ELSEVIER Ultramicroscopy 61 (1995) 215-220

ultramicroscopy

New design and imaging concepts in NSOM

Aaron Lewis *, Klony Lieberman 1, Nily Ben-Ami, Galina Fish, Edward Khachatryan, Udi Ben-Ami, Shmuel Shalom

Division of Applied Physics, The Hebrew University of Jerusalem, Jerusalem 91904, Israel

Received 29 May 1995; accepted 18 August 1995

Abstract

A normal force sensing NSOM probe has been developed based on a bent optical fiber. Using this near-field optical element a unique near-field optical head has been designed that can convert any far-field optical microscope into a multi-functional near-field optical, far-field optical and scanned probe imaging system. The design of this head is based on a singular flat scanning stage that includes the ability to achieve confocal imaging by altering the plane of the sample in relation to the lens by over 20/xm. The microscope has been used to distinguish gold colloids with 80 and 40 nm diameters. This cannot be achieved by any far-field optical system. Extensions of these imaging applications into biological systems in aqueous media are straightforward.

A number of years ago several goals were set for our near-field optics program at the Hebrew Univer- sity of Jerusalem. The objective of these goals was to achieve a generally applicable, reliable, user friendly near-field microscope that was totally integratable into all the diverse capabilities of far-field optical microscopy. An additional goal was to develop a contrast enhancement mechanism for NSOM that was tailored to the present capabilities of this emerg- ing technique. This mechanism had to be chosen with regard to the limited signal available in a subwavelength point of light and with consideration of the shortest image acquisition times that are realis- tic for present day NSOM emulations. The results

* Corresponding author. E-mail: lewisu@vms.huji.ac.il. i Present address: Nanonics Imaging Ltd., The Manhat Tech-

nology Park, Malha, Jerusalem 91487, Israel.

that we report in this paper demonstrate that we have reached at important juncture in our development of near-field scanning optical microscopy as a powerful and generally applicable probe of many problems that require nanometric optical resolution.

The steps in this development began in the early 80's when our group unequivocally demonstrated that subwavelength apertures down to 15 nm, that were well-characterized with scanning transmission electron microscopy, gave readily detectable throughputs of transmitted [1] or fluorescent light [2]. Independently, Poht and coworkers transmitted light through a subwavelength aperture at the tip of an etched quartz rod. The subwavelength nature of this aperture was deduced by a line scan across a grating [3]. This was followed by our adaptation of a tech- nique [4] for pulling glass with controlled heat, tension and cooling to produce a tapered glass struc- ture. These glass structures had a subwavelength point at the tip that could then be coated with metal

0304-3991/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0304-3991(95)00112-3

216 A. Lewis et al. / Ultramicroscopy 61 (1995) 215-220

to produce a subwavelength aperture for NSOM. This approach is presently used by practically all workers in this field to produce subwavelength aper- tures for NSOM. In the initial emulation of this technology a wire based heat source was employed to pull micropipettes of borosilicate and aluminum silicate glass to tips of nanometer dimensions. To couple light effectively into such a structure an optical fiber was inserted up to the highly tapered region of the pipette. Subsequently, the identical pulling technology was employed, with a CO 2 laser heat source [5], and then the technology could di- rectly be used to pull optical fibers to a subwave- length tip. This tip was then similarly coated with metal to produce a subwavelength aperture. The result was a dramatic increase in the throughput that was achievable since the coupling loss between the fiber and the subwavelength aperture was approxi- mately 3 orders of magnitude.

The adaptation of the technology of pulling glass tubes with controlled heat, tension and cooling not only resulted in an enabling approach that resulted in the production of cheap, reliable and reproducible subwavelength apertures for NSOM but it also per- mitted the placement of the subwavelength aperture in the near-field of even the roughest surfaces and provided a structure with exceptional force sensing capabilities that allowed the scanning of the aperture in the near-field [6-8].

With such straight near-field optical elements only one method of force sensing has worked and this is the method of lateral force sensing [9]. To employ the technique of lateral force the tip is modulated with a few gngstri3ms of modulation and when this tip approaches a surface the frictional force of the surface alters the modulation and the amplitude of the tip [7,8]. Five independent methods of lateral force detection have been developed for monitoring the surface induced alteration in the amplitude and the frequency of the tip modulation. These are inter- ferometric measurement of the tip amplitude [7], detection of the light emitted through the tip which is then transmitted through a transparent sample [8], diffraction of a light source by the tip [10] and two non-optically based techniques one of which uses a simple method with a tuning fork [11] and the other employs capacitance detection of the lateral force [121.

