1. Introduction

Light-induced changes in glassy materials, often referred to as photosensitivity, can be understood as the response of an optical glass network to photons of a range of energies and pulse durations, including the structure and property modifications. The physical response of the glassy material can vary greatly, ranging from laser-induced refractive index changes, to photo-expansion/contraction, photo-darkening/bleaching, etc. Each of these effects may vary in magnitude, sign, or stability based on the constituents of the network (covalent, ionic, or other bonds), the illuminating light’s wavelength/energy relative to the material’s bandgap, as well as the overall timescale (CW, kHz, MHz) at which the energy is deposited. Thus, light-induced property changes in glass, while studied for over two decades for a diverse range of potential applications, still remains a rich field of research.

In this paper we review, from a materials science perspective, the mechanisms believed to dominate the response of glassy networks to light exposure with a final focus on femtosecond (fs), short pulse/high intensity laser exposure. Photoinduced processes are summarized in general for different families of glasses over varied timescales and exposure regimes, beginning with an overview of the mechanisms involved in photosensitive response, the impact of CW exposure, and a discussion as to how fs-laser exposure fits into the mechanistic landscape of material modification. In addition to illustrating photoinduced behavior in various glass systems, current knowledge of key irradiation conditions most suited to writing passive and active components in bulk glass is reviewed, and contrasts drawn to the variations of this response. An enhanced understanding of mechanistic variation in material systems will be the key to further optimization and adoption of laser-written structures in the future realm of optical components.

2. Light-Induced Structural Changes in Glassy Solids

Photoinduced property changes in amorphous materials, regardless of glass or radiation/laser type, are generated through the same general series of processes: 1) the material absorbs energy from the incident source/laser light, 2) the network structure of the material is modified by the energy imparted, and 3) the induced network modification is reflected as changes (either permanent or transient) to material’s physical properties. Each of these three processes will be discussed herein, contrasting two exposure conditions, continuous wave (CW) and femtosecond (fs), and their application to oxide and non-oxide glasses.

2.1 Material Absorption of Laser Energy

In the case of CW irradiation of glass, the most important parameter of the incident light in the context of material absorption is the exposure wavelength, λ_photon. Absorption of CW laser light by amorphous networks is generally a single-photon event for which the threshold equality is Δ = hc/λ_photon, where h is Plank’s constant, c is the speed of light, and Δ is the material bandgap. At energies greater than Δ, i.e., λ_photon < hc/ Δ, single photons are absorbed by the network by shifting an electron from the valence band to the conduction band. An additional difference in the way a glass responds to CW vis-à-vis fs laser light arises from the much larger irradiance or peak power density imparted by the latter on a sample. Thus nonlinear effects such as due to multiphoton absorption dominate the effects generated in the material by a fs laser, whereas for all practical purposes CW laser introduces only linear optical effects. Thus structural modification by photoelectronic mechanisms require that the CW photon energy be comparable to or larger than the energy bandgap of the glass. Only then is the photon absorbed and an electron-hole pair generated, which may modify the structure and properties before disappearing by recombination or being trapped somewhere in the structure. On the other hand, intensity-dependent nonlinear absorption can occur even for photon energies less than the bandgap, as two or more photons together excite an electron from the valence band to the conduction band. In other words, for the electronic effects to be observed, the CW laser must have a minimum frequency as determined by the bandgap of the glass, while there is no such requirement with high power fs lasers.

It follows that if photon energy of CW laser is less than the bandgap of the material, there will be no excitation of electrons from the valence band to the conduction band and the light will simply travel through the glass without any absorption (neglecting scattering). Thus, there should be no change in the structure or properties of the glass under these conditions. Notwithstanding, the low energy or long wavelength infrared (IR) photons from a CW laser can still cause photostructural changes if their frequency matches with the vibrational frequency of the molecular species that make up the glass. This resonant IR absorption offers a mechanism for photoinduced changes, which applies to the CW as well as fs laser irradiation case. Here, light energy is used to excite directly the atomic vibrations, which is equivalent to increasing the local temperature that in turn may modify the structure selectively in the irradiated region of the sample - see Section 3.

