1. Introduction

X-ray absorption fine structure (XAFS) concerns the processes, related to X-ray absorption by atoms at photon energies in the vicinity and above the core level binding energy of the atom under consideration. Through the detailed studies of absorption spectra in this energy range, a modulation in the X-ray absorption of an atom is related to its physical and chemical state.

Extended X-ray absorption fine-structure (EXAFS) is a method, that allows one to study detailed information on the local environment surrounding the atom, through the observation of the fine spectral structure on the high energy side of an X-ray absorption edge. From that information, using the theoretical expression for EXAFS [1] the length of the bonds, coordination numbers and chemical structure surrounding the investigated atom can be deduced. This technique has some advantages over the other methods, for example X-ray crystallography [2], such as it does not rely on long-range order in the material, thus the material does not need to exhibit crystalline-like structure, as well as it is element specific, since the measurements are related to a particular atom present in the investigated molecular structure. Thus, it is a valuable tool for examination multicomponent molecular systems [3].

In the lower part of the energy spectrum, spanning the energy range from a few eV below the absorption edge to typically 20-30 eV above the edge in the soft X-ray spectral region, a near-edge X-ray absorption fine structure (NEXAFS) is often performed. The NEXAFS is a well-known and established method employed for a compositional analysis of the samples, yielding information about its elemental composition through the observation of the spectral features in the vicinity of the high energy side of the X-ray absorption edge [4]. In particular, NEXAFS is often used to study the structure of intermolecular bonds of polymers [5]. This is accomplished by probing the electronic transition from the core level to the unoccupied states. Each element has a characteristic core binding energy, thus the NEXAFS spectra contain information which is element specific. Moreover, the energy levels of initial and final states are strongly dependent on the involved molecular bonds [6]. This results in strong spectral features of the near-edge fine structure. The subsequent analysis of these unique spectroscopic fingerprints makes possible the identification and characterization of various kinds of polymers [7].

Initially, for those techniques, synchrotron radiation was used. The first demonstration of EXAFS with synchrotron X-ray emission [8], kindled the interest in this technique. Currently, most of the large-scale facilities have a beamline dedicated to studying the matter using the X-ray absorption fine-structure technique. Synchrotron light allowed to apply EXAFS/NEXAFS methods to a variety of fields of science, studying both organic [9, 10], inorganic [11] and nanoscale [12] samples.

However, it was not long after the synchrotron demonstration that laser-produced plasma light sources were employed for this type of experiments as well. Using a single pulse of X-rays generated by a laser-produced plasma (LPP) source [13] a complete spectrum was recorded in 1 ns. Since then, many examples of laser-produced plasma sources were employed for X-ray absorption spectroscopy. Laser-plasma source based on bismuth and iron solid target, employing 50 J / 1 ns pulses from Vulcan laser system [14], copper rod solid target irradiated by a compact Nd:YAG laser [15], copper target irradiated with thin-disc Yb:YAG laser [16] and an imploded CH target on OMEGA laser system [17] were also drivers for NEXAFS experiments. Initial pump-probe experiments using compact gas-target based SXR source [18] were also recently demonstrated. The solid-state target sources, although bright, have an intrinsic problem related to the debris production. Large-scale laser plasma sources do not allow for developing of the compact system, necessary for the widespread of the NEXAFS technique. A single gas jet based laser-plasma SXR source was also successfully employed for NEXAFS experiments proving the possibility to build truly compact and low complexity systems [19], to study samples in He atmosphere [20], or soil samples [21] for environmental sciences. However, a very long exposure time to obtain a single NEXAFS spectrum, reaching up to 10 thousand pulses [15] is often required, limiting the possibility for high throughput measurements. More recently, high order harmonic generation (HHG) sources were also used for soft X-ray spectroscopy [22] reaching even the “water window” spectral range (λ = 2.3 – 4.4 nm) [23]. Even though the HHG sources made recently a significant progress in terms of conversion efficiency, they require a femtosecond laser source as a driver, thus the complexity of HHG based system dramatically increases. Additionally, due to the physics of HHG, they have difficulty in providing quasi-continuous emission spectrum, required for proper spectral sampling of the near edge absorption spectra to reveal in detail their structure.

