1. Introduction

Coherent scattering phenomena have long been exploited in a range of optical devices, perhaps none so elegantly as the Fresnel lens where both phase and amplitude zone plates have been used [1,2]. These tailor the properties of the material such that refraction occurs in such a way as to give optimal constructive interference (and hence diffraction) at the nominally chosen focus. In optimised versions where the surface profile is graded to accommodate the entire phase distribution of a Gaussian input field, the transformation efficiency is often 100%. It is therefore not surprising, given the standard lens analogy of a conventional fibre often used in text books, that one might consider an extension of such a lens to infinite thickness as forming the premise for a new class of optical waveguide, called the Fresnel waveguide [3]. The principle of coherent superposition is a universal one and conceptually extremely simple. Theoretically, coherent superposition has been examined in detail in attempts to overcome free space diffraction of laser beams and other light sources. The difficulty has been how to generate the multiple sources required in practice to demonstrate such phenomena. An axicon has been used as a linear Fresnel zone plate in this manner to generate closely related finite Bessel modes for example [4]. An enabling insight has been the recognition of the waveguide optical mode as analogous to a localised free-space electric field such as that of a diffractionless optical beam [3,5,6]. On this premise, the Fresnel waveguide configuration based entirely on coherent scattering was envisaged and demonstrated [3,7,8]. For practical demonstration, rather than involve rings of varying graded refractive index as suggested for the future development of Fresnel fibres [3,8,9], it involves the extension of radially distributed scattering sites positioned on the virtual Fresnel zones of the cylindrical waveguide along the entire length of a silica fibre tens of metres long. The technology for being able to make such long holes was first demonstrated by Kaiser et al [10] and subsequently used by Cregan et al. [11] to propose and demonstrate a special class of fibres, so-called photonic crystal bandgap fibres, using pure diffraction to generate propagation. It removes the need for complicated and high precision dopant tailoring of conventional fibre fabrication. Thus we were able to use the same fabrication technology to demonstrate a generic waveguide based on scattering phenomena and in doing so it was recognised that the Fresnel fibre in fact embraces these other special structures. Indeed, we first recognised the importance of scattering phenomena from the cylindrical interfaces and the relationship with such parameters such as loss and coupling efficiency in both low and high index cores [7,8]. Resonant-like scattering, in particular, within such dielectric media will influence heavily the efficiency of the waveguide in confining light and allowing propagation without leakage. This resonant scattering is analogous to Mie scattering traditionally associated with spheres [12]. Further, by recognising the underlying principle it was predicted and demonstrated that coherent superposition can be extended beyond the waveguide into free space. We showed, to our knowledge, the first ever waveguide structure where field spreading by diffraction is overcome at the output. Focussing was obtained -multiple foci were observed: the analogy with Fresnel optics is confirmed [8]. In this paper we review the results in somewhat more detail and bridge together some of the ideas, implications and applications that arise from them.

2. The Fresnel fibre

Since the simplest fibre is of cylindrical geometry it may be concluded that the optimal index distribution for obtaining a propagating mode with a peak intensity at the centre based entirely on coherent superposition of scattered light is a radial one about the centre of the cylinder where the main propagating axis might be. From this logic the phase reversal characteristic of zone plates should correspond closely to the positioning of the boundaries of the refractive index variations at the zeroes of, say, the ideal mode of a typical step index waveguide. The natural wave solution for a cylindrical waveguide is a Bessel solution, the simplest being one of the first order, J₀ . For the Fresnel cylindrical waveguide the intensity follows an Airy-like distribution: I ≈ I₀[2J₀(r)/r] ² where r represents the radial position of the field within the waveguide. It is informative to note that Bessel solutions, analogous to the ones solved in free space for ideal non-diffracting waves, exist in optical waveguides because of the confinement principle balancing the diffraction of the mode. Consequently, this balancing act between the physical method of confinement over the tendency of an optical mode to diffract in free space, is appropriately referred to as a soliton-like solution [3,8,9] despite the traditionally linear view of this particular problem. Further, the possibility of using coherent superposition from multiple source generation as the principle means of overcoming free space diffraction entertains some extremely interesting ideas and consequences. (Note: One can read the insightful diagnosis of Snyder [13] to see that even non-linear processes can be broken down into a series of linear processes. The implicit generalisation one may therefore make is that traditional solitons are no more a simple solution of linear processes overcoming another – this greatly expands the concept of a soliton and its underlying physics, supporting the argument mentioned above of soliton-like solutions when free-space diffraction or beam spreading is overcome in a waveguide).