Unlike in the case of CW irradiation, when glass is exposed to laser pulses with durations on the order of femtoseconds, the energy transfer from the laser beam to the material typically is a nonlinear process. In this case it is laser intensity and pulse duration which are the dominant laser beam characteristics. Absorption by the material of laser light with energies below its bangap occurs either through a multiphoton ionization process or a photoinduced tunneling process which excites an electron from the valence to the conduction band. Multiphoton ionization is the dominant process at high laser frequency and low electric field, whereas photoinduced tunneling dominates the low-frequency, high-field regime [1]. Additional details of these differences are discussed in the partner review article in this volume [2].

From a materials science perspective, intrinsic differences should be expected between the responses of oxide and non-oxide glasses during these energy transfer processes. In comparison to the wide-bandgap silicates (6-9 eV), the chalcogenides and other non-oxide glasses are semiconducting, with comparably small bandgaps (2-3 eV) [3,4]. The impact of this variation in bandgap on the energy transfer mechanisms and efficiencies is seen in both CW and fs exposure regimes. Because CW processes typically involve one-photon at a time, the wavelength of light necessary to induce energy transfer from the beam to the network varies widely across glass systems [3]. For fs exposure processes, the density of photons needed to promote a sufficient number of electrons to the conduction band through nonlinear processes, i.e. the laser intensity needed to make photostructural modifications, is much lower for chalcogenides than it is for silicate materials [5]. As will be discussed in detail below, the fs-ablation threshold intensity at which material damage occurs, as compared to the threshold for structural modification (I_threshold), is also a strong function of the bangap of the glass.

2.2 Photoinduced Changes in Network Structure

In many ways, the understanding of the photoinduced changes in glass network structure is less well-developed than that for the energy transfer between the laser and the material. A number of interesting models have been developed, but none of them currently provides a structural interpretation that is consistent across glass families and explains the wide range of photoinduced property variations. These changes in network structure in oxide glasses are discussed in depth in the companion review article in this issue [2].

A model originally developed by Street proposes that the absorption of a photon creates a self trapped exciton state in the network [6]. A photon with near-bandgap energy can be absorbed and create a loosely bound Wannier-Mott exciton in the network [7]. At this point the exciton may either recombine, or through a phonon-assisted process may drop to only a slightly lower energy state identified as a self-trapped exciton, or combine nonradiatively back into the ground state configuration. This self-trapped exciton appears in the network structure as a point defect (D⁺, D^-) pair. In a chalcogen network, this defect pair would contain one tri-coordinated (D⁺) chalcogen paired with a single-coordinated (D^-) chalcogen species [6] such as shown in the schematic diagram of Fig. 1 .

 

Fig. 1 (a) Schematic diagram of, and (b) defect pair created by the self-trapped exciton process.

Download Full Size | PDF

A second model for the photostructural modification process is the photoinduced annihilation of frozen-in homopolar bonds, and their replacement with heteropolar bonding structures. As an example for the chalcogenide glasses, Elliot argues that arsenic-rich binary glasses contain a finite fraction of As-As bonds frozen into the random network structure of the glass as a result of the rapid kinetic arrest during quenching [8]. Here, the As-As σ-orbital bond interacts with lone-pair electrons from a neighboring chalcogen atom to form a low-lying π* state in the bandgap. Absorption of an incident photon scissions the homopolar bond, and the network instead forms two heteropolar bonds, which have a favorable electronegativity difference.

A third model is rather empirical; it attempts to explain the measured increase in the fraction of homopolar bonds following photoirradiation of a chalcogenide glass [9]. The process is likely to be multiphoton, as it requires the scission of two energetically favorable heteropolar bonds and the formation of two homopolar bonds. Because of the multiphoton nature of this transition, it is more frequently observed in femtosecond irradiation studies because the pulse intensities are high enough to produce nonlinear effects [10]. This bond rearrangement process is shown in Fig. 2(b) below; the transition may proceed directly if the (multi)photon energy is great enough, or the process may proceed through a lower lying self-trapped-exciton intermediate state, discussed above.

 

Fig. 2 (a) Schematic of material’s response when subjected to CW laser irradiation, delineated into three regimes, mild laser heating/cooling, strong laser heating/cooling and evaporation/ablation, depending on sample surface temperature vs. irradiation time. (b) Bond rearrangement processes in chalcogenide glasses. After [6,8].