Another possibility for a NEXAFS system driver is laser-plasma sources based on a double stream gas puff target, which inject not one but two gasses into the laser-matter interaction region [24]. While the inner gas is the material of the target, to which a specific elemental emission can be attributed, the second gas surrounds the inner gas decreasing the density gradient of the target gas in the direction of the nozzle axis, increasing target density in the interaction region. Such laser-plasma EUV/SXR source based on a double stream gas puff target was already proven to be useful for various applications, including metrology [25], full-field imaging [26], nanoscale microscopy [27], photoionization [28], polymer surface modification [29], radiobiology [30], etc.

In this work, we demonstrate the possibility to perform NEXAFS measurements on thin organic samples with laser plasma SXR source based on a double stream gas puff target. We present the results of employing SXR emission from krypton/helium plasma, to perform near edge fine structure spectroscopy with few second exposure time. As a proof of principle, a 1 μm thick polyethylene terephthalate (PET) sample was used. Attenuation coefficient spectrum was obtained with an exposure time of ~10 seconds and composition of the sample was evaluated using three separated fitting methods to confirm the applicability of compact, laser plasma source, based on a double stream gas puff target to NEXAFS measurements.

As a result, a compact NEXAFS system was developed, desk-top in size, with a footprint of 1.5 × 1.5 m², including a pumping laser head. Simultaneous acquisition of both, sample and reference spectra results in a much more accurate data acquisition, independent on source energy fluctuations and mechanical instabilities of the system. The spectral resolution of this compact system is comparable with early synchrotron-based works. In the following sections, the details about this system will be presented and discussed.

2. Experimental setup

The experimental setup for the SXR compact NEXAFS system using the emission from krypton/helium plasmas is depicted in Fig. 1 and the photograph of the system, indicating all major components is depicted in Fig. 2. An Nd:YAG laser beam, emitted from NL 303 HT laser system (EKSPLA, Lithuania), with laser pulse energy of 0.65 J and 3 ns time duration is focused by an f = 2.5 cm focal length lens onto a double stream gas puff target. The target is formed by a collinear set of two nozzles, driven independently by two electromagnetic valves. The diameters of the nozzles are 0.4 mm for the inner nozzle and 0.7-1.5 mm for the outer, ring-shaped nozzle. The inner nozzle was pressurized with krypton gas (working gas) at an optimum backing pressure of 3 bar, while the outer nozzle was connected to helium pressurized to 5.5 bars. The double stream gas puff target was used to increase the gas puff target density, by shaping the flow of the inner gas into a vacuum through the use of the outer gas. In such case higher inner gas density can be obtained at 1.5 mm away from the nozzle to avoid degradation of the nozzle by a repeatable plasma formation. The valves were driven separately by a dedicated two-channel controller, which is capable of independent adjustment of the delay and opening time for each valve. Moreover, driving signals for both valves are synchronized with the laser power supply, which produces a triggering pulse 1 ms before the laser pulse.

 

Fig. 1 Optical arrangement for obtaining simultaneously reference and sample spectra a) and the scheme of the experimental NEXAFS system employing laser plasma source based on a double stream gas puff target b).

Download Full Size | PDF

 

Fig. 2 Photograph of the compact NEXAFS system with major components indicated.

Download Full Size | PDF

Due to the interaction of the laser pulses with the gaseous target a laser produced plasma is created, which emits radiation in the broad range of wavelengths, from soft X-rays to infrared, depending on the gas used as a target, laser beam and focusing system parameters. In this experiment, an efficient soft X-ray emission from krypton was achieved. The radiation from krypton plasma enters the second vacuum chamber, in which a thin film filters are positioned. The filters are: Ti, 200 nm thick; Al, 250 nm thick and Zr, 200 nm thick. Additionally, the measurements can be performed without the filter, if it is required. The filters can be selected remotely, using motorized, vacuum compatible filter wheel (Standa).

