2.1

Optical properties of UV-fibers

For low-power transmission, only the UV basic attenuation (unit: dB/m) and the UV-induced loss (unit: dB) of a fiber with defined length have to be taken into account. Using (pulsed) laser, the total transmission is additionally reduced by two photon-absorption.

2.1.1

Basic attenuation

In the spectral region at λ < 400 nm wavelength, the basic attenuation, α_B(λ) in undoped silica or silica fibers with an undoped core, essentially depends on UV-intrinsic losses and on Rayleigh scattering /2,3,4/. Using the standard cut-back method /9/, the difference length AL has to be adjusted to the increasing attenuation values and the decreasing dynamic of the measurement system with decreasing wavelength λ. A typical result is shown in Fig. 1.

Fig. 1:

Basic attenuation of a 600 μm core SI fiber due to Rayleigh scattering and electronic transitions; the cut-back method was used with different lengths to overcome the dynamic restrictions of the test system

2.1.3

UV-induced losses

Measuring UV-induced losses, Δa_L(λ) (unit: dB), the wavelength- and position(z)-dependencies of the UV-induced attenuation, Δα(λ,z), play a significant role. With a low-power broadband DUV light-source, the light-guiding core material will be damaged by generating E’ -centers and/or NBOHCs (see Table 1) by UV-photons with energy hv:

In a fiber with a defined length, L, the extent of defect generation decreases along the fiber with position z due to the fact that the fiber behaves as a long pass filter. Therefore, the induced loss, Δa_L(λ), does not increase linearly with the length of the fiber:

In addition, the damage is time-dependent, leading to the following equation for the total loss aL,tot with a well-defined fiber length, L, where tUV is the UV-exposure time:

2.2

Fiber manufacturing

The above mentioned UV fibers are manufactured in several steps, as shown in Fig. 2. A rod of high-OH synthetic silica is produced by flame-hydrolysis /10,12/ using silicon tetrachloride in an oxygen-hydrogen burner. With Plasma Outside deposition (POD) this rod is cladded with silica having a fluorine content of approx. 3%. The obtained UV-preform has a Cladding-Core-Ratio, CCR, typically in the order of 1.1 or smaller /2,10/. Such performs with diameters > 15 mm are drawn into so-called All-Silica-(AS-)fibers, with core diameters between 50 to 1000 μm. To improve UV optical performance, additional preform and fiber treatments can be carried out.

Fig. 2:

Manufacturing steps of UV-fiber with high-OH silica core /1,2,10-12/

2.3

Fiber testing

In Fig. 3, a principle setup for testing the UV damage of fibers during UV irradiation with a deuterium lamp is shown; this setup was used for confirming the transmission improvements in R&D and for quality control in production. All spectral measurements were performed using a standard deuterium (D₂) lamp DO660/05J from Heraeus Noblelight /20/ as a broadband light source and an optimized imaging system. The distal end of the Fiber under Test, FUT, is coupled into a fiber-optic spectrometer (FOS) USB4000 from Ocean Optics /21/ with an UV-enhanced grating for deep-UV used as detector system, DS.

Fig. 3:

Measurement setup for damaging the UV-fibers with the following components: D2: deuterium lamp; LS: lens system; C. SMA-connector; FuT: fiber under test; UV-FOS: fiber-optic spectrometer with deep UV detection; z_F: axial distance between lens system and fiber front face; N₂: nitrogen purge, when required

In order to reduce the noise in the FOS-signal and in the UV-induced loss spectra, either an adjacent average with 9 points or a Fast-Fourier-Transformation smoothing filter with cut-off frequency of 0.36 (arb. units) using 15 supporting points were employed. The entire system for measuring the spectral UV-induced losses at room temperature is described in details elsewhere /22-24/.

Optionally, a gas-tight box, which can be purged with inert gases, e.g. nitrogen, to avoid oxygen absorption below 195 nm, was used. The position of the shutter, Sh, is controlled by software via an electrical TTL output of a computer. Two cases are distinguished. With an open shutter, the photo-darkening of the FuT with the D2 lamp is measured. During recovery after UV damage, the deuterium irradiation is blocked to avoid additional damage. For each measurement point, however, the shutter is opened for 2 ... 5 s, only; within this short period in comparison to the long period of darkness, the UV damage is negligible /22-24/.

