The optical near-field: super-resolution imaging with structural and phase correlation

1 The origin of nanophotonics

Nanophotonics or for that matter nano optics was certainly not a field in the early 1980s when the first experiments were done that led to what is now called near-field optics. This issue in the journal Nanophotonics focuses on progress in various areas of near-field optics demonstrating its generality of application. The object of this paper is to act both as an overview of the field and an introduction to the articles in this issue.

In addition to super-resolution near-field optics addresses the vexing problems in optics of confinement of light in X, Y, and Z, the structural correlation of the light distribution in an image and determining the amplitude and the phase of the electromagnetic field at each point in the image. Within the context of the current issue of Nanophotonics, this paper aims at giving the reader a perspective of the breadth and importance of this field on the 30^th anniversary of the first experimental reports that founded what has become a critical field in modern optics with great future potential.

The field of near-field optics has been instrumentation limited since its introduction in the 1980s. Thus, in trying to address the potential and limitations of the field and to present as broad a view as possible we focus, in this paper, on the instrumental developments that have advanced the field in the past and those that have the potential to have a tremendous impact on the use of the technique in the future. What is clear today is that near-field optics has the potential to touch every area of fundamental and applied optics. We can already see that it will impact many areas of science and technology from device physics to chemistry and even to biology where the impact of near-field optics has thus far been limited but already its potential significance is clear.

2 The near-field: a crucial step

To understand the basis of these instrumental developments it is important to emphasize that a key component in the development of this new field in optics was obviously the concept of the near-field.

In terms of physics it had been well known from the time of Sommerfeld that a radiating dipole had a near-field. However, in all of the decades since that knowledge was clearly in the literature the challenge of using the near-field for super-resolution imaging was not considered. Thus, from a fundamental point of view a key first step was the stated realization that the near-field could indeed be used for imaging.

Once that realization was appreciated it was necessary to:

• Fabricate an element that could be analyzed by known means that would emulate a radiating dipole.

• Reproducibly make such an element so that it could be investigated, in a quantifiable way, for light throughput. This was necessary in order to make a judgment as to the ability to use such elements for either point interrogation or spectroscopy and ultimately optical microscopy in the near-field.

• Produce such an element in a way that would allow tracking of even rough surfaces so that the radiating dipole could be kept within the near-field of even rough samples.

A key paradigm shift in appreciating the criticality of the near-field appeared in the literature in an abstract that was published in 1983 in the Biophysical Society Meeting [1]. In this abstract it was for the first time indicated that there may be a method to obtain the non-radiating components in the near field that are associated with obtaining super resolution optically. This paper was the first published in the public record that fully appreciated that the near-field was the essential element that was going to achieve a new direction in optics.

The first paper published based on this abstract, highlighted the importance of the near-field for such imaging and was the first paper that also quantified such a potential radiating dipole with standard methods. This was the paper published from our laboratory in 1984 [2]. In this paper the words and the importance of the near-field was mentioned as in the 1983 abstract in connection with such imaging. Also the experiments that were done measured light by known methods and correlated the light from such a quantifiable man-made radiating dipole with electron microscopic imaging that accurately imaged the dimension of the radiating aperture. Thus, the first and second criteria mentioned above were already met in 1983 and 1984.

Specifically, to obtain such results a well formed aperture, with a known geometry, was produced by the then incipient technique of electron beam lithography. This was done using polymethyl methacrylate which was just then being realized as an important resist for such electron beam lithography. Of course subsequent developing, etching and coating was required to produce the apertures in a gold palladium film. This produced the first man-made radiating dipole in the optical regime with an effective near-field that could be investigated.

What was particularly exciting about this advance was that, as a result of the fact that the size of the aperture was clearly determined, the amount of light from such an aperture could be related exactly to the aperture size. What was surprising was that these apertures made in Au/Pd showed a throughput that was clearly larger than what would be expected from either electromagnetic calculations or simple illumination area versus transmission estimation. This led to the conservative statement in this paper “Although the transmission through the 1 μm apertures appears consistent with exact electromagnetic field calculations, there appears to be more transmission through the smallest apertures”.