The principle advantage of the above lateral force method is that it is applicable to straight near-field optical elements. In some cases this is what you are limited to. An example of such a case is direct ablation near-field lithography which uses a pulsed argon fluoride excimer laser emitting at 193 nm [10,13]. This laser can remove material without any deposition of heat and it has been used to create lines with 50 nm dimensionalities [10]. Nonetheless, there are several disadvantages of the lateral force tech- nique in terms of imaging. First, different surfaces can produce very different functionalities for the approach of the aperture to the surface. This means that constant lateral force does not always mean constant height above the surface [9,14]. Second, when soft surfaces are to be investigated it is well known that lateral forces are to be minimized at all costs [15]. Third, the ability to work under water is more complicated when lateral force techniques are employed. Fourth, the technique is prone to fail more than it is desirable for a reproducible and reliable method. Thus, it would be of great benefit if an alternate method can be found for feedback of the t ip/sample separation that would resolve all of the above issues.

We realized this several years ago and began to consider the resolution of the problem in terms of using the standard technique of normal force mi- croscopy. In fact, the early work in our laboratory to characterize the force characteristics of these glass structures were completed by developing a technique of bending the tip of a micropipette [6]. We have over the last several years perfected the technique of producing bent tips. The tips are first pulled from single mode optical fibers via the standard methods in a Sutter Instruments P-2000 CO 2 laser powered micropipette puller. The pulled tips are then bent by passing them rapidly through the focal spot of a second CO 2 laser with a micro-manipulator while viewing the tip through a microscope. Sharp, accu- rate and very reproducible bends of less than 50 ~ m from the tip can be obtained with this technique. By varying the time the tip is placed in the beam the bend angle can also be controlled. The fiber tip is glued to a small stub, leaving a free cantilever length of between 100-400 /xm. The precise cantilever length can be controlled to within less than 10 /zm and determines the force constant and resonance

A. Lewis et al. / Ultramicroscopy 61 (1995) 215-220 217

Fig. 1. Bent fiber tip.

frequency of the tip. The result of these technologi- cal advances is seen in Fig. 1.

These tips are exquisitely sensitive structures for monitoring normal force. The tips produced by the above method have characteristic resonance frequen- cies that can be varied from 180-380 kHz. They have force constants that can be varied between 0.01 and 10 N / m for contact methods of force imaging or from 10 to 100 N / m for non-contact methods. Tap- ping mode protocols can also be used in which the scanning algorithm first brings the tip into contact with the surface at each point in the sample to be scanned and then the tip is lifted from the surface between the pixels of the scan, out of interaction range with the surface. This algorithm (developed by Zhong et al. [15]) dramatically reduces the lateral or shear forces on the sample and spectacularly im- proves the images that have been obtained. These bent near-field elements have important advantages over conventional force cantilevers in terms of their tip profile which is excellent for imaging deep fur- rows and can be readily adjusted to the problem at hand [6,16]. The cantilever length can also be readily adjusted and this can be used to change the force constant which depends on the length of this can- tilever cubed [6,16]. Also this allows the cantilever to be clamped at various points which can alter the

resonance frequency of the structure. In addition, the tip size can be readily adjusted for changing the interaction area of the tip on the sample and thus altering the compressional forces at any one point on the sample [6,16].

Such bent glass structures can be used in any of the available designs for force microscopy either commercial or homemade and the results obtained are as good as with any commercially purchased silicon cantilevers [16]. As has been shown a variety of samples can be imaged with these structures with results that are comparable to the best images that have been obtained to date with force microscopy which has evolved as a result of normal force tech ~ niques into a reliable, reproducible and simple tech- nique to use.

What we have accomplished is to extend this achievement in force microscopy to NSOM. For this we had to develop methods of coating such bent glass structures with metal to transform them into near-field optical elements. Although pipettes can be used for such purposes using epi-illumination methodologies that we developed for subwavelength light sources [17] and this method has some unique characteristics, the technology is still under further development. Thus, in the present report we describe the results of our efforts at guiding light to the tip of a bent, normal force sensing optical fiber.