Download Full Size | PDF

2.3 Photoinduced Changes in Material Properties

As noted in the Introduction section, a wide range of photoinduced changes has been observed in the thermal, physical, and optical properties of glasses. Laser-induced refractive index change is one of the most interesting and useful phenomenon from an applications perspective, as it provides a means by which to direct-write waveguide structures for photonics applications [11]. Both UV- (CW or pulsed) and NIR- (fs) irradiation have been demonstrated to induce photodarkening in fused silica, through the same process of color center formation [12,13]. The identical nature of the structural changes confirms the multiphoton nature of the NIR-fs induced processes. These same results have been demonstrated in a number of silicate systems, where in each case the generation of free electrons from the laser pulse leads to the formation of trapped electron-hole pairs and the formation of E^’ centers consisting of an unpaired electron localized on a silicon atom paired with a nearby oxygen vacancy [14,15]. Studies of these E’ centers, induced through both CW UV-exposure and NIR fs-exposure, in silicate glasses have shown them to be transient structures, with their recombination lifetimes controlled by the OH content of the glass [16].

A first order assumption would posit that a photoinduced increase in the refractive index should be associated with a concurrent photoinduced compaction of the glass and its network structure. While it is true that the processes of photo-expansion/compaction are closely linked to photoinduced refractive index change, the correlation between them (critical for waveguide writing applications [17–19]) is not perfect, and depends critically on glass type. Fused silica, the prototypical oxide glass, when exposed to fs irradiation exhibits localized increases in both network density (photocompaction) and refractive index. However, in arsenic trisulfide, the prototypical chalcogenide glass, photoexpansion is shown to occur, along with an increase in refractive index, which is then linked to a change in the material’s dielectric susceptibility [20].

3. CW Exposure of Glasses

3.1. Materials and Mechanisms

From the viewpoint of photostructuring, inorganic glasses can be classified into two categories depending on the strength of their dominant chemical bonds: (i) strongly bonded hard glass, or (ii) weakly bonded soft glass. Oxide glasses such silica, germania, Pyrex, etc. are examples of hard glasses that are made of strong, three-dimensional network of cation-oxygen ionic-covalent bonds. By comparison, non-oxide glasses, most commonly represented by chalcogenide glasses, are examples of the other class. In general, the strongly bonded oxide glasses are harder to structure than the weakly bonded chalcogenides. Although the basic physics of laser-material interactions may not be different for the two classes, there are major differences in the effects and phenomena that are exhibited by them, which are discussed next.

3.2. Photostructuring of ‘Hard’ Glasses by CW Laser Exposure

The large strength of cation-oxygen bonds in this class of glasses is responsible for their large bandgap, which makes them optically colorless and transparent, electrically insulating as well as mechanically strong and inflexible. Thus visible light, such as from a typical CW He-Ne laser, has no effect on their structure. Only UV radiation of sufficiently high energy is absorbed, which is capable of creating electron-hole pairs. The so-generated photocarriers can be trapped in the structure, producing color centers and other electronic defects. However, the bonds are too strong and electron-phonon coupling is relatively weak, so that the atomic structure of glass remains relatively unaffected. All these characteristics of oxide glass, which lead to optical insensitivity, have been the reasons for their very broad range of applications developed over many centuries. Therefore, the pursuit of photostructuring in common oxide glasses by visible irradiation is against their inherent nature, but it can be accomplished rather readily with the use of intense IR lasers. For this purpose, most commonly CO₂ lasers are used, which are commercially available in the highest power under CW operation and most commonly used in industrial material processing. The main wavelength emitted by a CO₂ laser is 10.6 μm (i.e. 963 cm⁻¹ wavenumber), but there are other lines in the region of 9–11 μm, particularly at 9.6 μm (i.e. 1042 cm⁻¹). Fortunately, this wavelength region matches with the vibrations of silicate network, resulting in strong absorption of CO₂ laser and consequently effective local heating of common silicate glasses [21,22]. For example, bulk silica glass has transverse optic mode at 1076 cm⁻¹ from asymmetric stretching of oxygen (with in-phase motion of adjacent oxygen atoms). A downside of the strong absorption of CO₂ laser by silica (or other silicate) glass is that the laser can penetrate only to a depth of about 10 μm or less [23]. Therefore, any restructuring of such glasses will be limited to near the surface to the depths of several micrometers at most.