Next, the radiation illuminates the sample, which is being investigated. The sample holder is designed in such a way to allow simultaneously for the SXR light to be transmitted through the sample (sample beam), but a portion of the SXR light (reference beam) also enters undisturbed the entrance slit of the spectrometer, located ~75 cm from the plasma. This can be seen in the Fig. 1(a), which depicts the optical arrangement to obtain simultaneously the sample and the reference beams.

Thus, in this system, in contrary to other compact systems [31] simultaneous acquisition of the two spectra has been achieved. This solution has significant advantages since the system remains unaffected by the energy fluctuations of the source and mechanical instabilities of the system, which may lead to unpredicted spectral shifts and difficulty in calculating the optical density of the sample in the vicinity of the absorption edge.

To accommodate for simultaneous acquisition of two spectra the SXR beams enter the spectrometer through the entrance slit 15 mm in length. Due to the fact that the slit length is large, it was decided to fabricate five slits, approximately 3 mm in length, connected by bridges, which allowed for the slit to have a constant width. The slit was fabricated in a 50 μm thick brass foil by repetitive ablation of the material through its interaction with a focused laser beam. For that pulses from Nd:YAG laser (PL2210/SH/TH/FH from EKSPLA, Lithuania), operating at the third harmonic λ = 355 nm, with a pulse energy of 20 μJ and time duration of 60 ps at the repetition rate of 1 kHz were used. The system was also equipped with an electromechanical shutter, SC-10 from Thorlabs, and galvanometric scanner, SS-IIE-10 [TY] from RAYLASE, Germany. Fifty passes at the scanning velocity of 2 mm/s were sufficient to fabricate the slit arrangement used in this experiment. A set of slits was fabricated this way, with widths ranging from 12 μm to 100 μm, however, in this proof of principle experiment 12 μm slit was employed.

The spectra are obtained using a home-made spectrograph with a grazing incidence diffraction grating from Hitachi High Technologies America, Inc., USA, having 2400 lines per mm made in ZERODUR, model 001-0471 and a CCD camera, in configuration reported in [32]. The arrangement of the SXR spectrometer is depicted in Fig. 3(a), while all the geometrical parameters are shown in Table 1.

 

Fig. 3 Scheme of the grazing incidence SXR spectrometer with geometrical parameters indicated a). Three gasses were used for calibration of the spectrometer: SF₆, N_2, and Ar. The isolated transitions with known wavelengths b) were used to obtain the spectrometer calibration curve c).

Download Full Size | PDF

Table 1. Geometrical parameters of the SXR spectrometer grating (manufacturer’s specification).

View Table | View all tables in this article

The spectra produced by the SXR grating are recorded using a back-illuminated CCD camera (GE 20482048, greateyes, Germany), placed downstream the diffraction grating in accordance with the scheme depicted in Fig. 3(a). The camera has a chip with 2052 × 2046 pixels, each 13 × 13 μm² in size. During the experiments the chip was cooled down to −40°C to reduce its internal noise and the background.

3. Source optimization for maximum SXR emission from krypton/helium laser-produced plasma

For the efficient emission in the SXR region used in the NEXAFS experiment the krypton/helium target laser plasma source was properly optimized. The optimization was necessary from the point of view of acquisition time to record the spectra to be of the order of a few to a few tens of seconds.

Initially, the lens position was optimized with respect to the position of the nozzle. This lens refocusing allowed to obtain a sufficient power density to be delivered to the optimum region of the gaseous target. The optimization of lens position is depicted in Fig. 4(a). A peak intensity curve was obtained considering emission from Kr plasma in the vicinity of 4.5 nm wavelength (see Fig. 5(a) – reference spectrum). A spectral integration curve depicts the change of emission obtained through spectral integration from 1.5 to 5 nm wavelength. The horizontal axis represents the lens position, indicated in Fig. 1 as “x”. The relative value of 0.375 mm represents the optimum position of the lens which allowed to achieve the maximum SXR yield both in peak intensity measurements and in the integrated spectra. The zero position corresponds to lens to nozzle separation equal to the focal length of the lens (f = 25 mm).

