Laser-based X-ray and electron source for X-ray fluorescence studies

Abstract

In this work, we present a modification to conventional X-rays fluorescence using electrons as excitation source and compare it with the traditional X-ray excitation for the study of pigments. For this purpose, we have constructed a laser-based source capable to produce X-rays as well as electrons. Because of the large penetration depth of X-rays, the collected fluorescence signal is a combination of several material layers of the artwork under study. However, electrons are stopped in the first layers, allowing a more superficial analysis. We show that the combination of both excitation sources can provide extremely valuable information about the structure of the artwork.

1 Introduction

The incessant development of laser technology in the last decades providing systems with a growing energy per pulse, shorter pulses, and extremely robust with respect to the experimental parameters has paved the way to new and unexpected applications in several scientific disciplines. In fact, one can say, without fear of being deceived, that laser technology development has become a leading force for the progression of new laser-based tools that get advantage of all these new capabilities. One of the fields that has been clearly reinforced by this progress is laser-based particle acceleration. In simple terms, the acceleration of particles is achieved by focusing an ultraintense laser pulse with a pulse duration of the order of the central wavelength $\uplambda$, i.e., pulses that contain a limited number of optical cycles, in a spot of the order of $\uplambda^2$. In such conditions, the laser energy is concentrated in a spatial cube of the order of $\uplambda ^3$ being possible to achieve intensities up to $10^{17}\,\hbox {Wcm}^{-2}$ with moderate laser powers (GW). In these acceleration mechanisms, the key factor is the achieved intensity in the target point. For intensities of the order of $10^{16}\,\hbox {Wcm}^{-2}$, the electric field of the laser is higher than the electric field that the electrons feel from the nucleus and the atom is ionized. The expelled electrons are further accelerated by the laser and re-injected into the bulk material once the electric field changes its direction. In this process, deep ultraviolet (VUV), extreme ultraviolet (XUV), and X-ray radiation are generated (see for example [1–3] and references therein). For intensities of $\rm 10^{18}\,\hbox {Wcm}^{-2}$, not only electrons are accelerated but also protons, and if the laser intensity increases to $10^{20}\,\hbox{Wcm}^{-2}$ there are mechanisms for the acceleration of neutrons (see [4, 5] and references therein).

In the following, we will concentrate on the production of X-rays and electrons with moderate intensities in the range of $10^{16}{-}10^{17}\,\hbox {Wcm}^{-2}$. In simple terms, this generation process can be understood as follows. When the laser interacts with the target, as the intensity is considerably high, in the early states of the pulse matter gets rapidly ionized. Then, the rest of the pulse interacts preferably with an expanding plasma from the target surface rather than with an electrically neutral target. Accordingly, the description of the process must be done in terms of laser–plasma interaction. Since the intensity is still sufficient, the laser extracts electrons from this plasma, accelerates, and re-injects them into the bulk, producing both Bremsstrahlung radiation by the sudden loss of energy of the re-injected electrons and the characteristics X-ray emission of the material of the target. It is important to notice that the laser electric field does not only re-inject electrons into the target but also accelerate them in the direction of the laser reflection. According to this, this source is capable of providing X-ray pulses and bunches of electrons both with a temporal duration of the order of the laser pulse.