The optimal fabrication of such structures might lead to a radial distribution of index change, preferably graded inbetween the appropriate Fresnel zones where a phase reversal (shift) is introduced, usually in steps of 2π [1,2]. In this case a worthwhile practical parameter to determine is the critical angle of diffraction of the lens properties enabling light to propagate along extended lengths. Working backwards using standard mode field diameters, one of us was able to obtain a simple expression for this angle: θ _c = tan^-1 (1.64λ/(4${{\mathrm{w}}_{1/\mathrm{e}}}^2$)) where λ is the operating wavelength and ${{\mathrm{\omega }}_{\mathit{1}\mathit{/}e}}^{\mathit{2}}$ is the spot size [3]. For a graded Fresnel lens, the length is related to the index contrast, Δn, as L= λ/Δn (2π phase difference between zones). For a waveguide with a spot size equal to that of a typical step index fibre the required index change is ~0.03. Whilst this is in the range of typical modified chemical vapour deposition (MCVD) fabrication processes, the precision in depositing layers does not currently exist. Another method with the required accuracy in depositing layers may employ nanodeposition techniques [14]. The recently proposed utilisation of boil-off during dopant fabrication to obtain precise graded profiles and annuli distributions, in part determined by the cylindrical geometry of the preform, offers one solution. However, some research is required to perfect this method since the index contrast to date has been very low (~0.001). Nevertheless etching with hydrofluoric acid the end of the fibre was possible – this enhanced the index contrast substantially to allow Fresnel lenses to be made on the tips of standard fibres [9].

Therefore it was realised a useful index contrast with air was found to be eminently suitable for such work since the index contrast is much larger. As mentioned earlier, such an index contrast is readily available in fibre form as a result of a recently reinvigorated field focussing on the fabrication of air-silica structured optical fibre. No dopants need to be employed and in this case rather than well defined rings determining the Fresnel zones, we opted to employ air holes spaced along the virtual Fresnel zones of the waveguide. Our treatment is not dissimilar to that found in aperture based Fresnel zone plates where holes were employed in an opaque medium used to make up the Fresnel zones for lens applications in the microwave [15] and x-ray regions [16]. However, to our knowledge a transmissivebased phase plate incorporating air holes in a transmission medium (such as silica) has not been previously reported. Thus in later sections we show how such fibre “zone plates” can be used as very short lenses (~1mm or less) spliced onto standard fibre. The extension of the lens thickness to “infinite” lengths for operation as a novel waveguide also contributes to the functionality of the Fresnel lens since it demonstrates a “lens” that can be made flexible and long, enabling light to be transported around bends, for example, whilst retaining its far field capability to focus. The issue of loss over long lengths can be readily addressed in these fibres by improving the basic design we have sued to increase the air-fraction appropriately – recent demonstrations of ultra-low loss fibres (<0.5dB/km) indicate this is possible [17]. This could have significant future applications where light is collected and transported by the same structure – the use of plastics can allow large diameter versions for use, for example, in free space optical communications, significantly reducing the costs (by eliminating the requirement of bulk optics to direct and collimate light). Hence the potential of enabling such a technology to be “disruptive” – an oft-misused word - in certain communication links, including local area networks (LANs), exists. It should be noted that the long extension of these zone plates means that multiple scattering along the waveguide will also involve an angular or wavelength dependent resonant scattering analogous (though not localised to one volume of space) to Mie scattering in spheres. At the minimum critical angle of propagation, the interface Fresnel reflections between high and low index can be significant and hence resonant-like effects along the waveguide potentially strong. These larger Fresnel reflections are responsible for the low number of rings required in photonic crystal fibres to achieve efficient propagation.