The FOS signal, FOSS(λ,t), is proportional to the spectral power densities, at a given wavelength. However, the transfer factor from input power or density to photo current or charge is strongly wavelength dependent. The spectral and temporal UV-induced losses (unit: dB) can be determined as follows:

Either, the spectral losses Δa_L(λ,t_i) at given times t_i or the temporal losses Δa_L(λ_n,t_UV) at given wavelengths λ, are determined.

3.1

UV fibers with high-OH silica core

In 1993, standard UV fibers (type: FVP /11/; 1^st generation) drawn from standard UV preforms /10/ had a very low basic attenuation, with values of < 1.0 dB/m at 200 nm (see Fig. 1). However, these fibers were darkened below 230 nm during UV irradiation with deuterium lamps within 11 h (Fig. 4); the UV induced losses at 214 nm of such 1 m long fibers were between 10 to 40 dB, depending on different manufacturing and testing parameters (Fig. 5). In addition to the high values, the recovery of the fiber was an issue, too: the concentration of transient E’-centers was decreasing with time constants of several hours.

Fig. 4:

Wavelength-dependent transmission, expressed by photocurrent, before and after 11 h of UV-irradiation of a 1.0 m long UV-fiber of 1^st generation; the darkening below 230 nm is due to the generation of E ‘-centers /25/

Fig. 5:

Saturated UV-induced loss after (only) 4 h of UV-irradiation, using UV-fibers from the same draw and a stronger deuterium-lamp; the different graphs are generated with different input conditions /8,24/

The stimulating step was the hydrogen-loading of the standard UV fibers (type: UVI; 2^nd generation), described in /25/ for the first time. The E’-centers generated by UV-light (see eq. 1) were passivated as follows:

Comparing with Fig. 4, these fibers exhibit no UV-induced losses within 11 h of UV-irradiation (Fig. 6). Testing three different fiber samples, the same behavior was confirmed for 11 hours (Fig. 7). Only small differences were observed in the short-term behavior because the precursor and defect concentrations could have been different due to processing of preform and fiber. In addition, similar improvements were found with KrF-laser irradiations (Fig. 8).

Fig. 6:

Wavelength-dependent transmission, expressed by photocurrent, before and after 11 h of UV-irradiation of a 1.0 m long UV-fiber of 2^nd generation; the darkening below 230 nm is avoided due to hydrogen-content in the light-guiding core /25/

Fig. 7:

Comparison of the UV-damages of 1.0m long UV-fibers from 1^st generation (1 piece) and 2^nd generation (three samples) using deuterium-lamp, described by the temporal transmission at 214 nm wavelength (maximum of E’ -center absorption) for 1.0 m long UV-fibers; the defects generated by UV-light and the passivation with hydrogen is indicated /25/

Fig. 8:

like Fig. 7, however the fiber is damaged by a KrF-laser (248 nm) and transmission is monitored at 248 nm, too /25/

This first step of improvement enabled several new applications where DUV light was required. However, for many applications using small fiber diameters the predicted life-time due to hydrogen out-gassing was not sufficient /26-29/.

Because of an increased demands for non-loaded fibers, the preform manufacturing process was modified and a new product was announced in 1999 /30,31/. Using this modified preform without any significant changes in the drawing process, the 3^rd generation of UV fiber became commercially available /10,11/. However, transient defects were still too high leading to signal drifts and uncertainties in the application set-ups.

Finally, additional changes in fiber-drawing and post-treatments lead to a further improved fiber (4^th generation) with lowest permanent and transient defect concentrations /32/, using the proposed 4 h test described in section 2.3.

Over several design generations, the UV damage at 214 nm wavelength of high-OH fibers has been reduced significantly. These steps and the increased UV-stability are shown in Fig. 9 and Table 2.