The kind of sources described previously have attracted a large attention lately because of their versatility, as well as their reduced size and price when compared with conventional particle accelerators. One possible application of these sources not explored so far to the best of our knowledge is the use of a laser-based X-rays and electron source as a tool for non-destructive analysis of artworks. X-ray fluorescence (XRF) is a well-established technique that provides valuable information about the presence of chemical elements in a sample (see for example [6–8] and references therein). When the sample is irradiated with X-rays (or electrons), inner electrons of the atoms of the samples are excited. Once the produced vacancies in the electronic configuration are occupied by other electrons, the atoms emit characteristic X-rays that is a fingerprint of each element. Nowadays the importance of XRF is beyond doubt. For example, when applied to paintings, this technique reveals the elemental composition of the pigments used by the artist helping the art restorers to prevent the degradation of the colors as well as to answer questions related to authenticity and provenance. However, the interpretation of the XRF data is not always straightforward. X-ray emission from the instrument itself and/or the surroundings of the artwork under analysis, e.g., the holder, can induce misinterpretation. Another important factor that must be taken into account is the penetration depth of the X-rays. For example, considering that oil paintings have as a basis a drying oil, e.g., poppy seed oil or safflower oil, with densities lower than $1\,\hbox {gcm}^{-3}$, a material like PMMA with a density of $1.18\,\hbox{gcm}^{-3}$ can be used to determine an upper limit for the X-ray penetration. According to [11] 10 keV, X-rays are attenuated up to 90 % in 0.6 cm of PMMA, while for 100 keV this thickness is of around 12 cm. Thus, one must be aware that the obtained X-ray fluorescence is produced not only on the superficial paint layer, but also on those behind the visible one. This difficulty can be overcome if electrons are used as an excitation source.

In this manuscript, we describe the use of a laser-based X-rays and electrons source for XRF analysis. The obtained results allow us not only to study several layers of pigment simultaneously using X-rays as in a conventional XRF technique, but also to study just the surface of the artwork due to the limited penetration depth of electrons. In the following, we will first briefly describe our sources providing a characterization for the generated X-rays and electrons. Then, we will present the X-ray fluorescence of different pigments obtained by exciting them with X-rays and electrons. We will discuss the experimental data underlying the differences between both excitation sources. Finally, we complete our contribution with a brief summary and outlook.

2 Setup

For the production of X-rays and electron bunches, a Titanium-Sapphire femtosecond laser system with a pulse length of 120 fs, operating at a 1 kHz repetition rate, a carrier wavelength of 800 nm, and a circular spatial profile with a radius of 0.6 cm FWHM (Full Width Half Maximum) was used. The laser system can provide up to 7.5 mJ, but for the experiments discussed in this work an energy per pulse of just 0.9 mJ was enough for the generation of X-rays and electrons. This energy was focused into the target by a microscope objective of numerical aperture NA = 0.42 (see Fig. 1). The achieved focal spot measured at low intensity was of the order of $1.5\,\upmu \hbox {m}$ in the horizontal and $1.2\,\upmu \hbox {m}$ in the vertical direction. With this spot size, the expected intensity is of ${\sim} 5\,\times \,10^{17}\,\hbox {Wcm}^{-2}$. However, we must notice that for high-energy conditions the focal spot will be considerable larger due to filamentation (see for example [9] and reference therein). For this situation, we estimate a focal spot with a diameter of the order of $100\,\upmu \hbox {m}$ and an intensity of ${\sim} 1\times 10 ^{14}\,\hbox {Wcm}^{-2}$. It is noticeable that even at such low intensities X-ray and electron bunches are produced. For ensuring a fresh target surface for each laser shot, the target was continuously spinning and moving perpendicular to the laser (see Fig. 1). Without these movements, the laser focus quality will be damaged because the material is ablated for every shot. It is interesting to notice that although for this work we have used copper as a target, it is possible to use any material, e.g., plastics, mylar, or PMMA. For the deflection of the electrons from the X-ray emission direction, we placed a collimator with an aperture of around $1\,\hbox {cm}^2$ and a pair of magnets (see Fig. 1). To verify the direction of the electron bunches once deflected, we used Gafchromic films (EBT2). The X-ray fluorescence of the different samples was collected by an Amptek Silicon drift detector (SDD-132) placed close to them. For X-rays, the distance between the copper target and the pigments was 8 cm and between the pigments and the detector 5 cm. When excited by electrons, the pigments were placed over the magnets being the distance between the pigments and the detector of 12 cm. In this case, it was necessary to use a second collimator placed between the sample and the magnets (not showed in Fig. 1) to avoid the background radiation produced by the collisions between the electrons and the magnets.