Being conservative in those days was prudent since to make such apertures was extremely challenging. Therefore, to make a definitive statement about having more light transmitted than electromagnetic calculations one needed apertures with different spacings in numerous materials and this was experimentally inaccessible. Nonetheless, having the thickness of the metal film, the exact aperture dimensions of arrays of apertures from 1 μm to 15 nm in the same film as verified with electron microscopy and making accurate throughput measurements on many of the apertures in the array allowed us to accurately predict the ultimate resolution of near-field optical microscopy with apertures to be 50 nm. No verifiable example of higher resolution has been reported. Furthermore, our assertion of higher light throughput than predicted by electromagnetic calculations based on Maxwell’s equations were verified by the elegant results of Ebbesen published 14 years later [3]. These later measurements used fully the power of the availability in the late 1990s of ion beam lithography that opened the horizons of the nanofabrication needed to definitely prove without a doubt such an assertion.

3 Early concepts

Up until this point no work emphasized in any publication the near-field for imaging. This was in spite of the fact that there were four periods where related research appeared without any enunciation of the concept of the near-field originally associated with the electromagnetic radiation pattern of a radiating dipole.

The first was work done by Synge in 1928 when three things were highlighted. The first was a possible use of an aperture as an element for super-resolution. The second was placing the aperture at a tip and the third was possibly producing some sort of aperture by crushing such a fully coated protruding tip of a glass shard against a surface. Synge made no attempts to experimentally verify or theoretically quantify any of these concepts [4–6].

In 1956 another paper appeared by O’Keefe [7] who did not know of the work of Synge. This paper used as a paradigm for super-resolution what could be achieved by the “occultation of a star by the moon (which) yields (as noted by O’Keefe) a somewhat analogous situation. A study of the light variation with time during the occultation allows a fellow with a small telescope to measure one component of the separation of a double-star pair which cannot be resolved with the biggest optical telescope. Occultation by an asteroid gives another large factor in resolution; that is how the rings of Uranus were discovered” [8]. Obviously once again in this paradigm there was little connection to the optical near-field.

Then in 1972 a paper appeared by Ash and Nichols [9] who did not know of the work of either O’Keefe or Synge. They investigated the concept of a sample being scanned opposed to a closely spaced aperture in the microwave regime and obtained λ/60 resolution. In spite of this work the optical near-field is a very different problem since as was mentioned in an early paper [10, 11] metal films around apertures have infinite conductivity in the microwave region of the spectrum and so the contrast between the aperture and the film is large. In the optical regime the metal film around the aperture has finite conductivity and this limits the contrast and the possible effectiveness of such a concept in the optical regions of the spectrum.

Finally, in 1984 the first papers appeared in the optical regime. We were not aware in our paper of Synge and O’Keefe but a reviewer of a grant proposal alerted us about Ash and Nichols.

In concert with our paper a report by Pohl et al. appeared which emphasized Optical Stethoscoopy rather than the near-field [12]. It was noted in this paper that a stethoscope using sound waves also obtains super-resolution as an aperture scanning over a surface. As was noted [11] however, acoustic waves always exhibit a propagating mode even in a sub-wavelength aperture whereas electromagnetic waves are evanescent in subwavelength apertures. Nonetheless, Pohl, also unaware of Synge and O’Keefe used the concept of Synge of crushing a coated glass tip against a surface to obtain an aperture and scanning a sample under such a glass tip. The aperture in this study could not be quantified and thus, the origins of 25 nm resolution reported in this paper are not clear.

4 Some next steps

Within the context of the above it is worth noting that our emphasis of the near-field led to the first calculations also from our laboratory [10] that showed the near-field extent as applicable to near-field optical microscopy. And finally the acronym near-field scanning optical microscopy or NSOM which has become widely used was introduced by our laboratory in a paper in the Biophysical Journal in 1986 [11]. This paper also detected for the first time fluorescence through such apertures a concept recently expanded by Ebbesen et al. [13].

Such point fluorescence through subwavelength apertures in the near-field has also taken on several other names including “zero mode waveguide” [14] for the detection of point fluorescence with small illumination volumes as in our earlier paper. Such concepts have been applied to DNA sequencing [15] and fluorescence correlation spectroscopy which has the potential for great import in biology with the resolution of instrumental problems of NSOM as applied to soft samples as noted below [16, 17]. With such advances in the near-term there should be exciting applications in biology (see for example the paper in this issue by Bhattacharya and Mukhopadhyay on the fluorescence of amyloid fibers on page 51).