To accomplish this, the fiber is first bent by the procedure described above and then a multilayer metallic coating is evaporated onto the probe held in an appropriate geometry to coat the entire length of the cantilevered fiber while leaving only the aperture at the tip exposed. This is a rather delicate proce- dure, requiring that the tip be rotated around the axis of the bent region to achieve a uniform coating along the walls without clogging the tip. The exciting results that have been obtained are best illustrated with a photograph of the resulting light that is seen emanating from the tip of the bent optical fiber (see Fig. 2). These near-field optical elements that we produce give the same light intensity as the straight tapered optical fibers. Nanowatts of intensity are obtained for tip dimensions of approximately 100 nm. In retrospect this result is clearly understandable since the biggest losses occur a t the subwavelength aperture where, after a relatively low threshold of a few milliwatts of laser power into the fiber, the

218 A. Lewis et al. / Ultramicroscopy 61 (1995) 215-220

Fig. 2, Bent mounted and coated NSOM fiber tip with light emanating from the subwavelength aperture.

coating of the tip begins to peal from the deposited heat. This is due to the exponential losses experi- enced by the light in the subwavelength regions of the tapered optical fiber and these losses cannot be compensated for due to the threshold effect in the pealing of the coating. However, the bend in the fiber is not in a region of subwavelength fiber di- mensions. Thus, any losses felt in the light intensity are of a magnitude :that can be compensated for without any fear of causing damage to the coating of the fiber in these:regions.

This brings us one step closer to the goal of incorporating the ease, simplicity and reliability of normal force microscopy into NSOM. Nonetheless, in spite of these achievements in probes that have been borrowed directly from scanned probe mi- croscopy, the design of near-field optical systems, which have generally been based on the principles that have guided the design of scanned probe micro- scopes, could gain considerably by including some of the aspects of far-field microscopes. Specifically, in all reported home-built scanned probe systems or commercially available instruments the piezo scan- ning mechanisms are perpendicular to the plane on which the sample is placed. This has been also the case in near-field microscope systems and has re- sulted in many ingenious solutions to the problems

of fully integrating near-field and far-field mi- croscopy.

Through our efforts in Jerusalem we decided that a radical alteration in near-field optical design was required in order to achieve a full integration with far-field microscopy. Towards this end we have been successful in constructing a totally flat scanning stage which presently has the ability to scan in x, y and z over a range of 20 /xm with very fine resolution and can rough scan a sample over several millimeters (see Fig. 3).

The scanner is based on four cylindrical piezo- electric elements which are placed in a horizontal position and are oriented clockwise in a square ge- ometry with the sample mount in the middle. The x scanning is accomplished by one pair of these cylin- drical piezos while the y scanning is achieved by the other pair of piezos. The z movement is accom- plished by all four of the piezos simultaneously bending and because of the horizontal geometry of the piezos the z extension is the same as that of the x o r y .

Sitting on top of this flat scanner is an additional plate which holds the tip mount and the control laser and position sensitive detector for normal force sens- ing in a design that permits far field optical elements, i.e. lenses, to be brought in close proximity to the sample from both the top and the bottom of the sample stage. This is a result of having the optical axis above and below the tip and sample completely unencumbered by the generally accepted methods of scanning in scanned probe microscopies.

Fig. 3. Integrated near-field, normal atomic force and confocal microscope system.

A. Lewis et al. / Ultramicroscopy 61 (1995) 215-220 219

there is a complete correspondence between the two images. Finally, in the images found in Fig. 5C there is an additional demonstration of zooming in from medium to high resolution where three higher den- sity balls and three lower density balls are, clearly

Fig. 4. Normal force image of a grating produced by interfering an Ar + laser on a photoresist with subsequent development and coating. The repeat of this grating was 0.4 /zm.

The result is a microscope with exceptional quali- ties for both near-field optical imaging and normal force imaging. For such imaging the tip is easily attached with a magnetic mount (see Fig. 2) and with these tips the microscope has produced excellent normal force images. An example is shown in Fig. 4.

The first near-field optical imaging application that we have focused on in our work is one that fits several objectives. First, we wanted to achieve with the near-field microscope a task that far-field mi- croscopy is unable to resolve. For this we chose the task of identifying gold balls with 80 and 40 nm diameters. This task had several other advantages since such colloidal gold is also used as a standard in transmission electron microscopy which is the ac- cepted technique for resolving such structures. The balls can also be readily attached with selectivity to proteins by connecting them to antibodies and this permits us to address several interesting ~problems in the micro-domain structure of biological membranes. In addition, such high contrast labels have the poten- tial of increasing the speed and the resolution of the NSOM probe since both of these issues :are rela~ed to signal to noise criteria.

In the first image of these gold balls seen in Fig. 5A, a field of 8 /zm has been scanned. The balls in this image are seen as holes having less intensity. In Fig. 5B is a high resolution scan of one section of the field of view from the above 8 /~m image. Notice, that in zooming into this field we have not moved any of the positions of the dark balls and that

Fig. 5. Images of colloidal gold with 80 and 40 nm diameters. (A) An 8X8 /zm scan, (B) a 6X6 /zm scan and (C) a 2.5X2.5 /zm s c a n .