Besides laser-heating of glass utilizing the material’s inherent vibrational modes, alternatively one can exploit optical transitions of rare-earth or transition metal ions that are either a major component of the sample or are added as a dopant. In this case, one would use a laser of matching wavelength in the near IR or visible range. For example, Komatsu et al. [24] have used readily available Nd:YAG laser of λ = 1064 nm for fabricating single-crystal patterns by the photostructuring Sm or Dy rare-earth doped borate glasses, or BaO–TiO₂–GeO₂ and BaO–TiO₂–SiO₂ glasses doped with less than 1 mol% of NiO, Fe₂O₃ or V₂O_5. More complex methods combining irradiation for nucleation followed by thermal treatment for the growth of a second phase with the help of sensitizing agents like silver and cerium have been known for many years starting with the classic work of Stookey using a UV lamp [25]. More recently, this concept has been expanded as photothermorefractive writing of holograms in robust aluminosilicate glass matrix using UV lasers [26], which has resulted in new optical components commercially available [27] for laser beam steering, filtering, combining and other types of manipulation.

Veiko et al. [28] described the basic processes that follow when laser energy is deposited as heat in a glass by any one of the mechanisms described above. The photostructural changes that follow depend on the spatial distribution and time dependent variation of local temperature as effected by laser irradiation. Thus heating and cooling rates, time of laser action and ambient temperature are the external parameters, whereas physical and thermomechanical properties are the intrinsic characteristics of the sample that determine the final outcome of photostructuring. Three different regimes of sample response may be identified depending on the conditions of irradiation: (i) Mild laser heating/cooling (MLH/C) regime where the heating and cooling rates are below the critical value at which cracks start forming, (ii) Strong laser heating/cooling (SLH/C) regime where the laser-induced stresses are high enough for creating the cracks but the intensity is insufficient to cause material loss, and (iii) Evaporation regime where the laser intensity is very high, so the glass is evaporated or ablated locally. Figure 2(a) shows these regions schematically, which have been separated depending on the surface temperature of glass as a function of laser exposure time, the former being determined by the power of laser and the absorption properties of the sample. Here the evaporation regime (E) at the top is obvious, but the MLH/C and SLH/C regimes are complicated as they are strongly affected by the material heating/cooling rate, r = ∣dT/dt∣. If r is greater than a critical value, r_c, cracks form, during heating due to thermal shock or by slow crack growth during cooling. A particularly interesting sub-regime of MLH/C regime is at the upper right corner, where glass begins devitrifying with the formation of crystallites. Alternatively, it may simply undergo viscous flow and anneal out all the stresses in this sub-regime.

The following processes may occur in the MLH/C regime, which are useful to know and control for determining the suitability of laser-induced structuring for most practical applications [28,29]: (i) thermal expansion and surface deformation, (ii) viscous flow under surface tension and gravitation, (iii) thermomechanical stress formation and consequent tempering, (iv) devitrification of glass (or amorphization of crystallites if the sample is partially or fully crystalline), (v) densification and shrinkage of porous glasses and coatings, (vi) fusion and reformation of glass. Note that (iv) and (vi) are opposite of each other, and specific laser irradiation conditions would determine which process will dominate. Indeed, it is possible to accomplish reversible transformation between glass and glass-ceramic with appropriate irradiation [30].

3.2.1. Applications of Photostructuring of ‘Hard’ Glass

All the three regimes of laser photostructuring shown in Fig. 2(a) have potential for introducing novel methods for fabricating optical elements and devices. Among them MLH/C conditions have been exploited the most in fabricating micro-optical components, such as for making microlenses and more complex optical elements by densification of silica gel [31,32], by laser ablation and localized heating of low thermal expansion silica [33] or silicate glass [34], etc. Although still in the research and development stage, there is increasing interest in the selective devitrification of glass whereby a single crystal is ‘written’ near the surface by controlled MLH/C. Initially, the laser heating devitrifies a small region of the sample into multiple crystallites. When the laser beam is moved away from this region, the most suitably oriented crystallite acts as the nucleus for the growth of a single crystal line along the temperature gradient that ensues the laser motion [14]. Figure 3(a) shows an example of a ferroelectric LaBGeO₅ single crystal fabricated within the glass of same composition [35]. Here the single-crystal character of the line is maintained through its turn without introducing any grain boundary, demonstrating the potential of this process for fabricating complex optically active architecture in an inexpensive glass for integrated optics. Very recently the concept of single-crystal laser writing has been extended to fabricate single-crystal layer on the surface of glass by rastering the laser beam back and forth. Such a glass structure has the potential of replacing expensive single crystals that are used as substrate for fabricating epitaxially grown devices.