3. Fibre fabrication

To simplify the fabrication process we have adopted a straightforward approach where the phase zones are determined by the ring of holes spaced at radii intervals of r_n ≅ r_n-1 + d²/2r_n-1 where r_n-1 is the radius of the previous zone and d is the radius of the outermost zone. This classical approximation for a series of rings of equal area is close to the distribution found for the zeroes of the Bessel function described above. The equation holds when the effective Fresnel lens focus is a lot greater than r₀ , the radius of the first zone. Much more efficient, and often complex, hole distributions exist for aperture based zone plates that will more than likely improve the coupling efficiency of our fibre but we chose this configuration because it is straightforward, albeit tedious, to fabricate the preform for such a fibre using glass drilling. Fibre is then readily drawn from such a preform. By controlling the temperature or draw conditions we are able to fine-tune the hole size as a function of hole collapse. Figure 1 indicates the degree of tuning possible.

Figure 2(a) shows a cross-section of a typical preform with a high index centre region (~1.45) surrounded by the low index air holes (~1.00). It is noted that the spacing of the holes appears relatively large. This disparity is enhanced after the preform is drawn into fibre form. Figure 2(b) shows the fibre cross-section (125µm diameter) where hole collapse has led to a hole spacing ten times greater than 1.5µm light. However, despite this apparent spatial incongruity we were able to sustain single-mode propagation of 1.5µm light over a straight 30cm piece of this fibre with relatively low loss (between 0.2 and 0.5 dB/cm) [7] (Fig. 2(c) shows the near-field profile). This indicates that whilst a considerable portion of coupled light leaks out between the holes at the input, some light scatters under appropriate phase conditions in the forward direction to lead to the generation of a stable mode along this fibre. At 632.5nm, the output of a HeNe laser, no mode generation is possible and the light leaked out between the holes, indicating the sensitivity to dimensions of the scattering phenomena (Fig. 2(d)). Arguments using an effective step-index approach to explain these results are not valid precisely because the interstitial hole spacing is not only commensurate with the wavelength of light, it is significantly larger. If the hole spacing is sufficiently small compared to the launched wavelength of light then the ring of holes can be treated analogously to conventional rings of dopants and an effective index picture can approximate quite accurately a number of properties of the waveguide. However, in practice, the high effective index contrast will result in multi-moded behaviour if the core area is not decreased, and hence the number of holes needs to be reduced. This results in hole spacing that may be substantially larger than the wavelength of propagating light leading to two desired effects: 1) it can readily generate single-moded behaviour for the appropriate wavelength since the next few higher order modes have a circumferential propagation vector which is not symmetrically radial around the centre of the fibre and therefore leak out more rapidly, and 2) the contribution of coherent scattering from the glass-air interfaces becomes increasingly important in determining whether propagation occurs. In these typical cases, it is therefore not reasonable to treat the surrounding areas containing holes as an approximate cladding with an average index lower than the central core region. Indeed, Fresnel optics using phase zone plates work because coherent scattering, from both interfacial reflections and edge diffraction, takes place. It is precisely the superposition of these fields that give rise to the modal properties of our waveguide. What is also interesting to note is that the position of the holes in bulk zone plates is not crucial within the zone regions in determining the focusing properties [see for example Ref. 16] – although they do play a role in determining the efficiency, or amount, of light focused to a point.

 

Fig. 2. (a) cross-section of fibre preform; (b) cross-section of drawn fibre; (c) near field profile observed at 1550nm; (d) near-field profile observed at 632.8nm.