220 A, Lewis et al. / Ultramicroscopy 61 (1995) 215-220

seen. Thus, we have been able to differentiate be- tween these two sizes of gold balls. An important point is to note that such an identification is possible even though the sizes of all of the balls in the image are close to being the same. This arises from the fact that these images are obtained with light being passed through a subwavelength aperture at the tip of a fiber that was larger than the largest ball. Nonetheless, notice that significant dark regions in these structures indicate that, with such high contrast objects, there is enough signal to noise that the resolution that is being obtained is considerably better than the dimen- sion of the fiber aperture.

All of these images were obtained with a far-field collection lens of 0.5 N.A. and 50 × magnification. In addition, the detector that was used was a standard low dark noise photon counting photomultiplier with a quantum efficiency of 14%. Thus the signal to noise in these images can be improved considerably by using an 1.4 N.A. oil immersion objective and an avalanche photodiode. Furthermore, a smaller tip size will allow us to significantly improve the resolu- tion that we have achieved thus far.

In summary, we feel confident that the advantages of normal force feedback in near-field microscopy will be recognized by the wider community of near- field microscopists. The ability to fully integrate far-field optical techniques with near-field method- ologies will lead to a synergistic interplay that will provide beneficial improvements in far-field resolu- tion while expanding the view of the near-field into the depth of the sample. When this is combined with the multi-functional imaging capabilities of near-field probes such as near-field ion sensing, conductance and force the future looks bright for near-field mi- croscopy. In order for the explosion in the interest in NSOM to continue NSOM will have to move from proof of concept to solving real problems which no other technique is able to accomplish. As this occurs a flood of ingenious microscopists will enter into our field and continue the explosion in the scope and applicability of NSOM that we have seen in the past several years.

Acknowledgements

This work was supported by grants from the Israel Ministry of Science and Technology and the United States Air Force. N.B. and U.B, are Levi Eshkhol Fellows. RHK Technology Inc., Rochester, Michi- gan, is gratefully acknowledged for the loan of an STM 1000 control system, an AFM 100 interface module and associated software.

References

[1] A. Lewis, M. Isaacson, A. Harootunian and A. Muray, Biophys. J. 47 (1983) 405a; Ultramicroscopy 13 (1984) 227.

[2] E. Betzig, A. Lewis, A. Harootunian, M. Isaacson and E. Kratschmer, Biophys. J. 49 (1986) 269.

[3] D.W. Pohl, W. Dank and M. Lanz, Appl. Phys. Lett. 44 (1984) 651.

[4] A. Harootunian, E. Betzig, M.S. Isaacson and A. Lewis, Appl. Phys. Lett. 49 (1986) 674.

[5] E. Betzig, J.K. Trautman, T.D. Harris, J.S. Weiner and R.L. Kostelak, Science 251 (1991) 1468.

[6] S. Shalom, K. Lieberman, A. Lewis and S.R. Cohen, Rev. Sci. Instrum. 63 (1992) 4061.

[7] R. Toledo-Crow, P.C. Yang, Y. Chen and M. Vaez-Iravani, Appl. Phys. Lett. 60 (1992) 2957.

[8] E. Betzig, P.L. Finn and J.S. Weiner, Appl. Phys. Lett. 60 (1992) 2484.

[9] C.M. Mate, G.M. McClelland, R. Erlandson and S. Chiang, Phys. Rev. Lett. 59 (1987).

[10] A. Shchemelinin, M. Rudman, K. Lieberman and A. Lewis, Rev. Sci. Instr. 64 (1993) 2528.

[11] K. Karrai and R.D. Grober, Appl. Phys. Lett. 66 (1995) 1842.

[12] J.-K. Leong and C.C. Williams, Appl. Phys. Lett. 66 (1995) 1432.

[13] M. Rudman, A. Lewis, A. Mallul, V. Haviv, L Turevets, A. Shchemelinin and I. Nebenzahl, J. Appl. Phys. 72 (1992) 4379.

[14] A. Jalocha, M.H.P. Moers, A.G.T. Ruiter and N.F. van Hulst, Ultramicroscopy 61 (1995) 221.

[15] Q. Zhong, D. Inniss, K. Kjoller and V.B. Elings, Surf. Sci. 290 (1993) L688.

[16] K. Lieberman, A. Lewis, G. Fish, T. Jovin, A. Schaper and S.R. Cohen, Appl. Phys. Lett. 65 (1994) 648.

[17] A. Lewis and K. Lieberman, Nature 354 (1991) 214.