J. Canning,     University of Sydney, Australia

S. Bandyopadhyay,     Central Glass and Ceramic Research Institute (CGCRI), India

9.1 Introduction

Silica remains the heart of nearly all modern optical transport systems. Engineering of silica in its various forms ranges from one-dimensional (1D) to three-dimensional (3D) waveguide and periodic structures, including recent interest in 3D photonic crystals. Most of the preparation methods involve complex vapour deposition and various co-dopants, which have an advantage of overcoming the lack of finesse involved with general formation of glass structure through high temperature processing and quenching. Nevertheless, to obtain micron or submicron precision over the processing of glass, post-processing methods are often used, ranging from lithographic etching of systems with dopants to laser processing using UV to mid-IR lasers, through to tapering and drawing of patterned preforms and capillaries into nanostructured fibres and wires. Concrete examples of laser-based micron scale processing of glass include direct written waveguides, Bragg gratings in waveguides and optical fibres and photonic crystals. The drawback with non-thermal post-processing techniques is that, to date, they generally produce glass that is structurally less stable than the starting phase; this includes that produced by femtosecond laser-induced melting and solidification. The relationship between glass stability and the rate of freezing in a structure warrants further investigation.

In contrast, a great deal of control of glass properties is achieved with simple thermal processing, varying both the softening and quenching rates of the glass. By simple processing, a striking range of properties can be achieved with just pure silica alone – controlled quenching of liquid and partial liquid phases can produce a variety of polymorphic phases of varying density and structure. In fact the common distribution of crystalline states of more than a quarter of the Earth’s crust is strong evidence that the Earth has cooled down slowly. By comparison, rapid quenching leads to a potential continuum of polyamorphic states – these need not be distinct but can span a significant density range. Rapid quenching is also characteristic of chemical precipitation. Recent discoveries, by one of NASA’s rovers, of amorphous silica on the Martian surface would suggest a surprising amount of chemically precipitated silica given the high hydrogen content and amorphous nature (http://www.sciencedaily.com/releases/2007/05/070521154427.htm) (alternatively, rapid thermal quenching of an eruption could have also prevented crystallization, potentially indicative of a distinct thermal history on Mars). This extraordinary range is, within the context of our planet and our brief history, unique to silica and arises principally from the topological rearrangements possible from a simple, but rigid, tetragonal structure of four oxygen atoms and one silicon atom. On a local scale the structure is paradoxically highly ordered whilst on a longer scale that order disappears rapidly into a so-called random network. The rigidity (which tends to lead to large ring collapse within silica) and flexibility of the structure (which is critical for the very low propagation losses enabling the wiring of the internet) can be further expanded by the addition of small dopant quantities, materials which coordinate reasonably well, although the number of defect states rises dramatically. It is no wonder silicon and silica mark the photon epoch we have just entered.

