1. Introduction

There is continuing interest in fabricating fibre Bragg gratings (FBGs) by side-writing using pulsed [1], ultra-short pulse [2, 3, 4] and continuous wave [5, 6] lasers. Holographic interferometry [1] and zero-order suppressed phase masks [7] have proved suitable for defining the sub-micron period refractive index variations needed for FBGs in telecommunications and sensor applications. The shortest exposure wavelength for FBGs has been mainly restricted to the 193 nm ArF laser [8], although potential advantage could acrue if this were reduced because one-photon photosensitivity mechanisms could be expected to be stronger. The possibility of this allowing FBGs to be written efficiently in some common fibres without the need for sensitisation by H₂ loading could be beneficial, as it would avoid ageing/relaxation problems associated with these treatments. There might also be a writing speed advantage, even for fibres that are intrinsically photosensitive.

In earlier work Chen et al [9] and Herman et al [10] measured 157 nm laser-induced refractive index changes in fibre and planar glass samples, noting increased photosensitivity at this short wavelength and the possibility of improving the efficiency of writing index-change microstructures. 157nm exposure has also been employed for fabricating long-period gratings [11], for pre- and post-treatment of FBGs formed at longer laser wavelengths [12, 13], and for micro-machining fibre sensors [14]. However, the application of the 157nm laser for direct FBG inscription has not been widely reported other than brief mention in conference presentations [15,16], though it has been the subject of theses [17,18].

Here we describe experimental results on writing FBGs in standard telecommunication fibres by 157nm F₂ laser exposure with a VUV phase mask. It is shown that strong gratings (> 80% reflectivity) with narrow linewidth (~0.18nm) can be formed in pristine (non-sensitised) Corning SMF28 and HI 980 fibre using a modest VUV dose.

2. Experimental setup

Our experiments used a Lambda Physik LPF 202 F₂ laser producing up to 35mJ in an 11ns full-width at half-maximum (fwhm) pulse and a pulse-repetition-rate of ≤20Hz. Fibres were mounted in a chamber that was initially evacuated to ~10^-5 mbar and then filled with high purity Argon at 200mbar. Working in a gaseous environment, rather than vacuum, had the advantage of reducing contamination on the VUV optics and also of restricting the maximum temperature rise of the VUV exposed fibre [19]. Prior to mounting on a V-groove holder in the chamber the fibre was stripped of its jacket mechanically and the cladding cleaned with isopropanol. Gratings were inscribed using a Λ=1060.4 nm period, zero-order suppressed, CaF₂ phase mask (StockerYale Inc), placed in contact with the fibre cladding. For 157nm illumination we measured a relative energy distribution in the transmitted orders of 4% in zero order and 43.3%, 2.8%, 1% and 0.74% for each of the ±1, ±2, ±3 and ±4 orders respectively, the ±5 and ±6 orders falling below the detectable level. The F₂ laser has good temporal coherence on account of its narrow linewidth [20], but poor spatial coherence when operated with a multi-mode stable resonator as here. The fibres were oriented in the direction of the narrow dimension of the rectangular output beam with the grating grooves normal to this direction. The reduced beam divergence in this direction gave an estimated coherence width of 53µm at the mask. Beyond the mask coherence between ±1 orders was maintained over a calculated range of ~300µm, for propagation in fused silica cladding of refractive index 1.67.

A cylindrical lens converged the 157nm beam onto the phase mask to increase the fluence, and the beam size was restricted by an aperture to expose 8mm of fibre with approximately uniform fluence. Grating evolution was monitored in reflection with a spectrum analyser (Anritsu Model MS 9710A, resolution 0.07nm) located outside of the chamber and connected to the sample fibre via a vacuum feed-through. The fibres were removed from the chamber in order to measure their spectral transmission.

