1. Introduction

Each distinct application has its unique requirements on the spectral transmission and absorption. Within the field of laser optics, fused silica has become the most applied representative of high quality and transparent materials in the spectral interval between the ultraviolet and the near-infrared. In spite of the wide and flat spectral transfer function, fused silica shows certain absorption peaks within its transparent spectral range [1]. Well investigated ones are ascribed to the absorption of hydroxyl groups chemically bonded to the silica network. The fundamental absorption band of the OH^- group is located at approx. 2.72µm and the overtones as well as combination modes line up towards shorter wavelengths. A detailed overview on the matter is given by Efimov et al. [2]. Hydroxyl caused attenuation in fused silica mostly affects the transmission in the NIR spectral region. However, the OH^- content also has an influence on other spectral regions such as the UV and VUV. In general, the adjustment of discrete levels of hydroxyl content in quartz for specific applications is one major task in the optics industry. In any case, the necessity of an exact characterization of the OH^- proportion in the fused silica matrix with highest accuracy is motivated by numerous applications within telecommunication, and industrial as well as scientific work.

As a consequence of this OH^- caused absorption fused silica can not be used for some NIR wavelengths unless the manufacturing process can be optimized to produce minimum OH^- material. Some state of the art processes are capable of producing lowest OH^- contents, however, the actual content needs to be determined for the application. Spectrophotometry is commonly applied for this kind of analysis [3]. Nevertheless, the low sensitivity requires long absorption paths in order to detect very low contents of OH^-. The approach presented here is to utilize direct absorptance measurements. Laser calorimetry (LCA) is able to resolve very low absorptance values, as has been demonstrated many times before [4]. By applying radiation with wavelengths at the peak absorption of the OH^- caused resonance, the highest possible sensitivity can be achieved.

This paper reports on the detection of lowest OH^- contents in synthetic fused silica by means of wavelengths tunable laser calorimetry. Furthermore, with a combination of LCA and a photothermal absorption measurement technique the lateral homogeneity of the OH^- concentration can be determined.

2. Measurement procedure

Laser calorimetry is an ISO standardized [5] measurement procedure to determine the absorption in laser optics. It has been developed to characterize both, optical materials and dielectric coatings [6, 7]. This is achieved by a high-precision temperature measurement on the sample of interest. A calibrated set of NTC resistors is applied in order to perform an absolute temperature measurement. With this configuration it is possible to detect increases in temperature well below 1mK. Once the temperature increase of an irradiated sample, the effective heat capacity (m · c_p) as well as the applied laser power (P - see Fig. 1) are measured, the absolute absorptance can be derived. It must be taken into account that LCA is not independent of sample geometry and the location of the temperature sensor on the surface of the sample. But considering these respective specifications, absorption below 0.1ppm can be measured for prominent laser wavelengths.

To achieve a lateral resolution of the absorption and, hence, of the OH^- concentration in a fused silica sample, an additional measurement procedure is required. The thermal lensing technique is a versatile tool to determine the absorption at a test spot (irradiated with a pulsed laser beam) by probing variations of the index of refraction within the bulk material [8]. A separate collimated laser beam (probe laser) is used to monitor the absorption-induced changes in the material. In transmission, the signal of a pinhole-photodiode combination will vary resulting from thermally induced changes of the index of refraction. This variation will either focus or defocus the collimated probing laser beam, which in turn alters the intensity passing through the pinhole in front of the photodetector according to the pulse repetition frequency of the pulsed test laser as shown in Fig. 2. In dependence on the temperature behavior of the material’s index of refraction, the signal will drop or increase after the excitation by the pump laser beam. Considering fused silica, the differential temperature coefficient of the index of refraction is positive: ^dn/dT>0. Therefore, a rise in signal amplitude is expected. A sensitivity study of the PTL method has been performed by Li et al. [9].

 

Fig. 1. Combined laser calorimetric and thermal lens technique (PTL) set-up

Download Full Size | PDF

A direct calibration of this procedure is difficult, since the set of thermo-elastic parameters of each test sample is different. But this test is independent of the sample geometry, which allows for a lateral scan resulting in a relative 2D mapping of the absorptance distribution over the sample. Combining LCA and PTL gives the opportunity to perform an absolute and calibrated measurement which additionally enables the recording of a surface mapping of the tested optic. This combined set-up was used to determine the OH^- content and the respective homogeneity. For the detection of OH^- concentrations with maximum sensitivity, the test laser must supply a wavelength within the peak absorption of the hydroxyl-group bonded in the silica network. Only special laser-active materials supply radiation at the fundamental or overtones of this resonance [10]. Therefore, an OPO system has been incorporated within the LCA test-bench. With this laboratory OPO system, which has been set up in accordance to the work of Ruffing et al. [11], the spectral range of 690-2000nm can be continually scanned. (Since wavelengths and rep.-freq. of the pump laser differ from Ruffing’s set-up, a few OPO resonator parameters were changed) For the investigated samples, the strongest absorption peak of the OH^- overtones within this range is found at 1385nm, which is why the presented measurements have been performed at this wavelength.

3. OH^- Concentration Measurements

In the open literature different values varying between 1370 and 1390nm for the spectral location of the OH^- absorption bands were found [12, 13, 14]. Therefore, it is necessary to first find the wavelength of maximum absorption of the individual sample for the subsequent lateral mapping. Having determined the spectral position of the peak absorptance of hydroxyl caused attenuation, the actual concentration within the fused silica matrix can be calculated:

with sample thickness d, the OH caused extinction α_eff, and the measured absorptance A of the sample. α_eff is deduced from α_OH, which has been reported for the different absorption peaks. For 1385nm, α_OH is 61,9 ^dB/km·ppm [15], resulting in α_eff=15.49 1/cm·ppm. With the here reported set-up, it is possible to find smallest OH^- contents with a sample thickness of a few mm.

