Energy. R.E. Thomson is suppor ted by the Director, Office of Energy Re- search, Office of Basic Energy Sciences, Division of Materials Science of the US Depar tmen t of Energy under contract no. DE-AC03-76SF00098.

[1] J.E. Demuth, R.J. Hamers, R.M. Tromp and M.E. Welland, J. Vac. Sci. Technol. A 4 (1986) 1320.

[2] Zero-Max Division of Barry-Wright Corp., Minneapolis, MN, USA.

[3] The ABEC (Annular Bearing Engineering Committee) standards are part of ANSI/AFBMA Standard no. 20, American National Standards InStitute, New York 1987).

[4] Kaman Instrumentation Corp., Colorado Springs, CO, USA.

[5] Joachim Uhing KG GmbH, Mielkendorf, Germany.