 

Fig. 3 (a) A curved LaBGeO₅ ferroelectric single crystal ‘written’ in glass of the same composition by heating with a CW Nd:YAG laser of λ = 1064 nm. The photo shows second harmonic generation by this active crystal in the background of inactive glass matrix. Here the orientation of crystal lattice does not change as the laser-written line bends and forms U-shape structure. (b) Evolution of local structure around As atoms in a-As45S55 film, as seen in average bond length, when irradiated with 488 nm Argon ion laser. VP and HP represent vertically and horizontally polarization of the laser

Download Full Size | PDF

Finally, SLH has proven to be very useful for scribing or cutting the glass sheets such as for liquid crystal displays [36,37]. In this application very high temperature gradients are required. Therefore, the heating of glass by the scribing laser beam is closely followed by rapid cooling with a cold water jet. A tensile stress is generated on the surface along the scribing direction, while the interior is still hot creating a compressive stress ultimately. The latter stress enhances the tensile stress and the crack formed at the surface grows. The cut formed in this way has higher accuracy, and forms with fewer debris particles. Laser power, the scribing velocity and the distance between the laser heating and water cooling regions are the key process parameters that determine the acceptable minimum and maximum scribing speeds.

3.3. Photostructuring of ‘Soft’ Glass by CW Laser

Typical soft glasses, that is, from the perspective of photostructuring, are chalcogenides of the elements of Group IV (e.g. Ge) and V (e.g. As, Sb). Unlike oxide glass insulators, they are wide bandgap semiconductors. Therefore, electrons are readily excited across the bandgap by the absorption of visible light, introducing the possibility of photosensitivity to readily available lamps or sunlight as well as common CW lasers. Often their atomic structure consists of 1D chains (as in a-Se) and 2D layers/sheets (as in glassy As₂S₃), which are more flexible in response to any stress than 3D network of strong oxide glasses. Additionally, they have strong electron-phonon coupling, which implies that any changes in the electronic structure from exposure to light will also affect the atomic structure. Consequently, chalcogenide glasses are highly photosensitive, and light may affect their volume [38], amorphization [39], devitrification [40], mechanical (e.g. plasticity [41]), rheological (e.g. viscosity), optical (e.g. darkening, birefringence, luminescence) [42], electrical (e.g. conductivity, dielectric constant), or chemical (e.g. etching [43], dissolution [44], doping) properties. The effects based on these changes have very large commercial potential. Therefore the subject has been investigated extensively, and the reader interested in details is referred to excellent reviews and texts [45–48]. Here we present only a broad overview from a materials perspective.

The broad range of observations of photoinduced phenomena in ChG can be divided according to their temporal response. On one hand, there are permanent photostructural changes which cannot be reversed once the sample has been irradiated, unless it is prepared again from the melt. On the other hand, there are temporary or transient changes which exist only in the presence of pump beam. That is, as soon as the laser is turned off, the sample begins to revert back to its pre-irradiation stage. There is also an intermediate reversible response, which may require that the sample is heated to its glass transition temperature (T_g). Of course, the kinetics of the development or reversal of a given transient effect can vary from ps to hours or longer. All of these effects can be further grouped as vector or scalar, depending on whether their magnitude depends on the polarization of light. Photofluidity is an example of a scalar effect, which represents increased mobility of glass structure when exposed to appropriate bandgap light. Photoinduced optical anisotropy is the best known example of a vector effect [49]. It is observed as anisotropic transmission of a polarized probe beam, when the isotropic ChG sample is irradiated with a polarized pump laser beam. Usually, the transmission of probe beam with polarization parallel to that of the pump is higher than that of the probe beam with perpendicular polarization.

