1 Introduction

Laser-produced plasmas (LPPs) have been one of the most successful applications of lasers since the first advent of Q-switched lasers [1]. Such plasmas have found application across a wide range of physics and chemistry, from fundamental spectroscopy to inertial confinement fusion. Recently, the LPPs have been found to be necessary emission sources in industry for extreme ultraviolet (EUV) lithography [2]. The LPPs have been utilized as excitation sources to demonstrate vacuum UV (VUV) laser oscillation using femtosecond LPPs [3, 4]. We have been working with LPPs as intense emission sources in the VUV and EUV spectral regions. As an EUV emission source, we have demonstrated a 13.5-nm emission source with a conversion efficiency of over 2 % using a Li LPP with dual laser pulses [5]. A spectrally continuous VUV emission source in the wavelength between 120 and 200 nm has also been demonstrated in an Ar LPP as a spectroscopic emission source to measure VUV absorption of materials [6].

We have been developing broadband EUV emission sources for surface spectroscopy, in which adsorbed molecules on surfaces are to be desorbed and dissociated as a result of absorption of the wavelength-selected high-energy EUV photons. The desorbed and dissociated atoms and molecules are detected by a mass spectrometer. This photon-stimulated desorption spectroscopy (PSDS) should have superior characteristics over conventional thermal desorption spectroscopy (TDS) in terms of high-energy resolution, less heat effect, and so forth. Basic surface chemistry of various organic materials, aiming to the PSDS applications, has been investigated using synchrotron radiation [7]. In practical spectroscopy, however, the use of LPP emission as a broadband light source may be more realistic than the use of the synchrotron radiation. It is thus necessary to have high-power, continuous LPP emission sources in the EUV for the PSDS. It is well known that continuous EUV/X-ray emissions can be produced in the LPP using a high-Z element as a target. The use of a high-Z alloy target may be appropriate for such emission sources. Several groups have already pointed out that the use of alloy elements could even increase the LPP temperature by increasing plasma opacity [8, 9]. Using the alloy target, low-opacity regions of one material were filled with high-opacity regions of another material, resulting in the temperature raise of the LPP. This opacity effect was significant in the X-ray spectral region rather than the EUV spectral region [8–11]. Even in the EUV spectral region, however, the opacity effect should improve the emission intensity.

In this article, we found that the LPP continuous EUV emission intensity in the wavelength between 40 and 200 nm for a copper–tungsten alloy target was observed to be 1.3 times higher than that for a tungsten target. The intensity increase was explained by the opacity effect of the alloy target, in which the increase of a calculated Rosseland mean opacity compared with either of the constituents was demonstrated at a certain plasma temperature. This opacity effect could raise the LPP temperature, resulting in the increase of the broadband EUV emission intensity.

2 Experimental setup

Figure 1 shows the experimental apparatus for emission measurements from LPPs. A Q-switched Nd:YAG laser at the wavelength of 1,064 nm produced a maximum energy of 500 mJ with a pulse width of 10 ns (FWHM). The laser was operated at a repetition rate of 10 Hz. The laser beam was focused by use of a lens with a focal length of 17 cm, and was irradiated onto a rotating target placed in a vacuum cell with an intensity between 4 × 108 and 6 × 1010 W cm−2. The intensity was adjusted by moving the lens position. The optimum intensity for the EUV emission was thus determined by the size of the plasma as well. A sheet of tungsten (W), copper (Cu), or copper–tungsten (Cu–W) alloy was used as a target. Atomic composition of the Cu–W alloy was 30 and 70 % for Cu and W, respectively. Emissions from the LPP were observed in a direction perpendicular to the laser axis by a photomultiplier (PMT) with a sodium salicylate scintillator coupled with an EUV spectrometer. Typical spectral resolution was approximately 1.0 nm. In this setup, debris from the target was minimized according to the cosθ spatial debris expansion observed in our previous research [12]. The photomultiplier was connected with a boxcar integrator for data acquisition. The boxcar signal was monitored with an oscilloscope and digitized with a data logger.

Fig. 1
figure 1

Schematic diagram of experimental apparatus

Full size image

3 Results and discussion

Figure 2 shows emission spectra using W, Cu–W, and Cu targets in the wavelength between 40 and 200 nm. The optimal laser intensity was 1.3 × 109 W cm−2. These spectra have not been corrected for the spectral response of the detection system. Spectral shapes using the Cu–W and W targets were similar to each other with continuous emission structures, which were appropriate for the PSDS applications. On the other hand, the spectral shape using the Cu target was not continuous and structures were dominant. The emission characteristics using the Cu target thus may not be investigated any further. The wavelength-integrated emission intensity using the Cu–W target in the wavelength between 40 and 200 nm was 1.3 times higher than that using the W target. The intensity difference was not negligible and may be explained by the difference of the W and Cu–W plasma parameters.