Fig. 9:

Comparison of the different hydrogen-free fibers in respect to the spectral UV-induced loss after UV-irradiation of 4 hours (saturation); the fiber lengths are different: 1 m for FVP, 2 m for FVP-UVM and FDP

Table 1.

History of different high-OH fibers with reduction of UV-induced losses at 214 nm

Focusing on the two main defect centers, Fig. 10 shows the temporal damage of a commercially available FDP-fiber (4^th generation) with 100 μm core diameter. In the proposed 4 h test (section 2.3), the UV-induced loss at 214 or 266 nm loss in a 2 m fiber sample will be below 0.8 dB or 0.3 dB, respectively. Starting with the FVP-fiber (1st generation), the normalized transmission at 214 nm wavelength increased from 0.1 % to > 90 % for the FDP-fiber. However, a longer UV test of 48 hours was carried out leading to an additional increase of 214 nm absorption band after a first saturation (around 4 hours), as shown in Fig. 11 /33/. For 260 nm band, on the other hand, it is obvious that no saturation can be observed. After 48 hours, both absorption bands seemed to have the same value due to different rates of increases:

Fig. 10:

Temporal UV-induced loss of 2 m long FDP-fibers (4^th generation) samples having <90 pi m and 100 μm core diameter, shown for 214 nm (E’-center) and 266 nm (NBOHC)

Fig. 11:

Long-term fiber damage with UV-light using a deuterium-lamp like in the 4 h test /33/

Further improvements lead to the newest results, shown in Fig. 12. Now, both absorption bands are nearly constant for at least 72 hours or 3 days of continuous UV-irradiation with a deuterium-lamp. These newest results were published in /34,35/. In addition, the optical UV properties can be determined leading to a certified UV-fiber /11,36/, using the described measurement set-up above.

Fig. 12:

Long-term over 72 hours for a further improved FDP-fiber with 100 μm core diameter /35/

3.2

UV fibers with low-OH silica core

On first view /37/, high-OH fibers seemed to be favorite for DUV applications. As seen in Fig.13, the basic attenuation is very high, especially around 214 and 250 nm. These drops in transmission are related to the well-known E’-centers and Oxygen Deficient Centers (ODC) in low-OH material.

Fig. 13:

Spectral transmission due to basic attenuation in low-OH SI-fibers with 200 μm core diameter; E ‘-centers at 214 nm and Oxygen Deficient Centers (ODC) at 250 nm are dominant /37/

Similar to the improvement from 1^st to 2^nd generation of high-OH, hydrogen loading will increase the basic UV transmission per meter or reduce the basic UV attenuation significantly (Fig. 14). The gain in attenuation for this selected sample of a low-OH fiber with 600 μm core diameter was: > 10.0 dB/m at 214 nm and > 5.0 dB/m at 250 nm (Fig. 15).

Fig. 14:

Enhanced UV-transmission of low-OH SI-fibers after hydrogen-loading /37/

Fig. 15:

Spectral gain, normalized to 1.0 m, using the graphs in Fig. 12 /37/

A slight decrease in attenuation at 330 nm due to chlorine modifications is obvious, too. The overall output power in dependence of the wavelength or the spectral attenuation is very similar for low-OH and high-OH fibers. As discussed in /37/, however, the starting performance of the low-OH fiber has a strong influence on the final one, after hydrogen loading. In some cases, the ODCs could not be suppressed totally.

Further processing steps have been added leading to a hydrogen-free low-OH fiber with excellent IR-performance and excellent UV-stability (Fig. 16). The final product FBPI /11,3 8/ is used in several new applications, e.g. in astronomy.

Fig. 16:

Spectral UV-damage of broadband low-OH fiber after 4 h UV-irradiation with deuterium-lamp /38/

5.1

Liquid-Core- Waveguides and flow cells

Some time ago, Liquid-Core-Waveguides (LCW) and flow cells (FC) for UV applications were studied in detail. Using Teflon on the outer side of a silica capillary, a light-guiding waveguide can be built up with all liquids in the capillary bore as long as the refractive index is higher than that one of Teflon /49,66/. In Fig. 21 an optical fiber based flow cell is shown /51/. Using the well-known Lambert-Beer’s law, the effective path length in the LCWs or FC is smaller than the distance (physical path length) between the two optical fibers due to the partial propagation of light in the non-absorbing silica tube (see l_cap in Fig. 21). This issue was discussed in detail in several papers (see /51/).