The elements in a ChG are primarily joined by covalent bonds, which are expected to follow Mott’s ‘8-N’ rule. Accordingly, the coordination number (CN) of an atom should be simply eight minus the number of its valence electrons. Therefore, for example, in an As-Se glass CN for As and Se should be 3 and 2, respectively. Furthermore, as the constituent atoms form stable compounds, the so-called heteropolar bond (like As-Se) between dissimilar elements should be more stable than the average homopolar bonds (i.e. As-As and Se-Se). With these restrictions the glass of As₂Se₃ composition should consist of only three-coordinated As and two-coordinated Se heteropolar bonds. However, as the glass is made from the melt or vapor state, these rules are broken, and one generally observes bonding defects, which include atoms in over/under coordinated configurations and homopolar bonds [50]. The concentration of these defects depends on the preparation method and thermal history. So, for example, thin films prepared by thermal evaporation can be expected to have a much higher concentration of bonding defects than the bulk glasses made by the slowly cooled melt-quench method. When such films are exposed to bandgap light, the defects are annealed out making their structure closer to that of the bulk glass with fewer defects.

Figure 3(b) shows a recent atomistic result describing the evolution of average interatomic distance from As atoms in As₄₅Se₅₅ glass film that is exposed to 488 nm Argon ion laser [51]. This type of information about the local photostructural changes vis-à-vis macroscopic measurements of photoexpansion suggests that there is an additional relaxation of the structure beyond the nearest neighbor, which involves Van der Waals interactions between the chains and sheets of covalently bonded atoms. Then the overall photostructural change is a combination of all these contributions, which then determine the nature and magnitude of various photoinduced effects.

The understanding of photoinduced phenomena in terms of bonding defects has provided scientific basis and helpful guidelines for optimizing composition to yield the desired results. In this regard, we note that the concentration and stability of relevant defects can be controlled by the way the sample is fabricated (e.g. by the deposition rate of thin films), or by changing the composition of starting glass [52,53] An interesting aspect of photosensitivity of certain properties of ChG is that they can have either positive or negative response, depending on the composition or irradiation condition. Thus, for example, arsenic-based chalcogenides exhibit photodarkening and photoexpansion when exposed to bandgap light. On the contrary, germanium-based chalcogenide compositions exhibit photobleaching and photocontraction. So for applications requiring photostable glass that does not change with time, one can mix these two types of glasses to obtain the desired results [54]. Yet another useful characteristic of ChG is photodiffusion of silver and few other metals [44]. It allows the formation of new phases with significantly different composition and properties only in the regions irradiated with bandgap light. This phenomenon has been exploited in the application of ChG as a photoresist for grayscale dry etching lithography with superior results than with currently used polymer resists [55].

4. Pulsed Exposure of Glasses

Due to the nonlinear, multiphoton nature of the laser-material interaction during femtosecond irradiation, the field intensity, I, is of critical importance in controlling the response of the glass for the desired application. Above a threshold intensity, I_threshold, which is not strongly material dependent [56], the pulse causes permanent structural changes in the network. This threshold intensity is a function of the pulse duration, the pulse energy, and the numerical aperture of the focusing lens used to launch the beam into the material. A linear dependence between the threshold intensity for modification and pulse duration has been demonstrated, indicating that avalanche impact ionization, discussed in Section 2.1, is important in generating the excited electron density necessary for structural rearrangement of the network ions [57]. Thus, the variation in bandgap between the silicates, tellurites, phosphates, and chalcogenides is critical in determining the threshold intensity for material modification and ablation. The transition between these two phenomenon is what has been defined as the material’s ‘ablation threshold’.

In addition to the laser intensity, the repetition rate is critical in establishing a material’s ablation threshold. Several researchers have studied the interplay of these two parameters in determining the threshold for material ablation as a function of glass type. A schematic of this relationship is shown in Fig. 4 , indicating the relative ease of either intentionally or unintentionally reaching the ablation threshold in various material systems. The much weaker network bonding in chalcogenide materials, evidenced by their low glass transition temperatures, makes photomodification of their structures possible at much lower total dose or fluence, (i.e., the product of the repetition rate (sec⁻¹) times the energy density (J/pulse) times the number of pulses, N) than is possible, for example in materials with a more covalent nature (determined by quantifying the fractional ionic or covalent character of the bonds present).

 

Fig. 4 Variation of ablation conditions as a function of glass type. From [5].

Download Full Size | PDF

From a materials engineering perspective, these relationships translate directly into a minimum writing energy and intensity, i.e. number of Joules per pulse at a number of pulses per second, needed to make either permanent or transient material modification, which in turn dictates the laser system necessary for production. For use in industrial processes, where high throughput and low-defect writing are of critical importance, this becomes one of the key driving issues in making laser processing economically viable [58,59].