5 Near-field optical approaches within the context of other more recently developed super-resolution methodologies

So where does near-field optics stand today? Obviously it is today recognized as a critical and important field of optics that has made numerous contributions and is set to make even more. Some of these advances are described in the papers included in this issue. As noted above however, in this report we would like to place these papers within the context of the numerous other areas of application that could obviously not be covered in one issue of a journal. Our goal is to place near-field optics within the wider field of super-resolution optics or attempts to break the Rayleigh diffraction limit. Furthermore, it is an objective of this report to indicate the areas of application and developing application that have led the optical near-field to have a pervasive influence in nearly every area of optics.

From a subjective perspective we can say near-field optics not only has made clearly critical contributions since its introduction in 1984 but is in an exponentially rising phase where its influence is expected to be ever more pervasive in the future than in the past.

What are the highlights of today that can lead to this future?

First, it should be noted that near-field optics has already demonstrated the highest resolution of any method of optics in the most general of circumstances.

Over the years since its introduction the concept of breaking the diffraction limit of Rayleigh however has become a central focus of optical imaging. This has led to many new concepts.

An example is stimulated emission depletion or STED microscopy [18] where depending on the intensity of light that the sample can withstand fluorescence super-resolution can be obtained in the far-field below 100 nm resolution. The technique uses ultrafast lasers to stimulate the emission of dye molecules thus achieving a narrower point spread function than would be obtained normally. This technique is limited to certain types of dye fluorescence.

To this can be added the exciting recent advances of PALM [19] and STORM [20]. Both of these techniques are once again limited to fluorescence and depend on a specific sub-class of dyes that can undergo reversible photobleaching. As a result of such bleaching closely spaced dye molecules can be interrogated individually and placed accurately as a result of an analysis of their far-field radiation pattern. This subclass of techniques also obtain super-resolution in far-field optics.

All of these far-field techniques, in addition to their intrinsic difficulties and limitations of being limited to a specific class of fluorophores have important limitations that have not been solved by any technique other than near-field optics. These are out-of-focus light contributions, the inability to be correlated with 3D structure of the object under investigation and the inability to obtain phase information on the electromagnetic field distribution of the object under study. Thus, when one wants to determine accurately the point spread function of the beam in STED a near-field optical element has to be used due to its generality of application [21].

Within this concept of super-resolution in far-field imaging an exciting new development is the use of the near-field optical point source radiating dipole as a reference source for a far-field image obtained with a CCD camera in order to obtain without iteration the exact phase and 3D structure of an object [22]. Added to this is the connection of the placement of this radiating dipole reference source with the precision of atomic force microscopy which has the potential to offer super-resolution imaging even in non-fluorescent mode by using the instrumental implementation of a non-obstructing scanned probe microscope/NSOM system. Furthermore, such imaging implementations based on near-field optical developments has already the potential to be combined with such modalities as STED in order to achieve 3D structure and phase together with STED super-resolution. These concepts of a reference source should also be very effectively combined with other modalities of far-field imaging such as structured illumination [23] which also like STED and PALM/STORM can achieve far-field super-resolution.

In spite of the generality of the imaging and super-resolution modalities available with near-field optics as shown in this issue, the optical near-field does have of course inherent depth limitations not shared by far-field techniques noted above. Nonetheless, this disadvantage has to be viewed within the ability of near-field imaging to provide phase information without any out-of-focus light contribution. Such information when combined with the amplitude that is known at each point in the image allows for back propagating the wave front which could in principle define the nature of the wave front at any depth of the object being imaged.

The power of this combination was clearly enunciated early on in the application of near-field imaging to photonic band gap materials [24]. This has applicability in silicon photonics [see cover page of this issue where the propagation of the amplitude and phase of a wave in a silicon waveguide in 3D was readily obtained using the unique feedback features offered by normal force tuning fork feedback discussed below]. Obviously such information is also of importance in plasmonics and a report on plasmonic applications is included in this issue on page XX. Other aspects of the importance of phase will be discussed below within the context of advancements of different available probes in near-field optics.

6 An instrumental perspective of present and future advances

In order to achieve the bright future for near-field optics predicted in this article new concepts in optically integrated scanned probe microscopy have to be emphasized. Within this goal it is important to note that the 1984 introduction of near-field optics occurred 2 years before the introduction of atomic force microscopy (AFM) [25]. And, even after this critical development of AFM which near-field optics depends so integrally on, the probes used, the scanning mechanisms employed and the feedback implemented all had serious limitations as far the development of the optical near-field. Thus, in this section we will deal with each of these elements of an NSOM system since, as noted, instrumental limitations have been the main hurdle in its full and general applicability.