Fig. 1

Experimental setup: a microscope objective, b copper target, c collimator, d magnets, e sample, f collimator, g spectrometer

Figure 2 shows a typical spectrum of the generated X-rays using copper as a target. We can clearly see two distinctive features. On the one hand, the Bremsstrahlung emission produced by those electrons extracted from the target by the laser, accelerated by the electric field, and finally re-injected. This emission has two different origins: the electrons directly accelerated by the laser and the electrons accelerated by collision mechanisms within the plasma during the plasma expansion. As a result, the electrons energy spectrum is described by a bi-component Maxwell–Boltzmann distribution. We can estimate the temperature of these components by adjusting the produced Bremsstrahlung X-ray spectrum to be $\rm T_{\mathrm{hot}}=9.67\,\hbox {keV}$ and $\rm T_{\mathrm{cold}}=4.38\,\hbox {keV}$, respectively [12, 13]. On top of the Bremsstrahlung emission, we can see the characteristics $\rm K_\alpha$ and $\rm K_\beta$ lines of the target material, Cu in this work, at 8.0 and 8.9 keV, respectively [10]. When the accelerated electrons are re-injected into the plasma, they have acquired enough energy for ionizing the atoms of the target material. More specifically, via electron impact ionization an electron from an inner shell is promoted into the vacuum level producing the characteristic X-ray emission when the electron vacancy is filled by another electron. It is important to note that this dependence on the target material allows us to “tune” both the characteristic X-ray emission and the Bremsstrahlung. As a general rule, denser materials produce more energetic Bremsstrahlung than lighter ones.

Fig. 2

X-ray spectrum obtained from a cooper target. The Bremsstrahlung emission has its origin in an electron energy spectrum described by a bi-component Maxwell–Boltzmann distribution with temperatures $\rm T _{\mathrm{hot}}=9.67\,\hbox {keV}$ and $\rm T_{\mathrm{cold}}=4.38\,\hbox {keV}$. The characteristics $\rm K_\alpha$ and $\rm K_\beta$ lines of cooper with energies of 8.0 and 8.9 keV are also visible. The acquisition time was 6 min

As was stated in the introduction, we can assume for X-rays of energy of 10–100 keV a penetration depth in oil paintings larger than 0.6–12 cm, respectively. For electrons, the situation is more complex. Due to the strong energy loss of electrons when they travel through any material, even air, only those with the higher kinetic energy are capable to reach the sample. For example, to pass through 5 cm of air, electrons with a kinetic energy of 60 keV or higher are needed [14]. If we assume as an extreme case that electrons of 60 keV excite the sample, according to [14] the CSDA (continuous slowing down approximation) range for PMMA is $0.006\,\hbox {gcm}^{-2}$. Thus, the electrons penetrate up to $50\,\mu \hbox {m}$ in the material.

3 Experimental results and discussion

For obtaining the experimental results discussed in this section, we proceeded as follows. We used as pigments emerald green and cerulean blue (oil pigments Marie’s, usual composition $\hbox {Cu}(\hbox {CH}_3\hbox {COO})_2\cdot 3\hbox {Cu}(\hbox {AsO}_2)_2$ and $\hbox {CoO}\cdot \hbox {nSnO}_2$ [15]) over watercolor paper in four different samples: green or blue solely, green over blue, and blue over green (see Fig. 3). In a first set of measurements to obtain reference spectra, we exposed directly the samples to electrons and X-ray radiation. For this, we removed the deflecting magnets shown in Fig. 1. Also, we obtained a spectrum without any sample to calibrate the X-ray emission of the materials of the setup. Then, we measured the composed samples with X-rays and electrons, comparing the results with the reference spectra. The acquisition time for the different measurements was around 15 min for X-rays and 25 min for electrons.