The precisely controlled energy transfer and structural modification were first implemented to write waveguides in several families of glass substrates by Miura et al. [60]. These waveguides were written above the ablation threshold, where the beam intensity was high enough to cause microscopic explosions, leaving visible voids in the glass. This method was shown to produce both single and multimode waveguide structures through an induced index change. At almost exactly the same time Glezer and Mazur demonstrated arrays of fs-laser induced voids in silica, quartz and sapphire which acted as diffraction gratings in the bulk material [61]. The ability to write ferroelectric single crystal architecture in glass has opened the possibility of incorporating active 3D optical functionality in glass [62]. These demonstrations of direct laser writing of optical structures utilizing fs pulses (at nJ/pulse levels) have provided a powerful tool to the growing field of glasses for photonic applications and optical data storage [63].

4.1 fs Modification of Silicates, Borosilicates, Borates and Phosphates

The silicate, phosphate and borate glasses serve as the proving ground for structural studies, models, and applications of fs-induced reactions in glasses [64,65]. Hirao et al. first studied the refractive index change in the silicate, soda-lime silicate, and borate families of glasses for waveguide writing using fs lasers and suggested the possibility of writing three dimensional structures into these glasses [66,67]. The structural mechanism of waveguide formation below the ablation threshold in silicates was identified to be the formation of color centers and a corresponding refractive index increases due to non-ablative densification due to a change in oxygen bond angle [13,68]. Femtosecond irradiation was subsequently used to write optically active elements into rare-earth doped silicate glasses [17,69].

Structural modification in fused silica was demonstrated to be dependent on the fs-beam polarization and the translation direction of the beam relative to the glass substrate [70]. Translation of the beam parallel to the direction of optical polarization produces Bragg-like gratings in the glass [71]. In addition to polarization of the incident beam, the profile of the induced structural change in the glass is determined by the nature of the focusing optic used to launch the beam into the material. Low numerical aperture optics produce elliptically shaped focal spots, meaning that translation of the beam transverse to the substrate produces highly asymmetric waveguide geometries, which depending on the target shape for the modified region, may not be acceptable [72].

Investigation of the network structural changes induced in waveguide writing by the fs-laser beam as the intensity is increased first above I_threshold, and then above the ablation threshold in silica reveals that microexplosions in the glass lead to increasing fractions of 3- and 4-membered silica rings, leading to a decrease in the average network bond angle, and hence densification of the structure [68,73]. The decision of where to modify material in the range I_threshold < I < I_ablation will of course dictate the resulting optical quality of the modified region and determine its suitability for subsequent light propagation [74–76].

In addition to the use of photoinduced refractive index changes for writing optical elements, fs-induced processes have been used to generate selective precipitation and phase change. Controlled crystallization of Ba₂TiSi₂O crystal phases with high optical nonlinearities has been demonstrated in erbium-doped silicate glasses using femtosecond laser irradiation, both lines and arrays of crystallized volumes have been made [77,78]. 2-D spatially selective phase change in materials has been used for memory devices for many years; the prospect of the use of fs-laser irradiation as a means to produce 3-D spatially selective optical arrays has also provoked much interest [79].

4.2 fs Modification of Tellurites

To a large extent, the response of the tellurite glasses is similar to that seen in silicate glasses, though the weaker Te-O bond decreases the ablation threshold for the material. Optical characteristics of waveguide-writing in tellurite glasses showed photomodification thresholds on the order of 150 kJ/cm² [80]. Waveguide writing in niobium tellurite glasses has been investigated as a function of pulse duration. It was found that the picosecond pulses produced only a transient change in the refractive index of the glass, due to an induced absorption, while femtosecond pulses produced permanent refractive index change, indicative of structural reorganization [81]. Femtosecond direct laser writing of waveguide structures in phosphorous-containing rare-earth doped tellurite glasses has demonstrated low loss structures [82], capable of optical amplification with gain bandwidths spanning the C and L telecommunication bands [83].

Studies comparing the variation in structural response between Ag⁺-doped silicates, fused silica, and zinc tellurite glasses have uncovered a wide range of structural phenomena ranging from nanoparticle formation in metal-doped glasses, to zinc and tellurium migration in the tellurites [84], and the formation of self-organized periodic structures in the oxide glasses [85].