Fig. 2
figure 2

EUV emission spectra using W, Cu–W, and Cu targets

Full size image

The absolute power of the wavelength-selected EUV emissions was evaluated. In other words, the absolute power at a certain wavelength diffracted by the spectrometer was measured, which was an actual power that should be irradiated on a sample in the PSDS. A typical power was approximately 150 nW at 120 nm or 10 eV, which corresponded to a photon number rate of 1011 s−1. The rate was high enough to detect desorbed and dissociated atoms and molecules within a sensitivity of a typical quadrupole mass spectrometer utilized in the PSDS that should be less than 103 s−1.

An alloy plasma using high-Z elements has been proposed as an efficient converter of the laser light to X-rays [8, 9]. It was theoretically shown that an energy band of small opacity for one element could be enhanced with that of a large opacity of another element. Therefore, the mean opacity for the mixed elements with an optimal mixing ratio became larger than the opacities of the individual elements. It was experimentally demonstrated that a higher opacity reduced the radiation conduction loss and led to a higher re-emission of the absorbed radiation [8, 9]. The increase of the opacity in the alloy plasma demonstrated in the X-ray spectral region may be utilized to explain the increase of the emission intensity even in the EUV spectral region as shown in Fig. 2. A Rosseland mean opacity has been widely used to describe the radiation transport in optically thick materials, when the matter and radiation are in thermodynamic equilibrium. It is defined as a weighted harmonic mean of the frequency-dependent opacity such as:

$$ \frac{1}{{\kappa_{R} }} = \frac{{\int_{0}^{\infty } \kappa_{\nu }^{ - 1} \left( {\partial B_{\nu } /\partial T} \right){\text{d}}\nu }}{{\int_{0}^{\infty } \left( {\partial B_{\nu } /\partial T} \right){\text{d}}\nu }} $$

where T is the blackbody temperature, B ν is the blackbody spectrum, and κ ν is the frequency-dependent opacity [9].

In order to calculate the Rosseland mean opacity, it was important to evaluate the blackbody temperature and frequency-dependent opacity under our experimental conditions. The upper limit of the plasma temperature was determined using the energy balance between the rate of the laser energy and the rate of heat loss [13]. The calculated plasma temperature was approximately 50 eV at the laser intensity of 6 × 1010 W cm−2, which was the maximum laser intensity used in the measurements. We, therefore, made calculations of the Roseland mean opacity in the energy less than 50 eV.

The frequency-dependent opacity may be computed for bound–bound (b–b), bound–free (b–f), free–free (f–f), and electron scattering (e-s) processes in the hydrogenic approximation [14]. We, however, have utilized only the b–f process to calculate the approximate frequency-dependent opacity. The b–f absorption can be calculated from the photoionization cross sections [15]. This approximation reproduced the behavior of the opacity as a function of the plasma temperature and of the fraction of Gd appeared in [9]. When the approximation was justified in the X-ray spectral region, it should be justified in the EUV spectral region as well, since the photoionization should be one of the primary processes in such a plasma.

Figure 3 shows the Rosseland mean opacity of the Cu, W, and Cu–W plasmas as a function of the plasma temperature. The ordinate shows the relative values because of the approximation. The mean opacity of the Cu–W plasma became larger than that of either the Cu or W plasma at the plasma temperature between 5 and 30 eV. The mean opacity of the W plasma became dominant at the low temperature region less than 5 eV. The Cu plasma opacity increased at the temperature larger than 30 eV. The behavior was reflected on the frequency-dependent opacity calculated from photoionization cross sections of these materials. There was a temperature region in which the Cu–W opacity was larger than those of the Cu and W targets. The opacity effect may explain the increase of the EUV emission using the Cu–W target as depicted in Fig. 2.

Fig. 3
figure 3

Rosseland mean opacity as a function of plasma temperature for three target materials

Full size image

Figure 4 shows the Rosseland mean opacity as a function of tungsten fraction with different plasma temperatures. The plasma temperatures were chosen since the optimum laser intensity to have spectra shown in Fig. 2 should lead to the plasma temperature of approximately 5 eV [15]. Since the pure Cu target became transparent at the low temperature, the fraction of tungsten less than 10 % may not be realistic, and is not shown here. In addition to this, the spectral structure should become different as shown in Fig. 2 as the fraction of W decreases. At the plasma temperature of 5 eV, the mean opacity increased 1.1 times at the fraction of 70 % of W compared to the pure W. At the plasma temperature larger than 5 eV, the mean opacity at 70 % W was always larger than that at 100 % W. But the opacity values at 70 % W decreased as the plasma temperature increased. Therefore, the optimum plasma temperature should be approximately 5 eV at the fixed tungsten fraction of 70 %. The increase of the mean opacity using the Cu–W alloy target may justify the increase of the EUV intensity.

Fig. 4
figure 4

Rosseland mean opacity as a function of fraction of W at different plasma temperatures

Full size image

4 Summary

In conclusion, we observed that the LPP continuous EUV emission intensity in the wavelength between 40 and 200 nm for the Cu–W alloy target was 1.3 times higher than that for the W target. The increase was explained by the opacity effect of the alloy target, in which the increase of the calculated Rosseland mean opacity compared with either of the constituents was demonstrated at the plasma temperature of 5 eV.