 

Fig. 4 Plots depicting optimization of the laser plasma source based on a double stream gas puff target. The optimization is based on obtaining a maximum signal from Kr spectrum. Peak intensity curve was obtained considering emission from Kr plasma in the vicinity of 4.5 nm wavelength (see Fig. 5(a) – reference spectrum). Spectral integration curve depicts the change of emission obtained through spectral integration from 1.5 to 5 nm wavelength. Optimization of a lens position a). The lens position was changed in the direction indicated in Fig. 1 as “x”. Optimization of a valve position b) in the direction indicated as “y” in Fig. 1. Optimization of two gases forming the target: Kr – c) and He – d). Timing optimization of the source e). The opening times for two gasses B = D = 1 ms was kept unchanged, while Kr delay time A and He delay time C, with respect to the laser pulse, was changed.

Download Full Size | PDF

 

Fig. 5 Typical reference and sample spectra from a 1 μm thick PET foil, obtained with 100 SXR pulses.

Download Full Size | PDF

The optimization of the valve position was performed in an orthogonal direction to the previous one. It is because the plasma position in respect to the differential pumping cone has to remain unchanged to avoid obscuring the SXR emission by the cone aperture. The direction of optimization is depicted in Fig. 1 as “y” distance. The result of optimization of a valve position is depicted in Fig. 4(b). The optimum value of 0.3 mm indicates the distance of the valve displacement away from the differential pumping cone. In this case, the laser plasma is formed between nozzle axis and the pumping cone. The explanation for the highest yield at this position is the reabsorption of the SXR radiation by a neutral, cold krypton gas if the focus of the lens is located on the nozzle axis. On the other hand, if the displacement is larger than 0.3 mm, the krypton density in that region is too small to produce intense emission in the SXR spectral region.

Next, the proper values of krypton and helium gas pressures were found by measuring the peak and integrated SXR spectral yields as a function of the gases pressure. Optimization of two gases forming the target: Kr – (c) and He – (d) is depicted in Fig. 4. The curves show that the highest emission from Kr/He double stream gas puff target was achieved at the optimum pressure of Kr equal to 3 bar and He equal to 5.5 bar. The typical curve for He gas optimization is single peaked. Initially, if the pressure of the gas is being increased, the density of the target increases as well. After the optimum value of the pressure is reached, further increasing of the backing pressure causes the target density to decrease due to the fact, that high backing pressure in the valve reservoir prohibits the valve to fully open, because of limited energy stored in the capacitor bank that is discharged through the valve coil while it operates. The Kr gas curve is not typical since it is double peaked. We attribute this behavior to the particular construction of this valve.

Finally, the timing optimization of the SXR source was also performed. For the source to work in a repeatable way the synchronization of the valve opening to the arrival of the laser pulse has to be established. Thus, the laser power supply produces a trigger pulse 1 ms before the arrival of the laser pulse. Within that time both valves have to be properly adjusted. To do that four different time delays are set (two parameters per valve). The first one is the delay time (A) from the moment of arrival of the trigger pulse to the moment of opening of the krypton valve. The second parameter defines the duration of the valve opening (B). The same parameters are for the second, helium valve: C and D, respectively. During the optimization B and D parameters (the valve opening durations) were kept at its maximum values defined as B = 1 ms - A and D = 1 ms - C, to provide high gas densities in the laser-matter interaction area. Two other parameters (A and C) were optimized to achieve the highest SXR spectrally integrated emission yield. The result of that optimization is depicted in Fig. 4(e). From the measurements, the optimum set of timing parameters was found to be A = 100 μs, B = 900 μs (Kr) and C = 200 μs, D = 800 μs (He), respectively. The Nd:YAG laser energy stability for each laser pulse is < 1%, measured as a standard deviation based on 20 pulses. Due to plasma instabilities and mechanics of the valve, the SXR energy, and spectral stability was measured to be ~5%. More details about the timing and valve synchronization can be found in [33].