Fig. 3

Samples used for the measurements. We used as pigments emerald green and cerulean blue over watercolor paper: a solely green, b solely blue, c green over blue, and d blue over green

Figure 4 shows a spectrum obtained without sample. This measurement allowed us to know the background produced mainly by the X-ray fluorescence of the materials the source is made of, i.e., post-holders, optical mounts. We identified the $\rm K_\alpha$ and $\rm K_\beta$ emissions of the following elements: Cr 5.4 keV and 5.9 keV, Fe 6.4 keV and 7.1 keV, Ni 7.5 keV and 8.3 keV, and Cu 8.0 keV and 8.9 keV [10]. All these materials were present in the experimental setup. At 2.7 keV, we can see the emission lines of Ar in air (natural abundance 0.934 %). Argon is the only element in air with emission lines that are energetic enough to be observed in our detector (lower limit of 1 keV). Since this peak is present with the same magnitude in all the samples, we have normalized all spectra to it.

Fig. 4

Background spectrum

Figure 5 shows the obtained XRF results for the blue and green pigments (see Fig. 3a, b), as well as for the watercolor paper. The structure of the obtained spectra is very rich containing peaks from the background already identified in Fig. 4, the pigments, and the watercolor paper. Analyzing Fig. 5 from low toward high energies, the first unidentified peaks are seen at 3.7 and 4.0 keV. We assigned these emissions to the $\rm K_\alpha$ and $\rm K_\beta$ lines of Ca. Our hypothesis is that Ca is present in the form of calcium carbonate ($\hbox {CaCO}_3$). This compound is normally used in the pigment industry as a whitening agent. Interestingly, for the green pigment, this element was not identified although we used the same watercolor paper for both (blue and green) samples. Different hypothesis can be proposed to explain this fact. Likely, the green pigment has a strong absorption at these energies preventing the detection of the Ca lines. For the blue pigment, either it does not absorb these energies or it absorbs them, but the pigment has been whitened with a Ca compound.

Fig. 5

Reference X-ray fluorescence spectrum for the blue and green pigments and for the watercolor paper

Continuing the analysis of Fig 5, we clearly identified, for the blue pigment, at 4.5 and 4.9 keV the $\rm K_\alpha$ and $\rm K_\beta$ lines, respectively, of Ti. Titanium is normally used in the pigment industry in the form of titanium dioxide ($\hbox {TiO}_2$). This compound is usually known as titanium white. At 8.5 and 9.5 keV, the $\rm K_\alpha$ and $\rm K_\beta$ lines of Zn are visible. This element has been detected for both pigments. Note that it was not present in the watercolor paper, although for the green concentration is significantly higher than for the blue. Zinc is normally presents as zinc oxide (ZnO) and is usually known as zinc white. Finally, at 17 and 17.9 keV, no reasonable element assignment could be made. According to this, we attributed these signals to pile-up events of the Zn lines. Pile-up happens when the detector is not capable to discriminate between two events that are close in time. In this case, rather than two events the detector emits a signal corresponding to a single event but with double the energy. Unfortunately, this is a common problem for particle and radiation spectrometers that becomes specially relevant for pulse radiation due to the high instantaneous fluxes achieved. Nowadays, due to the fast development of laser-based ultrashort particle and radiation sources, there is a growing interest in this instrumental artifact.

According to the chemical composition of the pigments, it was expected to find the emission of Co and Cu for the blue and green colors, respectively. In the first case, the $\rm K_\alpha$ emission of Co lies at 6.9 keV which is close to the much more intense background $\rm K_\beta$ line of Fe located at 7.1 keV (see Fig. 4). For Cu, the situation is similar. There is a strong background emission from the setup and the target itself, hiding probably the Cu emission of the pigment. It is also possible that the amount of these elements necessary to generate the color is tiny, being therefore below our detection limit. Alternatively, it is plausible that the color was generated adding organic pigments like, for example, indigo $\hbox {C}_{16}\hbox {H}_{10}\hbox {N}_2\hbox {O}_2$, whose X-ray emissions lie below 1 keV and are normally not detected by XRF techniques.