 

Fig. 2. Results of the PTL measurement on a fused silica sample.

Download Full Size | PDF

In Fig. 3 an example for the LCA measurement of a fused silica sample with an expected low OH^- content is displayed introducing the sensitivity for OH^- concentration of the set-up. From spectrophotometric transmittance measurements, it was not possible to see any OH^- based absorption for the tested sample thickness. The sample for this test has a thickness of 4.89mm, which along with the parameters necessary for the LCA result (spec. heat: 0.772 J/gK [16], mass: 5.44g, laser power: 703mW) allows for an estimation of the OH^- content. In this case, the application of Eq. (1) resolves to an OH^- content of 1.4ppm. With the given maximum temperature increase, a concentration of considerably less than 1ppm OH^- is expected to be detectable. One could assume that an increased sample thickness will also give a longer interaction path and will contribute to the detected temperature increase. However, since the total temperature of the whole sample is measured, a thicker sample will also add to the total heat capacity m×c_p, which is to be heated with the available laser power. Therefore, in first order the sample thickness does not contribute significantly to the lower detection limit of the OH^- concentration. Also, a few remarks concerning the error budget have to be mentioned.

In general, the LCA measurement is limited by an absolute error of 13%. On top of this value, one has to consider the basic absorption of the fused silica bulk and surfaces. Not the full absorption of 249.7ppm can be attributed to the OH^- caused absorption. By determining the absorption apart from the OH^- absorption peak, this contribution can be assessed. For this test, the wavelength of 1064nm has been chosen, because there is no specific absorbing impurity known for this wavelength. However, a small uncertainty about the basic absorption at 1385nm still remains. From measuring the absolute absorption of a fused silica sample at 1064nm, there is no information gathered to what extent hydroxyl molecules are bonded in the bulk or located on the surface of the test sample. This fact is also accounted for in the estimated error budget. As a final part accounting for the measurement uncertainty, it has been found, that the ambient atmospheric moisture also shows significantly increased absorption within the OH^- caused absorption peak. Hence, the test chamber needed to be dried of any moisture. Avoiding a drop in sensitivity due to the moisture issue, the test chamber has been purged with dry nitrogen, lowering the humidity to a sufficient level. The remaining amount leads to an additional increase of the temperature of approximately 0.4mK. In the case of the presented measurement (see. Fig. 3) this would lead to an error of additional 13%. Also, because of the nitrogen purge, the random temperature drift within the test chamber increases. The inflowing dry gas introduces an increased disturbance resulting in a decreased temperature stability. This can be observed in the measurement curve displayed in Fig. 3. Towards the end of the cooling phase, the noise of the recorded temperature curve is significantly increased. In a non N₂-purged measurement assembly, the noise is in the order of ±100µK. In summary, the OH^- concentration in fused silica can be determined with an absolute accuracy of approximately 30%.

 

Fig. 3. Temperature curve of a LCA measurement for determination of OH^- content.

Download Full Size | PDF

In addition to the laser calorimetric measurement of the OH^- concentration within the fused silica, the thermal lens technique was employed to attain a lateral distribution over the circular sample with a diameter of 25mm. The result is displayed in Fig. 4. This illustration shows a linear scan from the center of the sample towards the edge. The first and central PTL data point at radius 0mm was recorded simultaneously with the LCA data followed by the scan of the sample only by PTL. Due to the manufacturing process the sample shows radial symmetry, and this linear scan represents the hydroxyl caused attenuation over the whole sample. So far only linear scans are possible, due to the geometry of the sample holder for the LCA measurement. By means of a few design changes, an actual 2D mapping will be possible. From this data, within the error budget a constant hydroxyl concentration is observed. Merely towards the edge of the sample the OH^--induced absorption drops to a lower level. Reasons for this characteristic can be traced back to the manufacturing, refining, and storing conditions.

In summary for this combined measurement technique, the lateral resolution depends only on the diameter of the pump laser beam. A spatial resolution far better than that of a spectrophotometer could be demonstrated. In this case the test beam diameter was set to 300µm.

 

Fig. 4. Radial scan of the laser-induced absorption.

Download Full Size | PDF

4. Conclusion

In this paper it has been demonstrated that by means of laser calorimetry smallest amounts of contamination and impurities can be traced and the corresponding content can be measured, even on thin samples. The major feature of this technique is a wavelength tunable laser source. Depending on the production parameters and the corresponding dopants, impurities, and refining procedures, the spectral location of absorbing centers can vary. The presented set-up is capable of probing at the center wavelength of an absorbing impurity, enabling a maximum sensitivity. In addition, a calibrated lateral scan is introduced by combining laser calorimetry with the photothermal thermal lens technique.

Hydroxyl content in fused silica as the major loss contributor in the NIR spectral region was investigated. Within these efforts, a sensitivity of less than 1ppm OH^- in the fused silica network could be achieved. This sensitivity has been obtained for samples with a thickness of less than 5mm and a diameter of approx. 25mm. Additionally to the presented results, numerous other applications can be assisted and advanced by this absorption characterization tool. Any absorbing impurity or dopant within transparent optical material can be found and measured. For example, the offered lateral resolution open an excellent way to investigate in fiber preforms concerning losses and homogeneity before the actual fiber is produced.