4. SXR spectrometer calibration

Most important part of the system, except, of course, the source, is a high spectral resolution grazing incidence SXR spectrometer. The calibration of the spectrometer is essential due to the fact that the spectral features near the absorption edge ought to be properly defined and distinguished with energy accuracy of a fraction of an electronvolt. The laser plasma SXR source based on a double stream gas puff target allows one to perform such calibration easily because the change in spectral emission can be obtained by changing the inner, working gas. For that purpose, three different gasses were used: argon, SF₆, and nitrogen. The criterion to choose those gasses was to have in their emission spectra single, well visible, isolated emission lines with well-known energy/wavelength [34]. Such lines are shown in Fig. 3(b) for those three gasses and their wavelength range spans from 1.6807 nm line from F⁷⁺ ion in SF₆ gas, through three well-defined nitrogen lines from 2.377 to 2.878 nm from N⁵⁺ ions, till 4.918 nm line from Ar⁸⁺ ions. Those lines were used to obtain the calibration curve for the spectrometer, Eq. (1). To do that a parabolic function was fitted to the data with R-squared fitting equal to 0.(9)₆668, resulting in maximum wavelength error for 2.489 nm N⁵⁺ line equal to 0.057% and minimum error of −0.004% for 4.873 nm line from Ar⁸⁺ ions.

Additionally, the resolving power of the SXR spectrometer was also estimated. The spectral sampling around 250 eV was measured to be 0.15 eV/pixel. However, for 12 μm slit, which was the one that was used in the measurements, and in the vicinity of the carbon absorption edge, an isolated line at 4.409 nm wavelength (2s²2p³-2s²2p²(3P)3d transition from S⁹⁺ ions from SF₆ gas, photon energy of 281 eV) was observed to have FWHM width of 0.3 eV. Thus E/ΔE was estimated to be ~940. This resolution is defined by the width of the spectrometer slit, but also by the pixel size of the CCD camera. In principle the slit can still be narrowed down to 4-5 μm in width, however, the slit itself cannot be imaged onto a CCD chip as a sub-pixel feature. Thus, an isolated spectral line will still appear as being one pixel wide. Narrower entrance slit will also reduce the photon flux, making necessary to increase the exposure time for each spectrum.

5. Experimental results for near edge X-ray absorption spectroscopy of PET

To gauge the performance of the compact NEXAFS system based on a double stream a proof of principle experiment for obtaining NEXAFS spectra from 1 μm thick PET foil, (C₁₀H₈O₄)_n, Lebow, USA, was performed. The foil was partially covering the elongated aperture in the sample holder to allow both sample beam and the reference beam to enter the spectrometer slit, as depicted in the inset of Fig. 1. The sample S_sam(E) and reference spectra S_ref(E) were acquired simultaneously with the exposure of 100 SXR pulses at 10 Hz repetition rate (10-second exposure) of the SXR krypton/helium source. Typical spectra for PET foil obtained with this system are depicted in Fig. 5.

In the system the reference spectrum spans the wavelength range from 2.5 nm to 5.5 nm, corresponding to the energy of 225 eV to 500 eV. The sample spectrum shows clearly the carbon K-α absorption edge, in the vicinity of 4.3 nm wavelength, above which the absorption is low. The attenuation coefficient spectrum was obtained from the optical density spectrum, taking into account the thickness of sample d and using Eq. (2).

 

Fig. 6 Attenuation coefficient of the sample obtained near the carbon K-α absorption edge showing spectral features which were directly compared to the synchrotron data [5] showing good correspondence and exhibiting the expected features at the characteristic positions.

Download Full Size | PDF

Table 2. Energy positions and assignments of features depicted in Fig. 7(a) and b), taken from [35], in the C-1s NEXAFS spectrum of PET foil, based on synchrotron data.

View Table | View all tables in this article

To validate the experimental measurements the attenuation coefficient spectrum of 1 μm thick PET foil, obtained with the use of the compact NEXAFS system based on a laser-plasma SXR source with a double stream gas puff target, was compared directly with the spectrum of 150 nm thick PET foil, obtained with a synchrotron radiation [5]. The comparison can be seen in Fig. 6, where the correspondence between spectral features in both spectra is visible.