Figure 6 shows the results obtained for a green layer of paint over a blue one (see Fig. 3c) when it was excited by X-rays and electrons separately. For comparison purposes, an XRF spectrum of solely the green pigment when excited by X-rays is also shown. It must be mentioned that, although X-ray measurements are normalized to the Ar signal as discussed above, for the electron measurements this is not possible. The absence of the Ar signal can be due to the second collimator used that reduce the signal collected by the detector (see the description of the experimental setup in Sect. 2), and the different excitation yields of Ar for X-rays and electrons. Let us focus on the Ti lines at 4.5 and 4.9 keV. As was discussed previously, this element is present in the cerulean blue but not in the emerald green pigment (see Fig. 5). However, for the composed sample green over blue, one can easily identify the signature of this element. The penetration depth of X-rays for these energies (${\sim} 600\,\upmu \hbox {m}$ for 10 keV) is sufficient for producing the fluorescence not only from the upper paint layer but also from the lower one. Thus, the obtained fluorescence signal is a combination of different layers of paints, complicating, therefore, the identification of the pigments. This difficulty can be overcome if electrons are used as excitation source as shown in Fig. 6. Since the penetration depth of electrons is much lower (${\sim} 50\,\upmu \hbox {m}$ for 60 keV electrons), only the superficial layer is excited and, consequently, the measured spectrum does not have any contribution of the blue color underneath.

Fig. 6

X-ray fluorescence spectrum for the green pigment when excited by X-rays, and for the composed sample green over blue when excited by X-rays and electrons

To obtain further confirmation of the results discussed above, we collected the X-ray fluorescence of green and green over blue when excited by electrons. As is shown in Fig. 7, both curves are essentially identical not showing any evidence of the blue pigment under the green one. Specifically, there is no signature of Ti as can clearly be seen in Fig. 6. This clearly indicated that when complex samples are excited by electrons only, the superficial layer produces an appreciable fluorescence signal.

Fig. 7

X-ray fluorescence spectrum of the green pigment and the composed sample green over blue when excited by X-rays

To finalize the analysis of Fig 6, it must be said that, in contrast to the measurements shown in Fig. 5, the Ca lines at 3.7 and 4.0 keV are visible. We attribute this to the slightly larger collimator used for these measurements. The same argument applies to the large Cu peak at 8.9 keV (compare Figs. 5, 6). This modified setup allowed us to increase the signal but, as a drawback, part of the watercolor paper without pigments was also excited.

Figure 8 shows the obtained results when a blue over green composite sample (see Fig. 3d) was irradiated by X-rays and electrons. In this case, the interpretation is more complex than for the previous sample (green over blue) because of the richer fluorescence spectrum of the blue pigment (see Fig. 5). Let us focus on the ratio between the Zn and Ti $\rm K_\alpha$ lines located at 8.6 and 4.5 keV, respectively. As shown in Fig. 8, when the blue sample was irradiated by X-rays, the fluorescence ratio between the Zn and Ti lines was 0.7. However, when the composite sample blue over green was irradiated, this ratio increased to 2.7, i.e., an increase of ${\sim} 387\,\%$, because of the high concentration of Zn of the green pigment (see Fig. 5). When electrons were used as irradiation source because of their limited penetration depth, this ratio was reduced to 1.2 which is only ${\sim} 70\,\%$ larger than the original ratio. We attribute this ${\sim}70\,\%$ difference to the difference in the relative excitation yields when X-rays or X-rays and electrons are used.

Fig. 8

X-ray fluorescence spectrum for the blue pigment when excited by X-rays and for the composed sample blue over green when excited by X-rays and electrons

4 Conclusions

In this work, we have shown, with a proof of principle experiment, the potential of laser-based particles acceleration techniques in X-ray fluorescence analysis of complex samples. In particular, we have shown the potential of the combined use of electrons and X-rays, both emitted simultaneously from our laser-based source, in carrying out XRF analysis of paint layers. While X-rays give us information about several paint layers as well as the canvas, electrons, because of their limited penetration depth, excite just the most superficial layer. According to this, using our source, it is possible not only to simplify the analysis of the XRF spectra, but also to obtain extremely valuable information about the superficial layers of the artwork.