Fig. 21:

Sketch of a fiber-coupled Liquid-Core-Wave-guide (LCW) or flow cell; due to total reflection at the Teflon-coating, some portion of the light is propagating in the non-absorbing silica capillary /49,51/

Some examples of spectral absorption measurements are given, for acetone (Fig. 22) and toluene (Fig. 24) with different concentrations in pure water. As expected, the absorption of acetone at the maximum absorption at 266 nm is linearly increasing with the concentrations, in the range from 20 mg/l to 300 mg/l (Fig. 23). Above that level, saturation is obvious caused by the dynamic range of the measurement set-up.

Fig. 22:

UV absorption spectra of acetone in water, for different concentrations /66/

Fig. 23:

Concentration-dependent absorption of acetone in water at 266 nm (absorption maximum), based on the raw data in Fig. 20 /66/

Fig. 24:

UV absorption spectra of toluene in water, for different concentrations /66/

5.2

Hollow-Core- Waveguides for gas analyses

With a similar set-up, gas absorption can be measured in the UV-region (Fig. 25). Using short-length fibers, the possible wavelength region can reach 185 nm and below easily; however, the dynamic range is decreasing with decreasing wavelength, mainly caused by stray light effects in the polychromatic detector and in the application. As shown in Fig. 24, the different gases have specific structures in their absorption spectra: the UV “finger-print” of the gas under test. Only ethanol has a non-specific spectrum with only broad absorption band. More details are in /52-54/.

Fig. 25:

Overview about fiber-coupled set-up for measuring gas absorption with long-length Hollow-Core- Waveguides (HCW, /67/) in the UV-region

5.3

Diode-array (DA) Thin-Layer-Chromatography (TLC)

The last example of our selection is the spectral Thin-Layer-Chromatography in the UV-region with diode-array which can only be realized with fiber-optic bundles /45-47/.The separation technique itself including the preparation will not be explained in this paper; details are in /48/.

In Fig. 27, the principle set-up for spectral TLC with fiber-optics is shown: without any analyte (a) on the plate the light from the “emission fiber” will be reflect and/or re-emitted from the surface layer and collected by the “detection fiber”. The numerical aperture and the distance to the plate are the key parameters for the light coupled between the two fibers /46,47/. With analytes (b), the measured spectrum will be changed by the spectral absorption losses in the analytes. For the first time on a TLC-plate, fluorescence can be used for analyses, too. In mode A (c), the fluorescence of the analyte will be collected directly by the “detector fiber”. However, having a fluorescent layer on the plate, this emission light can be modulated by the analyte due to absorption, as seen in mode B (d).

Fig. 26:

Examples of absorption spectra of acetone, nitric oxide, ammonia, ethanol and oxygen, measured in a 1 m long HCW /52-54/

Fig. 27:

Fiber-optic Thin-Layer-Chromatography; principle light-path and emission and acceptance cones of moving fibers (a), absorption mode with absorbing analyte on the plate (b), fluorescence mode A with fluorescent analyte (c) and fluorescence mode B with additional fluorescent layer on the plate plus an absorbing analyte (d)

In Fig. 28, the difference spectrum (in dB) of an absorbing and fluorescent analyte is shown. For positive sign, the light is attenuated due to absorption; for negative sign, the measured signal is higher due to additional fluorescence. As an example, spectral TLC with fiber-optics is capable to distinguish between 16 Polycyclic Aromatic Hydrocarbon (PAHs); using absorption and fluorescence. In addition, the positions of the separation are different (Fig. 29). While only one PAH can be detected easily by fluorescence, 5 PAHs will be optically separated using absorption spectra only.

Fig. 28:

Difference spectrum of an absorbing and fluorescent analyte on the TLC-plate /46/

Fig. 29:

Spectral results with 16 PAHs, having only absorption, only fluorescence or a combination of both /46,47/