In the NEXAFS the bonding structure of the sample is probed using SXR photons, resulting in peaks in the spectra. The parameters of those resonance peaks, such as their heights and widths, are proportional to the number of transitions in the sample and allow an approximation of elemental composition of the sample. A similar approach was employed for NEXAFS spectra of sulfur [36]. To perform such analysis the peaks, corresponding to certain bonds in the molecular structure, as well as a function describing the profile of the absorption edge, are fitted. For the step, usually, arctan or erf functions are used, while for the peaks a combination of Gaussian and Lorentzian functions are being utilized [37]. To perform the fitting a dedicated software can be used, such as Athena [38]. The software has many interesting and useful features devoted to the processing of such data. We have found, however, certain limitations of the software, related to the automatic fitting of the peak functions to the spectral envelope, thus we decided to use our own Matlab based fitting algorithm, which performs much better in terms of automatic peak fitting to the spectral data. It is based on a nonlinear programming solver that searches for the minimum of an unconstrained multivariable function using the derivative-free method. During the data processing, we found that arctan step function together with Gaussian peak functions is sufficient to perform accurate fitting to the spectral data.

To assign all spectral components of attenuation coefficient spectrum three methods, found in the literature, were employed. In the first method, an arbitrary height of the global, single step function [37] was used, in our case, it was 50%. The result of the fitting is depicted in Fig. 7(a). In the second method, a global step function was also used, however, its height was set to 10%, as depicted in Fig. 7(b). The physical meaning of this approach comes from the fact, that for PET the ratio of absorption lengths around the carbon edge is equivalent to ~10%, according to the CXRO data [39]. The final, third method was to use local step functions for each spectral contribution. This method was developed since it is not trivial to determine the exact position of a single, global arctan step function [37] for the measured data. Thus, to solve this problem and for the asymmetry of the peaks, a combination of Gaussian, Lorentzian, and arctangent curve for every peak was proposed [31]. The height of each step was set to 10% (based on the approach in Method 2) of the Gaussian peak amplitude. The result of the fitting is depicted in Fig. 7(c).

 

Fig. 7 Results of NEXAFS spectroscopy of a 1 μm thick PET foil with a compact desk-top system using three approaches. Measured data points are indicated with circles, thick solid line depicts the fitting. Each contribution to the fitting is depicted by a thin solid line. The classical approach in which an arctangent step function was fitted separately from the Gaussian peaks (a global absorption edge) with a height of 50% - a) and 10% - b). A combined approach [31] in which for each peak a separate 10% step function was employed c). From the peaks composition of the PET sample was subsequently calculated.

Download Full Size | PDF

For the elemental composition analysis, it was assumed that the area under each Gaussian peak curve assigned to certain resonance is proportional to the frequency of occurrence of this binding form of the studied element [40, 41]. This allows one to approximate the composition of the bound elements. All ten spectral components, listed in Table 2, were fitted to the spectrum for each method. The energy positions and assignments of features were based on synchrotron data [35]. The spectral components listed in Table 2 are not only ones that contribute, however, those are the most probable transitions, according to literature, and were used for assignment of the peaks.

For composition analysis, the sum of areas under each fitted spectral component was normalized to 100% to obtain a probability of occurrence of a particular binding form. Not all spectral components from Table 2 were used for this normalization because the contribution of the σ* bonds has been accounted for from the π* orbitals, thus, components above 292 eV were left out of the estimation. Additionally, there is a disagreement in the synchrotron data about spectral component at 291 eV (Table 1 in [35], Okajima et al.) for PET.

Among various sources [35, 42] the other spectral components are usually defined with an energy accuracy of ~0.3-0.4 eV, while this component is assigned as carbonyl only in the data in [35], Okajima et al., while in the other reference it does not exist. Thus, since the existence and assignment of this spectral feature are not well established, we have decided to omit that contribution in the composition estimation.