Download Full Size | PDF

In order to demonstrate the effectiveness of this approach, we took a similar preform piece and drilled an additional hole in the centre, the objective being to show that we could generate a mode profile whose peak centre was in the middle of the waveguide [11]. However, the structure is changed and the superposition somewhat more complicated since the scattering of the central hole is also involved. Besides the obvious scientific interest, there is great practical interest for being able to generate such modes. A great deal of effort has gone into, for example, developing so-called photonic crystal “bandgap” fibres where Bragg diffraction is used to confine the light at the centre of a hole [18–20]. Likewise, the use of deposited dielectric layers to generate omnidirectional reflection along the waveguide length is also being investigated [21]. Such waveguides reduce the contribution of non-linear interactions between high intensity pulses propagating down the fibre and the host medium (in our present case silica glass) since the peak intensity is in vacuum or air (or another gas or low nonlinearity material). Further, being able to insert any material within the hole can maximise modal overlap with a chosen medium for a range of applications, including non-linear switching (enhancing the chosen non-linear coefficient in this case) and sensing.

 

Fig. 3. Cross-section of Fresnel fibre with centre hole.

Download Full Size | PDF

Figure 3 shows the cross section of the drawn fibre – the central hole is slightly off centre since it was not incorporated into the first phase of drilling and there was some error in realignment of the preform during the second phase of drilling. The air-fraction is chosen to be larger than the previous fibre. To improve confinement and reduce losses, we ensured little hole collapse in the drawn fibre, thus retaining the preform structure. The basic principle is to generate scattering between the holes such that field superposition leads to a maximum intensity inside the central hole. A schematic of this process is depicted in Fig. 4. It is the phase relationship generated between the holes that allow such constructive interference – since this relationship is of concern only in the radial direction, the system can be treated approximately as if it were a ring of Gaussian fields propagating along the waveguide around the central axis. Under ideal circumstances, where the annulus around the centre hole is made up of a continuous ring structure rather than a ring of holes, superposition of these fields leads to the generation of a super-field with a Bessel function profile. The analogy between waveguide mode fields with free space beam fields is strengthened since a Bessel beam is often described theoretically in this fashion [22].

4. Near field properties

Figure 5 shows the near field profile of the fibre using sources at three different wavelengths: 632.5nm from a HeNe laser, 1052nm from a Nd:YLF laser and ~1550nm from a broadband erbium doped fibre amplifier source. At 632.5nm the wavelength is too short relative to the dimensions of the hole spacing and it leaks out quite substantially. Because the air fraction is somewhat larger than the previous fibre (i.e. the hole spacing reduced), there is sufficient scattering of some of the light to generate a complex mode with four lobes around the central hole. It is effectively located in the high index “ring” around the central hole as one might expect for a principle based on index contrast confinement. Similarly, at 1052nm a mode is generated within the ring, although the leakage loss is reduced. The asymmetry of the fibre gives rise to much more pronounced asymmetry in the field distribution. At still longer wavelengths it is possible to remove most of the leakage loss and have a central mode whose peak is within the hole centre – this is despite the ring thickness being larger than the launched wavelength of light. At these longer wavelengths field overlap in the ring region is significant and since they are all coherent with respect to each other, constructive interference gives rise to a super mode across the centre hole. Resonant scattering of the light within the hole is also expected to contribute to the properties of the optical modes generated, particularly dispersion. Thus, we have experimentally realised the generation of a supermode based on superposition of the fields that would normally be localised to the ring region of high index. Multiple scattering properties within the holes are likely to contribute to these effects.

 

Fig. 5. Near-field profiles of Fresnel fibre with central hole at three wavelengths.