7 The probe

7.4 Far-field illumination and apertureless probes

An early example of light at the tip of a probe that could be used for near-field imaging was the production of fluorescent material at the tip of a probe [46]. This was later extended in a beautiful piece of work to the single molecule limit at the tip of a probe at 1.4°K [47]. This approach has great potential either as a far-field optically excited point of light in the near-field of a sample or as a fluorescent detector of a near-field effect as illustrated by Maier et al. [39].

Sensing ionic changes with such a fluorescent NSOM far-field excited probe has also been demonstrated [48]. Along this same line is an electrically excited light source in the near-field that was originally reported by Kuck et al. [49]. In this same vein and of considerable interest is an actively (electrically) excited near-field heat source on a cantilever that is exposed to the optical axis [50]. This has potential as a broad band infrared source. An FTIR spectrum of such a black body source is shown in Figure 1.

Figure 1

A glass encased dual wire thermal source AFM sensing probe with its associated IR spectrum for a variety of currents traversing through the junction. One readily sees in air the spectrum of water vapor and carbon dioxide in the surrounding excited by the probe.

Such probes have considerable potential in ultra-high resolution infrared microscopy. Initial work in this direction was done by Dekhter [51]. All of these approaches have, for generality of application, to be coupled with an appropriate general feedback mechanism that would work even with the softest samples. This issue will be dealt with in more detail when feedback in NSOM is considered.

In this vein is also the concept of scattering or aNSOM which is an apertureless NSOM technique with a passive rather than active probe tips as described above. This concept has evolved into a laser focused in the far-field onto a tip of an AFM probe. The far-field focus is of course much larger than the size of the tip and so there is a lot of unwanted scatter. Modulation is used in concert with interferometric techniques to try and extract the near-field of the tip. This has been recently described with great clarity by Scott et al. [52].

The technique has generated most interesting results for elastic scattering of photons. However, as noted in the literature “the technique consists of the presence of an intense background light reflected from outside the interaction area between the probe and the sample, which makes the detection of the near-field scattered light very difficult” [53]. Also reliability and consistency problems in comparing the same or different materials have appeared [54] and these probably arise from the fact that the technique depends a great deal on instrumental parameters. These include the oscillation amplitude of reference mirrors in the interferometric techniques used to try and isolate the signal of the near-field. A variety of other factors are also shown to have an influence including any slight change in the tip sample distance, tip roughness and shape etc and as the effect is further modeled new influences are uncovered. As a result, the literature is not without its artifacts.

In addition to the fact that there is a lack of generality since inelastic photons cannot be interrogated with such techniques there is also the issue of background free imaging which is inherent in many near-field optical techniques and is required in many areas. An example is fluorescence where especially organic materials are very susceptible to bleaching of the surrounding by the far-field illumination.

Additional examples of the need for background free super-resolution illumination not provided by other methods available today is in the area of photoconductivity described by Narayan in this issue on page 19.

Furthermore, in addition to such needs for background free illumination there is the need in photonics to be able to collect a light field attached to the near-field of a surface or for that matter profiling in free space a propagating fields from say a point source without out-of-focus contributions.

In a number of spectroscopies such as linear and non-linear Raman spectroscopy or second harmonic generation apertureless methods are certainly applicable since there is no issue of bleaching which requires light confinement and the scattered radiation is at another wavelength. Thus, far-field illumination can be used to excite the tip of a probe and such approaches have been used in Raman spectroscopy (see article by Zachary Schultz in this issue) or such approaches have been employed in second harmonic generation [55].

In the vein of vibrational spectroscopy infrared interferometric methods are applicable if infrared lasers are employed. However, there are competing methods both with active NSOM sources as noted above [50, 51] and new modes of detection that could have significant import in this area also. This is especially the case in view of the relatively low signal to noise that results from infrared scattering NSOM. These new methods involve either nanometric temperature sensing which is appearing to be ultrasensitive in the milli degree centigrade range [56] or employ ultra sensitive capabilities of normal force sensing of the photon force [57].