Femtosecond lasers operating above the ablation threshold for the glass have been used to form structures in silicates and tellurites through a controlled series of microexplosions. These explosions are evident in the glass as voids surrounded by shockwave-induced density variations. In Al₂O₃ and La₂O₃ doped glasses, these density variations were found to correspond to mass transport of the dopant ions away from the focal point of the laser pulse [86]. Linear arrays of voids have been proposed as possible waveguiding structures.

4.3 fs Modification of Chalcogenides

As with the other glass families, a full range of photoinduced phenomena is observed in chalcogenide glasses when exposed to fs-irradiation. The physical, optical, and structural response of the glass to irradiation is critically dependent on glass composition. For example, both arsenic- and germanium-sulfide glass films exhibit a localized photoexpansion when irradiated, however in the former this photoexpansion is associated with an increase in the refractive index, whereas in the latter it is associated with a decrease in refractive index [87]. In the case of arsenic sulfide, the increase in refractive index, on the order of 8 x 10⁻³, is due to the photoinduced formation of As-As homopolar bonds [88] via the mechanism shown in Fig. 2(b); whereas in the case of the germanium sulfide system the optical energy serves to promote the formation of S-S linking bridges between germanium centers, which decreases the index [10]. A similar difficulty in deconvoluting photoinduced effects is seen in the example of Tanaka’s experiments with badgap irradiation of As₂S₃, where it was demonstrated that one-photon and two-photon processes occurring at the same frequency would produce different structural changes, with the one-photon process producing photodarkening and the two-photon process producing no photodarkening, but an increase in refractive index [89].

The refractive index increase observed during femtosecond irradiation of arsenic sulfide glasses is extremely advantageous to the formation of waveguide structures in the material [88,90]. Whereas Ge-containing materials typically exhibit negative index changes, believed to be due to the more constrained nature of the glass network (provided by the Ge species with CN = 4) other ‘compensating’ aspects of could be valuable to material/component design. Similar to the design of athermal (dn/dT ~0) optical structures, one might envision photonic device designs where use of multi-material solutions based on positive and negative induced index changes, could prove valuable.

5. Future Applications in Materials

As included in the multiple examples cited in the present article, the future for use of laser-induced modification of glasses and glass-ceramic materials is moving at a rapid pace, driven by materials development in conjunction with laser physics.

Recent efforts to understand the fundamental physics required to design and fabricate components that could be used in systems to manipulate light in a photonic network have shown promising designs that could be physically realized. In addition to efforts to model light propagation through 2- and 3D generated structures [91], recent experimental efforts have demonstrated the creation of multifunctional components in a single glass material (couplers and Fresnel zone plates) [92] and fully integrated solutions where a monolithic LiNbO₃ waveguide chip combined frequency doubling and a modulating unit [93]. Waveguide lasers have been direct-written into Tm³⁺:ZBLAN using MHz fs-irradiation, which demonstrate promising slope efficiencies [94]. Phase masks have been used in conjunction with fs-duration pulses to pattern Fiber Bragg Gratings (FBGs) [95] in germanate and tellurite glasses; comparison of the strain sensitivities of the FBGs inscribed into the two glasses indicated that the tellurite glass would be a better candidate material for fiber strain sensor applications, due to its lower Young’s modulus [96]. These techniques have also been employed in chalcogenide-based strain sensors as well [97]. Some of these efforts have resulted in fs-written commercial products [98,99]. While not exhaustive, these few examples highlight the broad promise of diverse materials and component designs possible.

6. Summary

This article has provided an overview of key material-specific examples of how light can be used to induce changes in glassy and glass-ceramic materials to enable the creation of unique photonic structures. As noted, the material modification mechanisms responsible for the observed photoinduced changes are varied, but glass scientists in conjunction with optical physicists have narrowed the gap considerably across the last decade making key advances in the ability to compositionally-tailor glasses with desired photoresponse. The fundamental science underlying many of these advances is discussed in this volume. These advances will continue to lead to the creation of novel components and devices in future years. However key future opportunities will rely on the creation of robust materials that can be fabricated in a rapid and repeatable manner (optimization of the laser manufacturing process) in a manner that allows retention of stable, permanent structures that do not decay or degrade in their use environment (the materials optimization component). With the active research taking place in these areas across the world, clearly new components and devices of unique design, form and function, will continue to evolve, transforming the way light is utilized in future photonic applications.