Despite the extraordinary versatility thermal processing offers, very little work has evolved to be able to control it at a dimensional level. There are only a few examples of this, a particularly striking one being the exploitation of two-photon excitation, normally a problem in the semiconductor industry, into the band edge of silica to achieve densification and damage of glass to write Bragg gratings in silica core (Albert et al., 2002) and allsilica (Groothoff et al., 2003) optical fibres. Extending this to higher multiphoton processing, often involving very rapid and localized quenching quickly freezing in large index contrasts, was made possible with the advent of readily available femtosecond lasers. These impressive achievements suggest lasers are one way to go in achieving highly localized, and rapid, glass transformation potentially offering a unique finesse in thermally tuning change compared to conventional radiative or convective heating. Band edge absorption relies heavily on the subsequent thermalization that occurs with localized excitation. A review of these photo-writing processes can be found in Canning (2008). More recently, we have considered the concept of attempting to localize thermal processing on a scale that is unprecedented by combining the localization possible using lasers with thermal processing (Bandyopadhyay et al., 2008; Canning et al., 2008–2010) following earlier work (Canning, 2004; Zhang and Kahriziet, 2007). In effect, the process of ‘regeneration’, carried out in optical fibres but certainly not restricted to them, involves patterning a seed structure (in any dimension – 1, 2 or 3D) using a laser, perhaps through defect excitation, to effect a difference between one region and another. In so doing, the changes in local structure and stress between regions will lead to a subsequent competitive struggle for change when the whole glass is heated. This competition arises principally from different relaxation rates as a result of changing glass structure by initial irradiation and varying stress fields between such changes. The general concept of patterning the relaxation of glass on a local scale using lasers to achieve localized change was proposed by Canning (2004). The possibility of glass transformation as the template for ultra high temperature gratings and ultra high temperature micro and nano processes was confirmed when we first saw the work of Zhang and Kahriziet (2007), an observation of regeneration in just standard photosensitive fibres. Subsequently, we noted similar work predating this from Juergens et al. (2005) – where Zhang and Kahriziet (2007) did not ascribe a mechanism, Juergens et al. (2005) assumed a so-called ‘chemical composition grating’ based on oxygen diffusion, despite some contradictory evidence. Instead, based on the results of Zhang and Kahriziet (2007) we recognized that what we had previously predicted was demonstrated in fibre form: spatial patterning of structural relaxation in glass and that this regeneration approach was one way to achieve this. Hence, we were able to optimize the annealing process to demonstrate strong regenerated gratings (Bandyopadhyay et al., 2008, 2011; Canning et al., 2008, 2009, 2010).

Given the immediate relevance to nanophotonics and the potential of nanoscale processing to produce sophisticated components, together with the growing area of ‘extreme’ photonics where high temperature operation is regarded as increasingly important, we review that work to show that the regeneration retains fully the seed structure in optical fibres with nanoscale resolution.

9.2 The regeneration process

In reviewing this work and its broader implications, we focus on our work in optical fibres where the picture may appear complicated somewhat by the presence of germanate doping and core–cladding stresses – however, so long as these are exploited correctly the underlying principle is identical. That is the germanate is essential for the fabrication of very strong seedgratings through conventional fibre Bragg grating writing; in the presence of hydrogen these seed gratings are proportional to the generation of OH via defect excitation and require no significant contribution to the polarizability change via densification to account for the local induced index change (although clearly this may be present). What is remarkable about hydrogen is that simple physical diffusion of the gas into the fibre alone leads, through hydrogen bonding and OH formation, to significant internal pressures that help reduce tensile stresses produced during optical fibre fabrication which can improve the size of the induced index change. This can significantly alter the distribution of relaxation processes, usually described by the relaxation term, β (Angell, 1995), since the local structural change and the surrounding stresses are altered differently. In contrast, when hydrogen is not used, thermal processing leads to a rollover in index change regenerating gratings that are identical in all aspects to type 1n (or type IIa) gratings obtained by continual laser processing (Lindner et al., 2009, 2011a) – this is very strong evidence that the formation of type 1n (type IIa) gratings is, in fact, thermally driven. Laser-based annealing becomes a very potent tool if examined in this light. The very slow formation rate along with the much lower annealing temperature suggests crystallization of the germanate glass rather than silica; it is dopant dependent.

By contrast, an important systematic study was reported using fibres with varying dopants to show that dopants do not directly have a significant contribution to the final regeneration when hydrogen is present (Bandyopadhyay et al., 2011), as was first predicted if the mechanism of regeneration was based primarily on patterning the relaxation kinetics of glass (Bandyopadhyay et al., 2008). Figure 9.1 summarizes a general plot of the obtained regenerated normalized reflectivity versus germanate concentration for seed gratings (all ~ 47 dB) in four types of fibre (SMF 28; two specialty fibres produced at CGCRI in India – NM113 and NM41 – and a UV photosensitive fibre UVS_EPS); in one fibre fluorine is contained in the cladding to see if these regenerated gratings are related to the ‘chemical composition gratings’ of Fokine (2002–2004) and Trpkovski et al. (2005), which are described in terms of chemically assisted diffusive mechanisms.