Download Full Size | PDF

5. Far field properties

Another feature of propagation achieved by coherent scattering is the extension of the principle of superposition beyond the near field into the far-field. As the various fields go in and out of phase away from the fibre end, complex interference effects are observed. For example, at 1550nm the profile changes from Gaussian-like to a ring distribution and back again, is seen before eventual dissipation occurs. Figure 6 shows the far-field profile at various positions away from the fibre end face revealing at least two effective foci of the fibre. Figure 6 (100µm) shows the first focus where the light is brought to a point with six weaker lobes around it. The intensity exceeds that of the light at the end face, indicating that waveguide field spreading at the output has been overcome. As the fields travel further out, interference leads to complex image reconstruction of the fields within the waveguide. The second “ring” focus (Fig. 6 (200µm)) shows the construction of light within the high index region of the waveguide where the light is inbetween the holes. Note the apparent π/6 rotation of the ring with respect to the six lobes surrounding the first focus point. This is repeated again at the second focus (Fig. 6 (300µm)) at approximately twice the distance of the original. This second focus has a central lobe of greater peak intensity and narrower transverse profile than the first. The π/6 shift in the reconstructed images at each point coincide between the superposed fields actually in the holes and the waveguide fields inbetween the holes. Therefore, image reconstruction at the focus is of the superposed fields that exist not within the high index region but in the low index region air holes, indicative of the role of multiple scattering phenomenon, akin to Mie resonances, in the propagation process of air-silica structured fibres generally.

 

Fig. 6. Far-field profiles at varying distance away from the Fresnel fibre end face. Image reconstruction is observed at ach plane. The white arrows denote a π/6 rotation between the various images in the far-field.

Download Full Size | PDF

The multiple image “foci” are consistent with those expected from phase zone plates and their position is approximated by ${\mathit{~}r}_{\mathit{0}}^{\mathit{2}}$/nλ where n is an integer multiple [1,2]. Unlike the recently demonstrated Fresnel lens fabricated by controlled chemical deposition and etching [9], it does not require any etching procedure or complex graded dopant distributions. Thus there is potential in constructing novel Fresnel lenses and Fresnel beam shapers for numerous applications. The phase sensitivity of this process can be used to generate or enhance numerous sensor configurations with possible applications in surface microscopy. It is anticipated that since the underlying physics of air-silica structured fibres generally relies on coherent scattering, similar complex superposition phenomena in the far-field should be observed within conventional air-guiding photonic crystal fibres. Some evidence is indicated in the observation of a π/6 shift of one previously reported photonic crystal fibre [23].

The potential for tailoring and shaping an arbitrary field profile using this method of waveguide fabrication is significant. By arbitrarily tailoring the phase profile and to some extent the amplitude profile it is in principle possible to generate multiple reconstructions of complex field structures within the waveguides in free space. This has enormous potential for beam shaping and positioning generally. When examined in 3-D space we have clearly generated an optical void or bubble where a volume of space is encapsulated in an optical field. A schematic of this is presented in Fig. 7. Such optical bubbles may have applications for example in micro- or nano- particle manipulation for a range of applications in areas such as nanotechnology and biotechnology (DNA cleaving and manipulation come to mind). The notion of wrapping a field around an object could in principle be scaled up to macrodimensions.

The optical bubble above can be used as a spatial field or phase interferometer by observing appropriate interactions with a desired measurand. Further, multiple image construction from several waveguide structures can be envisaged inbetween a point of combination to further enhance all these effects provided control of coherence is maintained (free space versions of optical devices based on interference effects are potentially realisable). Thus we have demonstrated the first steps to real all-optical manipulation in space, operating “remotely” from the generator source. Controlling the temporal and amplitude properties of the light, as well as phase, traveling along each guide, enhances it. This has the potential of significantly impacting the optical component and holographic industries. These concepts are not limited to device performance – they potentially underpin a range of phenomena hitherto unconsidered, including extending the ideas to other fields that invoke similar superposition principles. In particular there exists the possibility of generating more efficient means of controlling and extending free space diffraction well beyond the Rayleigh range.

 

Fig. 7. Representation of the optical field “bubble” generated between the two foci of the Fresnel fibre or lens. A micro- or nano- particle is caught within in.