In summary, apertureless NSOM within its limitations of applicability is making important contributions. Current instrumental advances to be discussed below have the potential to make this an even more fruitful source of research in the future. This is especially the case when one considers multiprobe SPM/NSOM described at the end of this article and illustrated in the paper by Berezin et al. in this issue. Such methodologies have the potential to reduce the extent of the scattered light component in order to improve the signal to noise. An example of this is readily seen as a gold nanoparticle probe comes within the near-field of an apertured probe. This is the case even when a slightly larger aperture where trillions of photons are visible in the link of the video included [58]. Also the increase in the signal to noise offered by such apertureless protocols has the potential to replace the scattering probe with a low dielectric glass probe which would have a highly reduced perturbation on the investigated structure especially those where plasmonic imaging is important.

8 The feedback mechanism

A repeating theme in this overview is the quest for generality of near-field optical application. Within this stricture another repeating theme is to achieve low background that reduces noise. In essence, as all experimentalists know the way to better characterization and imaging methodology is certainly to increase the signal but of even greater importance is to reduce the background and any noise sources.

This section deals with Feedback Mechanisms that can be used with near-field optical probes to achieve both these aims: Generality and lower noise.

In addition, this section deals with a great attribute of near-field optics the ability to provide the optical distribution within the three dimensional structure of the sample being studied. In order to fully achieve this, one needs to have the ultimate in AFM both in terms of resolution and in terms of force sensitivity and positional sensitivity relative to the sample.

What can provide such sensitivity?

To understand the answer to this question within what is understood about force microscopy and NSOM fiber probes which are the most generally applicable optical element in near-field optics let us consider the alternatives within the context of these probes.

The standard method that is used with atomic force microscopy to monitor mechanical properties of materials such as elasticity and adhesion is based on beam bounce technology (see Figure 2). In this technology a laser beam bounces off an appropriately chosen cantilever that is soft enough for the particular material that is being investigated. The laser beam then tracks on a position sensitive detector the bending of the cantilever or its amplitude as a surface is being approached and this is analyzed for trying to understand the mechanical properties of the material of interest.

Figure 2

Beam bounce AFM feedback.

With such an approach there are two major mechanical problems. The first is jump to contact. This occurs on the approach of an AFM cantilever to a surface. As the tip of the cantilever approaches within ∼10 nm of the surface there is jump to contact since the cantilever is bent by the surface forces. This effectively negates analyzing forces or other physical parameters such as the intensity of light at such ultra close distances where extremely interesting phenomena are occurring.

In addition, on retraction from the surface there is an adhesion force that keeps the probe stuck to the surface. This causes a second instability in monitoring the force as a function of the distance of the probe from the surface.

During this entire process an oscillation of the probe is induced to try and understand the change in amplitude of the probe that results from the elasticity of the material under investigation. Thus, these instabilities mask the true interaction of the tip with the surface and complicated methodologies both mathematical and instrumental are needed to try and unmask the behavior of the interaction. However, in spite of such valiant attempts even the critical exact point of touching the surface is masked and estimated.

What one would want is a smooth approach and retract curve from the surface without such discontinuities. In essence what is really needed is that the tip would smoothly approach the surface monitoring the force with ultra sensitivity even at ultra short distances while monitoring optical phenomena and force simultaneously.

In a pioneering series of papers it has been realized that the best way in which to avoid these and other problems is to use the frequency of the oscillation of a cantilever rather than the amplitude. However, soft cantilevers even those like near-field optical probes that are relatively stiff (1–10 N/m) oscillate at many eigenfrequencies and have a broad spectrum. Such a broad spectrum is very insensitive to the small changes in frequency that ultra weak forces impose.

An alternate to very soft cantilevers is the other end of the spectrum and that is the use of very very stiff cantilevers that have a very sharp frequency of oscillation since the stiffness prevents other frequencies of motion. When there is such a sharp frequency spectrum there is incredible sensitivity to small changes in ultra-small forces of a surface interacting with a tip. This method of feedback is called Frequency Modulation.

An ideal example of a stiff device that is even self-oscillating (since it is formed of a piezo material) is a tuning fork which has very sharp resonances. The sharper the resonance, the sharper the Q or the quality of the oscillator or Q factor. Correspondingly, the higher the Q the higher the sensitivity to any external perturbation such as surface forces. Thus, Frequency Modulation with tuning forks is ideal for force spectroscopy.

This has resulted in some spectacular successes [59–61] and have validated the fundamental force limits of a tuning fork that were estimated to be less than a pN over a decade ago [62].