Interestingly, it would appear that there is an inverse relationship between regenerated grating strength and germanate concentration. This trend immediately rules out germanate as being directly involved with regeneration beyond determining the seed grating strength – in fact, one might mistakenly assume it is deleterious to the process. However, when taking into account the numerical aperture (NA) (V parameter) of the fibre, and based on a model where the structural change takes place in silica and therefore most likely occurs at the core–cladding interface where stress is highest, the dependence of germanate is negligible and can be ruled out. Most interestingly, some support for this interface model is the deleterious effect of fluorine, present in the cladding layers of one fibre (NM41). This has a significant effect in reducing the regeneration and probably can be explained by diffusion of fluorine, which happens readily at these temperatures, leading potentially to a reduced fringe contrast. These results directly contrast with that expected from so-called ‘chemical composition gratings’ originally thought to arise from fluorine diffusion (Fokine, 2002) and later modified to enhanced fluorine-assisted selective oxygen diffusion (Fokine, 2004, 2009) (after suggestions by others that fluorine could not be responsible but rather oxygen partly based on stoichiometric, diffusion and structural considerations). An oxygen diffusion model was also taken at face value by Trpkovski et al. (2005) and Juergens et al. (2005), although the latter did point out conflicting results. It appears likely that the origin of these diffusive approaches can be traced to the work on valid thermally induced diffusion gratings (varying ‘chemical composition gratings’) demonstrated by Dianov et al. (1997) using localized heat sources (> 1000°C) such as CO₂ lasers to fabricate long period gratings. Although it is unclear how uniform heating of a standard grating, chemically activated by periodic irradiation of fluorine or OH formation, can generate such distinct and stable periodic structures based on chemically assisted diffusion alone at lower temperatures, the logical extension is that any dopant can be used to create regenerated gratings. Our results are inconsistent with such a mechanism as the primary cause of change – in contrast, our mechanism which is generic and not dependent on the local chemical details but rather glass structure, does explain the results and is based on a simple extension of well-known glass kinetics and relaxation (Canning, 2004) so it almost certainly must be able to occur.

It seems clear that regenerated gratings do not follow the type of behaviour expected solely based on diffusion. Further, whilst diffusion obviously can occur at these temperatures, such diffusive models would lead to fringe decay, often uneven, and therefore changes in optical phase as the index also changes in between the periodically laser processed regions amplifying non-uniformities. It is difficult to rationalize the results obtained in high NA germanosilicate optical fibre annealed at 1295°C (Canning et al., 2009), where the fibre core is soft, and at even higher temperatures until 1400°C was exceeded (Åslund et al., 2010) and whether it would be possible if the changes were dependent on diffusing dopants (since there is no reason why diffusion should stop at these higher temperatures). The evidence would indicate that dopant diffusion, whilst probably present, is not critically involved and that the major changes are associated with silica structural phase changes probably at the core–cladding interface. Since silica can operate at these temperatures, the exclusion of dopants in the first instance makes the system a very simple one to understand in terms of conventional glass processing as we have proposed (making the optical fibre potentially a wonderful laboratory for studying glass and its changes optically). From a fundamental science perspective, the broader picture we arrived at was that it is possible to tailor and pattern any amorphous material, with either laser or thermal systems or indeed any alternative means, by patterning its local relaxation kinetics and thermodynamics, either 1D, 2D or indeed 3D (enabled by multiphoton absorption processing such as those possible with near-IR femtosecond lasers). Practically speaking this is also extremely exciting – the higher temperature resistance of silica, in combination with laser processing conditions using dopants, means that potentially high resolution, nanoscale and thermally robust patterning of glass is possible, as originally predicted (Bandyopadhyay et al., 2008). Interestingly, after our original work, so-called chemical composition gratings were redescribed in terms of changing the local structure (through diffusion) to achieve differing relaxation profiles between processed and unprocessed glass (Fokine, 2009). Although limited to a single relaxation process, the description in terms of relaxation is identical to what we have been saying (Bandyopadhyay et al., 2008; Canning, 2004) and is an acknowledgement that nanoscale processing of glass will be critically dependent on the physical picture we have been espousing. Whether chemical composition gratings are indeed identical to high temperature regenerated gratings, but perhaps described in terms of details that are independent of the larger picture of glass transformation and consequently may not be optimized appropriately, may be an interesting topic for some readers to pursue – given some inconsistencies with our experimental results, we have continued to assume the two gratings to be distinct. More importantly, the general consensus is that a model based on glass transformation of the type we first suggested is probably correct. Undoubtedly, the opportunity for exploring new routes to such transformations will herald significant results not only in optical fibres but in two- and three-dimensional photonic waveguides and devices, for potential use in extreme environments.