To properly assign spectral components to the particular element we followed the rules that state that the peak contributions from carbon to other elements are accounted for that element and peaks assigned to more than one transition were counted in equal parts to the relative components of those transitions. For example 50% of the area under the peak at 288.7 eV (π* _{C = C}, π* _{C = O}) was added to carbon and 50% to oxygen. All those approximations result in the fact that the values for the elemental compositions should be considered as maximum content. The results of composition evaluation for PET, using three previously mentioned approaches, are presented in Table 3. By a comparison to the theoretical value, calculated as a percent by weight, w/w %, of the composition, the error for each method was evaluated using a root-mean-square deviation approach defined by Eq. (3):

Table 3. Elemental composition of the analyzed PET sample. Three methods were used for composition evaluation. The global error suggests the best methods are for 10% global arctan step function (method 2), however, lower error in hydrogen estimation was found for 10% step function for each Gaussian type peak fitting curve (method 3).

View Table | View all tables in this article

6. Discussion and conclusions

As can be seen from Figs. 7(a)-7(c) the spectral contributions fitted for each method are changing slightly, to accommodate the spectral envelope. In the peak fitting algorithm, the only assumed parameters were the peak/step positions (energies), the height/width of the step function in methods 1 and 2 and the widths of many step functions in method 3. Other parameters, such as all widths and amplitudes of the peak contributions in all methods and heights of the step functions in method 3, were fitted automatically. This lead to interesting results, such as, for example, the method 2, Fig. 7(b), is the only one that incorporated spectral component at 289.8 eV into the fitting. In this particular case (for this sample) this contribution to oxygen content is very insignificant, however, in cases of other small elemental contributions, such as hydrogen, such contribution would be important to be accounted for.

From the error analysis for each fitting method, the smallest global error was found for 10% global arctan step function (method 2). Lower error in hydrogen estimation was found, however, for 10% step function for each Gaussian type peak fitting curve (method 3). The second method significantly underestimated the amount of hydrogen in the sample, being, at the same time, the most accurate in the estimation of the oxygen content. Method 1, on the other hand, while is the simplest, gives the worst estimation of elemental content in the measured sample.

In conclusion, the proof of principle experiment, showing the applicability of laser plasma source based on a double stream gas puff target to NEXAFS, was demonstrated. The compact, desk-top system based on this source, was developed. A source optimization, details about the system and its application to study a polymer sample, showing its applicability to the near edge X-ray absorption fine structure spectroscopy, was demonstrated. In this work, a well-known sample was studied, however, complicated sufficiently enough to be able to test the system and entire data processing approach to the correct assignment of the peaks and evaluation of the sample composition. The obtained spectrum is comparable to the one achieved with synchrotron radiation and it exhibits the expected peaks at the characteristic positions. The system allows one to obtain the NEXAFS spectrum from the simultaneous acquisition of two spectra (sample and reference) with the exposure time from a few to a few tens of seconds. Simultaneous acquisition of both sample and reference spectra makes possible for much more accurate data acquisition, independent on source energy fluctuations and mechanical instabilities of the system. The construction of the system permits to change the filters remotely and it can be easily updated the allow for the possibility to study up to 8 samples, without the necessity to break the vacuum. The spectral resolution of this compact system, ~0.3 eV, is currently comparable with early synchrotron-based works.

Additionally, the gas puff target approach allows one to change the working gas to illuminate the sample with different emission spectra, currently, the Kr gas was used with an energy range of 250-500 eV. Moreover, different gasses allow one to perform precise calibration of the spectrometer as well. Such compact, desk-top NEXAFS system could provide the possibility to perform test experiments or to check initial and novel approaches to data processing with samples, which are later studied in more detail with X-rays at large scale facilities. On top of that, such compact desk-top system might, in the near future, allow for a broader spread of the NEXAFS spectroscopy to environmental, biological and material sciences. Thus, it will be possible to perform NEXAFS experiments using a desk-top system in the university laboratory or small company, to obtain preliminary data on novel materials and samples, without the immediate need to get beam time on the large-scale facility. This, in turn, might benefit the development of these areas of science and technology in the near future. Additionally, such compact system might allow researchers with limited resources to do research on materials that may be too fragile or have other constraints and limitations that preclude measurements at a synchrotron source.

So far, the instrument allows one to record spectra, which give averaged information about the sample. The acquisition of the spectral data is not yet spatially localized. This is planned in the future experiments. The compact SXR source cannot reach the brilliance of a synchrotron, thus the possibility of spectromicroscopy has to be experimentally verified.