The utility of tuning forks in near-field optics was due to the pioneering efforts Karrai [30]. These efforts focused on how one may apply tuning forks to straight near-field optical glass probes. For such probes the tuning fork was placed with its axis normal to the surface under investigation (see Figure 3). Such a configuration was used with a driving oscillation that monitors the shear force or frictional force of the sample.

Figure 3

Shear force tuning fork/fiber probe geometry [63].

Many instruments were based on this elegant work but in terms of both force sensing and generality of application this approach was not ideal. In terms of generality of application this geometry perturbs the optical axis from above and makes it quite difficult to work in liquid. In terms of force sensing shear forces are perturbing to surfaces and are less sensitive and more difficult to account for theoretically. On the other hand cantilevered near-field optical probes can be combined with the ultra high sensitivity using a normal force geometry as shown in Figure 4.

Figure 4

A cantilevered NSOM fiber probe mounted in a normal force geometry for ultra sensitivity of force even with stiff probes.

This geometry allows for the ultimate in force sensing and of greater importance can fully implement the pioneering theory of Sader and Jarvis that has shown theoretically that it is possible to derive accurate formulas for the force versus frequency in such Frequency Modulation feedback methods [64]. These formulas relate the change in Q which is related to the frequency change to the force.

Recently, Kohlgraf-Owens et al. has shown that the force of a photon can be measured with such glass probes and normal force tuning fork. A force of 1.6 pN was detected for the force of a photon. This is close to the ultimate in force sensitivity predicted by Grober et al. [64]. This latter work has been extended to use force for imaging the photon distribution in the near-field [57] and also see article by Dogariu in this issue. Such an elegant approach that results from the ultimate sensitivity of the tuning fork has many applications as noted by the author especially in the area of a new detection methods for imaging vibrational absorption in the infrared in an apertureless fashion.

In addition to the tuning fork attributes noted above there is an additional characteristic of tuning forks that is critical for achieving the ultimate in force spectroscopy. To appreciate this attribute consider, once again, that the Q and the associated frequency is used as the feedback. Thus, as the tip attached to the tuning fork approaches the surface the Q changes and this is monitored as a smooth force distance curve where there is no jump to contact or adhesion ringing. Along with the frequency change the tuning fork is of course being modulated in this Frequency Modulation scheme. As a result of this, the tuning fork amplitude of course can be monitored in a separate channel unrelated to feedback. At the very point at which the tip touches the surface one experimentally sees the amplitude change as diagrammatically illustrated in Figure 5.

Figure 5

An amplitude versus distance dependence seen simultaneously with the frequency alterations of the tuning fork as it approaches and retracts from the surface.

This diagram shows an Amplitude vs. Distance graph. As the probe approaches the surface from say the right on the graph, from I, the point at which the slope abruptly changes, T, is the point of contact without jump to contact. This point can readily be verified by taking the intersection of straight line I T and the straight line TM. At M one has the point of greatest penetration. Similarly the point of retraction from the surface without ringing is readily and smoothly determined by the intersections of MR with the straight line RF.

Obviously the forces are related by Sader and Jarvis [63] to the monitored Q at T, M and R which are F^T, F^M and F^R. The change in the slope of the amplitude distance curve gives the point of contact and release of the probe tip to and from the surface. Thus, one knows experimentally the point of contact and this gives all of the parameters for accurately determining the forces even 1 nm from a surface and for calculating with experimental parameters Young’s modulus. Of course one also gets energy dissipation of the probe on-line from the change in amplitude at each pixel without any need for calculation.

The above is opposed to beam bounce feedback where the amplitude is integrally tied in with the feedback and there are additional problems of jump to contact and adhesion ringing instabilities. This combination of interrelated effects require on-line digitization of the force curve and subsequent deconvolution from the above complications and even then one gets only an estimate of the amplitude and the point of contact.

Of critical importance to NSOM is that the NSOM probes have relatively large force constants (1–10 N/m) which precludes applications with beam bounce amplitude feedback methods. Tuning forks with their close to 19,000 N/m force constants readily accommodate such force constants while permitting the ultimate in force sensitivity and permitting ultraclose approach without jump to contact. This readily permits such applications as energy transfer between a tip and a surface that was initially attempted using tunneling NSOM feedback [65].