In summary, the dopants are therefore important in inscribing the initial seed grating since there is a direct correlation with regeneration and the seed grating strength, not surprising given that it should also correlate with the variation in local relaxation rates. Further, the retention of phase information (Canning et al., 2009, 2010) throughout what would otherwise be regarded as substantially harsh thermal processing, confirms nanoscale processing is possible. More immediately, we may conclude that if this is the case, then gratings that are regenerated should be able to retain all the complex phase information of the seed grating if thermally processed uniformly. This could have extremely important ramifications since in principle ultra high temperature stable complex patterning in glass can be achieved, something that is increasingly important in a number of industries, including semiconductor optical lithography and ultra high power fibre laser filters and reflectors. Further evidence decoupling the role of hydrogen in regeneration completely from the initial seed grating is the recent demonstration of posthydrogen loading regeneration of seed gratings written with no hydrogen (Canning et al., 2011; Lindner et al., 2011b): the presence of hydrogen leads to regeneration that appears identical to that so far reported, far exceeding the thermal stability from type 1n (type IIa) – like regeneration (Lindner et al., 2009, 2011a). In addition to its scientific importance, this decoupling allows for a significant expansion of the regeneration process to gratings that are produced without hydrogen, including on-line draw tower gratings (Lindner et al., 2011b).

9.3 Estimating the retention of nanoscale information in regenerated grating structures

In order to determine whether this process can be applied beyond simple Bragg grating writing as a realistic approach to the production of complex gratings and patterns and structures that can operate at high temperature whilst retaining the complexity of a nanoscaled device, the impact of the regeneration process on two complex grating structures was explored (Canning et al., 2010):

For the superposed gratings (L ~ 5 mm) were inscribed into a H₂ loaded (24 h, P = 100 atm, T = 100°C) GeO₂ doped core silica fibre ([GeO₂] ~ 10%, fabricated at CGCRI) using a pulsed KrF exciplex laser (248 nm, pulse duration = 20 ns, f_pulse ~ 70 mJ/cm², repetition rate = 200 Hz). The Moiré grating was written into a fibre which was similar to that used for the superposed grating but also had boron to increase the seed photosensitivity. Regeneration is carried out with an identical recipe to that described earlier but inside a short fibre micro-heater. The hot zone of this heater is supposedly uniform over 5 mm only (the exact variation along this length is not known but we suspect a Gaussian profile), and this dictates the grating length.

9.4 Conclusions

The preservation of optical phase information within a grating is evidence that the resolution of features within a laser-written seed grating can be preserved on a nanoscale after annealing and subsequent regeneration. Clearly information is retained; almost certainly the memory is stored through differences in glass structural properties between original seed processed and unprocessed regions. Hydrogen preferentially selects out changes in silica; without hydrogen it would seem that lower temperature effects dominate. The exact role of hydrogen is unclear, but above 500°C we know Si-OH is readily formed (Sørenson et al., 2005); laser irradiation may likely increase the defect sites where OH can form leading to a higher concentration and therefore greater stress changes (swelling) which can relieve core-cladding tensile stresses more within these regions, for example. This explains the original observed sensitivity of regeneration to fibre tension (Bandyopadhyay et al., 2008). However, more studies are necessary, including potential simulations such as that previously used to study radiation bond breaking of Si-O (Wootten et al., 2001).