Download Full Size | PDF

 

Fig. 8. Schematic illustration of Fresnel lens spliced onto fibre tip. Cross-section

Download Full Size | PDF

6. Micro-optic zone plates

The use of zone plates to obtain focusing has seen a major resurgence within applications particularly involved with the focusing of wavelengths of light for which there is very little practical conventional optics available, such as extreme UV and X-rays [1,16]. These wavelengths are of increasing importance as the drive for higher resolution lithography continues. Conventional Fresnel lenses of this sort are often associated only with amplitude zone plates, which consist of concentric rings of transparent and opaque material, including air and metal. However, there are also other geometric variations including the use of arrays of apertures and non-circular apertures [1,2 and refs therein]. The more recent variation at these wavelengths involved random hole distributions in a metal within defined zones [16]. This is a stochastically varying approach to the same solution found for microwaves [2]. Thus there is no doubt that there are defined applications in the microwave and short wavelength regions that benefit from the continued evolution of this technology.

In addition to the traditional Fresnel zone plate arising essentially from interference effects due to diffraction of light from slits of various distribution profiles, there are phase zone plates made up of concentric rings of, for example, air and other transparent media where the phase retardation of the optical field in these domains allows appropriate constructive interference of all the light at the focal point [1,2]. The Fresnel fibre described above is an example of a similar device using cylinders of air as apertures in a transmissive medium. Both types of zone plates, amplitude and phase, are generally characterised by several phenomena, including the existence of multiple foci and wavelength dispersion. These features are also characteristic of the Fresnel fibre. However, it is worthwhile noting that these zone plates can be modified at the boundaries by non-discrete edge topologies, such as a sinusoidal surface variation, to generate only one focus [24].

A 1mm section of our Fresnel fibre is spliced onto the end of standard single mode optical fibre (SMF28). Figure 8 shows a schematic of the device. The ease of splicing combined with straightforward fibre fabrication processes to generate hundreds of metres of Fresnel fibre can lead to mass production of cheap lenses. In this case we used a fixed wavelength from a tuneable laser source to characterise the device. The near field properties are somewhat different to the case when a long length of fibre is used. Both the near and farfield profiles at 1510nm are summarised in Fig. 9. It is observed that the asymmetry of the profiles appears more significant. By scanning the microscope objective such that the imaging plane is well inside the fibre section itself it is possible to crudely image the extent of the field profile at that point inside the waveguide (also taking into account the amount of light confined by the fibre beyond the image plane). With the long length of fibre, when the coupling is appropriate it is observed that there is no change of the mode with peak intensity inside the hole. However, over a short length of 1mm where the light coupling into various leaky states is not filtered out before probing, the field within the fibre was found to vary between ring and focus. Coupling is also not matched perfectly in this case as a result of the splicing an asymmetric Fresnel fibre to a symmetric conventional fibre. Within experimental uncertainty, it appears that the near-field is defined by a ring profile for this particular lens, indicating that the propagating solution is sensitive to coupling.

 

Fig. 9. Field profiles within, at the end and in the far field of the Fresnel fibre lens at 1510nm.

Download Full Size | PDF

The contribution from dispersion to the case where an EDFA is used, was determined by examining the performance of the lens at a few wavelengths spanning the EDFA spectrum. Figure 10 summarises these results. Initially the position of the reconstructed images are all identical at all wavelengths. The image position, ƒ _n , is approximately described by relationship, ƒ _n ~ nƒ. ₁ where n is an integer multiple and ƒ ₁ the position of the first focus point, which is close in agreement with the classical Fresnel lens formula for concentric rings: ƒ _n${\mathit{~}r}_{\mathit{0}}^{\mathit{2}}$/nλ. Further away from the end face, however, the distance between foci increases and there is growing difference in this position between wavelengths. At this stage the intensity is dropping rapidly and the light slowly diverging away (Fig. 9). Despite dispersion becoming noticeable at further foci, at practical working ranges available to the first two foci, there is no significant change in focus across the wavelength span shown. The increasing disparity further away may be useful for applications such as dispersion compensation. Alternatively, this form of spatial sensitivity to wavelength at greater distances could be used as a novel and simple spectrum analyser.