3. Results and discussion

Figure 1 shows results for the space-averaged (DC) change in effective refractive index, Δn_eff, and the spectral width of the grating reflection, Δλ, (fwhm) versus dose for SMF 28 fibre. Δn_eff was derived from the measured wavelength shift, δλ, of the reflection peak (Fig. 2) using Δn_eff=δλn_eff/λB, given n_eff=1.4466 from the initial Bragg peak at λ_B=1534.1 nm. Δn_eff is seen to initially increase rapidly and then more steadily with dose, reaching Δn_eff=2.8×10^-4 at 4500 Jcm-2 by which point the change has begun to saturate. The linewidth was 0.12 nm at the lowest dose a grating could be measured, rising to ~0.22 nm at 4500 Jcm^-2. The corresponding magnitude of the peak reflection for this grating (Fig. 3) reached 93% at the maximum dose of 4500 Jcm^-2. The absolute reflectivity, R, in this case was calibrated by measuring the grating transmission, T, and using R=1-T on the assumption of negligible absorption. Figure 2 shows a well-defined reflection spectrum for the FBG, though strong side-lobes are absent suggesting some degree of apodisation occurred, possibly because of fluence non-uniformity over the 8 mm long grating. The transmission, Fig. 2, showed not only the grating dip but a strong loss feature located ~1.3nm lower. Similar features have been recently reported for gratings written in low sensitivity fibres using a fs UV laser and attributed to asymmetry in the index change in the vicinity of the core that results in coupling to cladding modes [21].

Assuming the grating is linear with uniform cosinusiodal index variation, the standard expression for R [22] can be used to find the modulation amplitude of the refractive index change, Δn_ac. This is shown in Fig. 3 for SMF28 fibre, with the value of Δn_ac=1.24×10^-4 obtained at R=0.93 corresponding to an effective fringe visibility, s=Δn_ac/Δn_eff≈0.44. Although the majority of the diffracted energy from the phase mask is in the ±1 orders (~86%) which produce Λ/2 fringes, where Λ is the mask period, the fraction remaining in the zero order and higher orders can have a significant influence on the fringe quality [23]. Modelling of the mask inteference field using the measured order efficiencies and taking account of finite spatial coherence is shown in Fig. 4. Order mixing results in an x-periodicity that is predominantly Λ or Λ/2 depending on the distance from the mask [23], but when averaged over the 8µm core diameter the fringes are effectively spaced by Λ/2 with a y-averaged optical visibility s≈0.75. Support for the effective in-fibre fringe spacing being Λ/2 comes from the spectral resolution which is within 20% or so of that expected for an 8mm long grating of period Λ/2=530.2 nm.

The saturation of Δn_eff evident in Fig. 1 may arise because of competing positive and negative index change mechanisms in silica with 157nm radiation [24]. If this non-linearity was also reflected in the local index response then the smoothing effect would reduce index fringe visibility compared to the optical fringes. In fact, when a power law fit to Δn_eff (Fig. 1) was applied to the simulated fringes in Fig. 4, a depth-averaged visibility of s≈0.33 was obtained (c.f. 0.44 deduced above).

 

Fig. 1. Linewidth of spectral reflection response and effective index change Δn_eff versus dose for SMF-28 fibre exposed using the 157nm F₂ laser. Uniform grating of length =8mm; exposure fluence =58mJcm^-2 per pulse.

Download Full Size | PDF

At 157 nm there may be significant absorption in the cladding, even when this is ‘pure’ fused silica and it is possible that the gratings formed are not solely confined to the doped core. The reported photosensitivity for fused silica at 157 nm is Δn~1×10^-4 for a dose of 4500 Jcm^-2 [10] and as an estimate, for a mode confinement factor of η=0.8 for SMF 28, the contribution to index change from the cladding would be (1-η)Δn≈2×10^-5. This is small compared to our result of Δn_eff=2.8×10^-4 at this dose for SMF 28, suggesting index change is principally confined to the core. If photosensitivity is entirely attributed to the core, the corresponding index change is Δn_c=Δn_eff/η=3.4×10^-4. This can be compared with Δn=8×10^-4 reported by Chen et al [9] for a 157 nm dose of 4500Jcm^-2 and 25mJcm^-2 per pulse based on the Bragg shift of FBGs pre-written using a KrF laser in SMF 28 that had been treated with H₂ and then annealed. The 157nm photosensitivity for the fibre core falls well below Δn~10^-3 measured for a cladless 3% GeO₂ doped planar waveguide [10] that has similar composition to the core of SMF 28 fibre. A likely reason is that the dose at the core is significantly less than at the fibre surface itself because of cladding absorption. This has been noted previously [11] and measurements at 157 nm [19] indicate that loss can exceed 50% over the cladding depth depending on the fibre type. This will tend to offset any fluence gain through focussing effects in the circular cross-section fibre. The true core photosensitivity at 157 nm may thus be significantly higher than implied by our results, as no correction has been made for cladding absorption. Non-uniformity of exposure over the core cross-section (Fig. 4) could also degrade the effective fringe visibility and lead to transmission loss peaks through cladding mode coupling (Fig. 2).