With no jump to contact tuning fork feedback one can now switch on-line using an NSOM probe or for that matter any probe from force feedback to tunneling feedback. Shown in Figure 6 is the same atomic step on a highly oriented pyrolytic graphite (HOPG) surface done sequentially with force and tunneling using an etched gold wire probe similar to the one described above within the context of fsec pulse propagation. Why is this important for NSOM? First tunneling to a gold surface has been shown to be an effective method to enhance Raman of molecular species placed on gold [66]. Second tunneling is a known method to investigate plasmonic optical effects [67]. Third tunneling is a good method to induce carriers in semiconductors. Thus the ability to switch between AFM and tunneling in the same probe is most significant especially within the context of the discussion of multi probes in the next section of this article.

Figure 6

Monitoring the same atomic step in highly oriented pyrolytic graphite alternately with AFM and Tunneling.

The great utility of tuning forks has even been realized by the biggest AFM silicon probe manufacturer, Nanosensors (see link: http://www.nanosensors.com/news_10112008.html). In its website Nanosensors rightfully says that these new tuning fork probes are “>>for creating a new generation of scanning probe microscopy (SPM) instruments.”

Nonetheless, commercial manufacturers have been reticent to embrace tuning forks since a way had not been found for making them multifunctional and using tuning forks in aqueous conducting media. Over the years this problem has been solved by coating the tuning fork with a mono molecular layer of pyrelene. This has opened the great utility of tuning forks generally to a variety of problems in biology where the ultimate in force sensitivity is required.

In summary, normal force tuning fork feedback is the ideal feedback for NSOM for numerous reasons. These include:

• The ultimate in force sensitivity allowing either single atom manipulation [59] or single photon detection [60] or apertureless force based detection for imaging in a broad spectrum of wavelengths including the mid infrared [64].

• No jump to contact instability giving the ability to investigate optical effects even within one nanometer of a surface.

• No adhesion ringing.

• Direct experimental determination of energy dissipation in the sample by monitoring the amplitude of the tuning fork which is independent of the feedback based on frequency and thus requires no computational algorithms to image energy dissipation.

• The only feedback method that experimentally and instrumentally without computational estimation gives the point of contact with the surface.

• Obtaining all parameters experimentally for providing Young’s modulus using presently available equations such as those of Derjaguin et al. [68].

• Allowing for on-line switching between force and STM feedback.

• Overcoming the problem of the high force constants of NSOM probes.

• Having higher Q factors and correlated force sensitivity in liquids by several orders of magnitude over those achieved in beam bounce feedback even with NS0M probes.

• Having ultra sharp frequency spectra for lock-in, homodyne and heterodyne operation with no positional jitter in the Z axis.

• Allowing for exceptional interferometeric operation.

• Having no background light from optical feedback.

Thus, the turning fork fulfills fully the mandates noted in the beginning of this section of generality, low noise both optical and mechanical and the ultimate in force sensitivity. Furthermore, as will be discussed in the next section it also fulfills other required characteristics of simple optical integration and makes multiprobe operation readily possible.

8.1 The scanners

Thus, far in this instrumental perspective we have concentrated on probes and feedback as would be most appropriate for NSOM. We now speak about scanning mechanisms.

An oft employed scanner solution for fiber probes and for that matter all AFM systems are cylindrical piezo devices which are standing upright. These block lenses from above for reflection NSOM which in any event is complicated with either straight fiber probe solutions or silicon cantilevers with holes.

Figure 7

Cylindrical piezo scanners in a generally applied standing geometry [69].

To overcome these problems and to deal with the issue of extent of Z scanning while having a geometric profile that is amenable for optical and electron/ion optical integration, our laboratory invented one of the first reported flexure scanning mechanisms [32]. This is shown in Figure 8. Specifically these scanners use 4 such cylindrical piezo elements that are lying flat. This resolved the problem of having a large Z range which is critical in optics without affecting the X Y range of the scanner. This occurs since the motion of such piezos against the piezo axis is large while the axial extension of the scanner is very small. It also accomplished this in a very small form factor and a free optical axis. Thus, these ultrathin <7 mm scanners allow for a form factor that permits unique compact SPM geometries. An example shown in Figure 8 was the first probe and sample scanning SPM and had such an ultra compact geometry due to the form factor of these scanners. The importance of probe and sample scanning is seen in several papers in this issue. A particular interesting example is the use of such a system in an electron beam (see paper by Haegel on page 75). As noted in original work by Haegel with this geometry the platform “provides capability for independent scanning of both sample and tip. The electron beam is incident in a fixed location and the light is collected through a fiber probe, operating simultaneously as the tip for atomic force microscopy and near-field optical collection. The unique open architecture is critical to the ability to scan the collecting probe, while keeping an independent generation source fixed at a point of interest on the sample” [70]. Using these attributes of this combination light generated by carriers in this semiconducting material by the electron beam could be imaged with ultra high spatial resolution and full correlation with topography of even low profile structures difficult to profile in the SEM. In essence this provided spatially correlated ultra high resolution cathodoluminescence of the structures discussed.