The recent work decoupling the role of hydrogen from seed grating fabrication (Canning et al., 2011; Lindner et al., 2011b) points quite clearly to a common mechanism between both with and without hydrogen type I gratings in retaining memory, despite one being associated with mostly OH formation and the other with density change and defect polarizability. In spite of some differences, both lead to periodic stress along the grating, although where one involves compaction the other may involve dilation or net reduced compaction. This, together with the original observation of Bragg wavelength mismatch between regenerated and seed gratings as a function of applied tension, suggests stress and structural changes at the interface are critical. Separating out the contributions of dopants, especially those in the core, strongly supports the hypothesis that important change involves silica transformation most likely near the core–cladding interface (consistent with differences in regenerated Bragg wavelength shift from the original seed grating as a function of tension, as originally first observed). The nature of this transformation remains an area of important investigation. These structures can withstand temperatures up to and beyond 1400°C for short periods before rapid annealing is observed, although more detailed work is required to better quantify the conditions of these upper limits. At this temperature the majority of core dopants will lead to core glass softening, making it less likely that the important change is occurring in the core. Core-cladding interface changes are strongly suggestive of stress as a significant contribution to the type of glass polymorph or polyamorph that is induced. Thus further work with applied stresses is warranted.

The retained complex functionality of a laser-written seed structure after very high temperature processing, tantalizingly hints at the prospect of ultra high temperature patterning of glass, something vital for many industries, including semiconductor lithography where more robust masks are necessary and also in applications that require robust holographic filters or reflectors such as high power fibre lasers and a new generation of high power holograms (including a potentially lower cost approach to fabricating optical phase masks for fibre grating writing).

Given that much of the general process appears to involve glass re-quenching under a different thermal history, both the thermal stabilization and the regenerative processes described here are unlikely to be confined to silica fibres, or waveguides, loaded with or without hydrogen (as recent results indicate). Lower temperature regeneration without hydrogen suggests changes in structure of the germanate glass, which is less stable than silica. The correlation with laser-written regeneration such as type In (type IIa) gratings suggests that high multiphoton lasers can be fine-tuned to achieve similar laser regeneration in pure silica through continued exposure, another potential subject area of investigation.

This work indicates potentially superior performance may still be extracted by optimizing the thermal robustness (or indeed melting point) of the core glass relative to the cladding. Exploring other systems such as aluminosilicate, phosphosilicate and indeed other glasses (oxide or otherwise) to explore the predicted dependency, or in some cases limitations, is another area of important work.

There is no obvious reason why these changes could not be reproduced under different material systems and indeed different laser preparations given the foundation of glass transformation under appropriate cooling and quenching kinetics – all that is required is an amorphous network (and even some possible discrete transitions within crystals could be prepared analogously). As such, the first suggestion for controlling relaxation kinetics and thermodynamics was by laser (Canning, 2004) – here, we have reviewed a hybrid approach where a seed is patterned with a laser and thermal macro annealing in the presence of hydrogen completes the transformation. There is no reason why the entire process could not be done using laser or other variations. Anyone skilled in the area of amorphous materials can see the broader potential of such controlled transformations in their particular niches.

9.5 Acknowledgements

We would like to acknowledge various colleagues who have worked on several aspects of regenerated gratings at different stages. Some of the more recent results warrant special mention of the following: Palas Biswas from CGCRI, Michael Stevenson, Kevin Cook, Mattias Aslund from iPL, and Hongyan Fu and Hwayaw Tam from Hong Kong Polytechnic.

Funding from the Australian Research Council (ARC) and an International Science Linkage Grant from the Department of Industry, Innovation, Science and Research (DIISR), Australia and the Council of Scientific and Industrial Research (CSIR), India, under the 11th five-year plan is acknowledged.

9.6 References