 

Fig. 2. Spectral transmission and reflectivity of fibre Bragg grating written in SMF 28 fibre using the 157nm F₂ laser. Grating length =8mm; fluence =58mJcm^-2 per pulse.

Download Full Size | PDF

Experiments were also carried out for pristine Corning HI 980 fibre exposed at 157 nm. As is evident from Fig. 3, the peak grating reflection exceeded that for SMF 28 at doses ≤900 Jcm^-2 but thereafter increased less rapidly, with a maximum of R=81% being reached at 6000 Jcm^-2. For this high index fibre, values of Δn_eff=3.4×10^-4 (Δn_c=6×10^-4 for η≈0.55) and s=0.26 were obtained at 6000 Jcm^-2. The lower fringe visibility in this case possibly arises from the greater 157 nm beam attenuation in the more highly doped core resulting in non-uniform exposure over its cross-section.

It is of interest to consider whether there is an advantage to be gained working with the VUV F₂ laser rather than the 193nm ArF laser which is already known to be suitable for FBG inscription in unloaded standard fibres [25, 26]. It has been reported, for example, that a Δn≈1.7×10^-4 is produced in unloaded SMF 28 fibre at a 193nm dose in the range 3600-7200 Jcm^-2 (comparable to that used here) but delivered at a higher fluence of 0.2 - 0.4Jcm^-2 per pulse [26]. Evidence from fluence scaling suggests photosensitivity at 193nm is driven by a two-photon mechanism [25], whereas at 157nm increases above a quite modest fluence level can actually reduce photosensitivity [9]. This effect is confirmed in our own investigations [18] but is not yet understood, though it may relate to a balance between defect creation and damage at the core-cladding interface [9] or competing index change mechanisms with VUV pulses [24]. Even at the ‘optimum’ fluence it seems likely that enhanced photosensitivity seen in unclad waveguides with VUV exposure will be off-set in fibres because of cladding loss. In contrast, such loss is essentially negligible at ≥193 nm and this, coupled with the ease of use in normal laboratory atmosphere makes 193nm attractive from a practical perspective.

 

Fig. 3. Comparison of spectral reflectivity of FBG in HI-980 and SMF-28 fibre versus 157nm laser dose. Fluence per pulse ~58mJcm^-2. The solid line represents the modulation amplitude of refractive index for SMF 28 fibre.

Download Full Size | PDF

 

Fig. 4. In-fibre optical energy density (power density) as a function of normalized distances x/Λ and y/Λ parallel and normal to phase mask surface respectively. y/Λ runs from 58-66 to nominally overlap the fibre core. Mask period Λ=1060nm, un-polarized 157nm beam with full-angle divergence =0.003rad. Interference field calculated taking account of vector fields for s and p components using fibre refractive index =1.67 at 157nm. Phase mask order efficiencies based on energy distributions: zero =4%, ±1 order =43.3%, ±2 order =2.8%, ±3 order =1%, ±4 order =0.74%, ±5 and ±6 orders =0%.

Download Full Size | PDF

4. Conclusions

In summary these results show that useful quality FBGs can be written in SMF 28 and HI 980 fibre with the 157 nm F₂ laser at low dose and without the need for sensitisation. The findings point to potential advantages that might be gained using the VUV laser, though further work is needed to better understand the photosensitivity mechanisms. Cladding absorption is suggested as a limiting factor in realising the full benefit of VUV laser inscription and more work is required to quantify this and the extent to which absorption in the doped core leads to grating asymmetry. Additionally, modelling shows that contributions from unwanted orders in the phase mask interference field will produce significant non-uniformity over the core diameter that may also contribute to cladding coupling loss.