Figure 8

Ultra flat scanning mechanisms with large Z and XY range with free optical axes together with a probe design that permits a free optical and electron/optical axis from above or below using tip and sample scanning on-line.

Shown in Figure 8 the generality of the solution allows for even dual optical microscope operation and this is important for permitting both transmission, reflection, collection and illumination mode NSOM in the same platform.

As noted together with the cantilevered nature of the NSOM probe the combination has a free optical or electron/ion beam axis and readily permits insertion into many different types of spectroscopy and imaging systems. This includes standard Raman systems and SEMs as seen in Figure 9 and also permits (see Figure 10) multiprobe geometries in ambient and ultra low temperature (10°K) multprobe AFM/NSOM based probe stations. The latter have applications in many areas including studies in plasmonics at low temperature, photoconductivity at low temperature of 2D crystals such as graphene, MoS2, WS2 etc. Even these low temperature platforms have unencoumbered optical axes for transparent integration with other imaging modalities and can incorporate high field super-conducting magnets.

Figure 9

Examples of integration of these platforms in diverse imaging tools including Raman (left) and SEM (right).

Figure 10

The combination of cantilevered glass based probes, normal force tuning forks and unique scanning mechanisms allow for multiprobe operation with a completely free optical axis for integration of all such platforms in many imaging and spectroscopic imaging modalities. Shown in this figure above is an ambient multiprobe design (left) and a cryogenic version (right) of these AFM/NSOM based multiprobe stations.

The multiprobe platforms described in this article and developed as part of the research efforts of Aaron Lewis have been used in numerous experiments. Examples of such experiments are described in references [70–72] and are also described in this issue in the paper by Berezin et al. The goal of this article was to introduce the papers in this issue of Nanophotonics and to place them within the breadth of investigations in near-field optics. We end this overview with an example of a measurement sequence completed in our laboratories with the platforms that are based on the principles indicated in this paper. The example (in Figure 11) is chosen to highlight a super-resolution imaging task that near-field optics, with its generality, readily and uniquely achieves. In this task a v-grooved quantum wire laser emission is imaged by collection mode NSOM relative to the on-line AFM [73]. The correlated images accurately place the optical emission relative to the structure at the apex of the v-groove. They allow delineation of the offset of the optical distribution (Figure 11A) with the structure that is accurately positioned by the AFM on the flat facet of this laser. These measurement tools also permit full spectral correlation of the imaging to give the mode structure of the laser from the light collected and displayed in Figure 11C and D at 805 nm and 805.8 nm respectively. No other technique could detect this mode structure alteration with wavelength. The resolution clearly approaches 50 nm. A 0.5 mm bar is included for comparison since it is the resolution normally achievable in a standard optical microscope. Note that at 0.5 mm standard resolution only a very few points over the optical distribution would have been distinguished. Finally, the multi dimensional nature of these probe stations allow for comparison of the optical and the temperature distribution. The results of the two images show that the temperature profile is less connected with the optical distribution but rather is associated with bowing toward the p contact in this quantum semiconductor laser structure.

Figure 11

Multidimensional nanoptical, nanostructural and nanothermal characterization of a v-grooved quantum wire laser. (A) Collection mode NSOM. (B) AFM structure of the polished surface of the laser facet. (C) and (D) The mode structure of the laser at 805 nm and 805.8 nm respectively. (E) and (F) Comparison of the optical and thermal distribution of the laser facet indicating that the heating arises from phenomena other than the optical distribution.

In summary, as can be seen the advances in NSOM both in its instrumentation and its penetration into numerous areas of science and technology bode well for a continual series of important contribution in many of the areas of future scientific achievement.