7. Discussion and conclusions

 

Fig. 10. Position from the end face of the Fresnel lens for different wavelengths from a tunable laser source. The field within the lens is taken only at 1510nm.

Download Full Size | PDF

We have demonstrated a high-index core waveguide such that propagation at longer wavelengths is possible but not at shorter wavelengths. The fibre design chosen is based on optimising the size and position of holes to so-called Fresnel zones of a cylindrical waveguide such that the interstitial hole spacing can be significantly larger than the propagating wavelength. Propagation in this regime will be highly dependent on coherent scattering.

In addition, we demonstrated a similar waveguide with a higher air-fraction and a central hole surrounded by a high index region. This particular fibre has a propagating solution in the central hole at longer wavelengths but not at shorter wavelengths, where propagation in the “ring” is preferred. From the far-field data reconstruction of both the ring fields and the hole peaks are obtained at various distances. There is direct evidence of image reconstruction of light within the low-index holes as well as the central hole, indicating the importance of resonant, or multiple, scattering within the cylinders in determining the properties of the wave guidance in our fibres. This effect is optimised in terms of coherent superposition when there is a Fresnel (or Bragg in some cases) condition satisfied. We note that this resonant phenomenon also underpins the classical ARROW waveguide where there exists propagation in a low index region surrounded by a ring of high index medium [25]. A recent analysis of photonic crystal fibres by analogy with ARROW waveguides has also indicated some correlation between the two [26], further supporting the classification of photonic bandgap fibres within the Fresnel waveguide umbrella. Combined with the observation of π/6 rotation of the fields in free space, this contribution underpins the constructive interference effects leading to a peak optical field within the central hole. However, over short lengths such as that used by our lens, this contribution adds to that obtained by guidance within the high index region and hence we observed both ring and central point profiles propagating over short lengths at 1550nm when the image is probed within the waveguide. Over long lengths we could only observe a dominant centre profile where interference leads to the peak intensity within the hole and not the ring. We did not observe a ring profile within the fibre.

Clearly, the ability to tailor single and multiple scattering by controlling the size, position and distribution of the cylindrical (though they obviously need not be circular in cross-section nor uniform in size) interfaces enables us to design complex optical waveguides and optical components. Extremely refined control over properties such as dispersion is possible. By way of example, we have demonstrated the Fresnel fibre, which utilises the generic waveguide propagation principle of coherent scattering. Further, a new class of micro-optic phase zone plates operating in the visible to near-IR were demonstrated. The sort of technologies, which can be readily incorporated into a subsystem or system using standard technologies such as fibre splicing, can be fabricated cheaply and in bulk. They offer, for example, a competitive lens alternative to current micro-optical elements such as GRIN and ball lenses. In addition, they also offer a way of reducing losses in interconnects involving Fresnel fibres and compact photonic crystal circuits or devices both in tapered and untapered forms.

From a fundamental viewpoint, the analogy between free space localised optical fields and waveguide fields recognizes the generic commonality between the two field representations. In this instance, it has enabled us to understand wave guidance generally and to invent some novel forms of waveguides. Likewise, the waveguide itself has enabled the control and manipulation of optical fields in free space (i.e. in the far field). We have done this with one waveguide only, though the ideas scale to combinations of waveguides all with correlations in phase space that allow future sophisticated manifestations of the optical field structure. The imposition of time (for example with pulsed light or switching) can allow a dynamic restructuring of optical fields structures for numerous applications including holography and communication. It is now conceivable that many of today’s optical functionality in waveguide circuits could be achieved in free space using such architectures thereby negating some of the complex and costly fabrication processes involved with photonic circuit design. This would be particularly important for 3-D circuit functionality that has not yet been practically demonstrated. Playing around with interference effects can also remove material considerations for local switching and a new era where optical devices with no matter are involved (at least in the immediate vicinity) may come to fruition.