BEATS. 1 Introduction

Transcription

1 BEATS A.L. Bacci, M. Bellaveglia, A. Clozza, G. Di Pirro, M. Ferrario, A. Gallo, T. Levato, E. Pace, A.R. Rossi, C. Vaccarezza (Resp.) Laboratori Nazionali di Frascati P. Oliva Sezione INFN Cagliari M. Gambaccini Sezione INFN Ferrara V. Petrillo, L. Serafini Sezione INFN Milano U. Bottigli, P. Delogu, D. Giulietti Sezione INFN Pisa 1 Introduction The Thomson scattering X-ray sources show relevant features for several applications due to the capability of producing intense, quasi-monochromatic, tunable X-ray beams, after collimation, still with a reasonably small size apparatus. Applications to medical physics are straightforward, in particular in mammography where dose control in screening programs is the main relevant issue. The fluence rate is low if compared to those typically achievable by synchrotron sources,but still compatible with the requirements of radiography, moreover the facilities based on TS can be in perspective much more compact and less expensive than synchrotrons, well in excess of one order of magnitude in both items. Hence TS sources represent an appealing alternative to conventional X-ray tubes. The PLASMONX Thomson source combines a high brightness electron beam (SPARC [5]) and a high intensity laser beam (FLAME project [4]) see Fig:1; the source will be able to provide a high flux (up to γ/s) of quasi-monochromatic photons and the mean energy can be varied from 20 kev to hundreds kev by changing the electrons energy. The BEATS experiment is an imaging application of the Thomson radiation used on a standard mammographic phantom devoted to improve the contrast between normal and cancerous tissues while lowering the absorbed dose. Theoretical and experimental studies on the mammographic imaging suggest that the ideal X-ray source for mammography should produce a low tunable energy spectrum, with a narrow energy band, in the kev energy range [1, 2], and that the transfer of the radiological potential of monochromatic sources to a clinical diagnosis is advisable together with the reduction in size and cost compared with the synchrotron facility. In the 2011 the X-ray detection apparatus has been set up for the Thomson source characterization at LNF ; in spring 2012 it will be installed in the SPARC bunker and the first collision experiment is foreseen in fall The Thomson scattering process The Thomson scattering(ts) is the electromagnetic process in which each electron absorbs one (linear Thomson scattering) or more (non linear Thomson scattering) photons from (typically) a laser pulse, emitting one photon. If the electrons are ultra relativistic the scattered radiation looks frequency upshifted and it is emitted forward with respect to the motion of particles, in a small cone of aperture roughly given by the inverse of their Lorentz relativistic factor. The physics of TS is quite complex in the non linear regime which holds when the density of the incoming photons is large enough, i.e. when the laser pulse strength a 0 = (Iλ 2 ) 1/2 reaches unity, being I and λ the pulse intensity and the wavelength, respectively. At intensities above the so-called relativistic

2 Figure 1: CAD drawing of the PlasmonX electron beam transfer lines layout. intensity Iλ 2 = mm 2 W/cm 2 the extremely intense electric field makes the electrons quivering speed approaching the light speed, making the magnetic field relevant for dynamics thus generating a complex particle motion.the computation in the far-field of the scattered photons distribution Ng of pulsation ω can be performed in the classical regime provided that the energy of the electrons is far below tens of GeV, as it is our case. A detailed description of the analytical computation of the scattered radiation distribution valid in the case of a planar, flat-top pulse can be found in [6]; the analytic results show that the spectral distribution of the photons emitted by a single particle is almost completely correlated with the scattering angle and it is composed by a sum of harmonics of the fundamental pulsation: d 2 N γ dωω = V (n, θ, φ)δ(ω nω f ) (1) n 1 ω F = ωl 4γ 2 /(1 + γ 2 θ2 + a 2 0/2) (2) being γ the Lorentz relativistic factor, ω L the laser pulsation, V (θ, φ) a structure function with complex dependance on particle energy, incidence and scattering angles (θ e, φ e ), see [?] What is relevant is that in the linear regime (a 0 1), that is our case, only one harmonic is produced while with a weak non linearity (a 0 1) only a few harmonics are generated. 3 The electron beam The electron beam generation system must be able to produce and transport electron bunches characterized by an energy E beam 30MeV, (corresponding to γ = 60), reaching the focal spot with a transverse size comparable with the laser focal spot size (σ x, y 810µm) and minimizing the hourglass effect [7], in order to allow the optimization of the geometrical overlapping with the laser pulse on the whole interaction duration. These considerations imply severe constraints on the longitudinal energy spread and on the transverse emittance of the electron beam at the interaction point: the SPARC-LAB Thomson source is meant to produce high brightness electron beam able to accomplish to the experiment requirements, the electron beam transfer lines provide the beam transport from the phoinjector up to the interaction point preserving the 6D-phase space

3 characteristics and providing the required final strong focusing for the interaction with the laser pulse. In Fig. 2 the transverse beam rms size evolution is reported for the reference working point setup starting from the photoinjector down to the IP as obtained from the simulations performed with the Tstep code tracking 15 kparticles, for a beam energy of 30 mev, and a quite high value for the starting normalized emittance ɛ nx,y 2.7µm. Figure 2: Rms beams sizes evolution along the transfer line (left), and detail of longitudinal tunability of the beam waist at the interaction point. 4 The source simulation The properties of the photons emitted by the whole electron bunch can be simulated by summing up the intensity contributions of each electron (i.e.in the incoherent regime). To do that we have employed the Thomson scattering simulation tool (T S) 2 code developed by P. Tomassini et al [3]. The code works as follows: the secular trajectory of each particle of the bunch is first computed by neglecting transverse ponderomotive effects (this approximation is fully consistent with the laser pulse and electron bunch parameters considered in this paper, see Tomassini et al. [6]. Since the analytical outcome sketched in Eq. 1 and 2 are valid only for the case of planar long flat-top laser pulse, the code decomposes the pulse in a sequence of single cycles of the laser pulse, each cycle having its own phase shift and intensity. While the particle is moving along its secular path, it interacts with different cycles of the pulse and the coherent summation of the radiation emitted in each cycle gives rise to the radiation emitted during the entire interaction. 5 The TS laser pulse parameter optimization The laser used in this experiment is the FLAME laser at LNF: a Ti-sapphire laser able to deliver pulses with energy up to 6J, whose duration can vary from a few ps down to 20 fs, with a repetition rate of 10Hz [16]. In our case to fit the electron bunch length the laser pulse willbe only partially compressed to attain few ps duration. A well optimized laser is meant to generate the highest X-ray flux while keeping the relative energy spread of the radiation below 20% FWHM for the fundamental harmonic. An additional requirement is that high order harmonics should be as low as possible in order to prevent their enhancement after filtration. Free parameters are: pulse focusing size w 0 (waist size),

4 pulse duration T, acceptance angle θ M of the scattered radiation. Two, competitive phenomena play the major role: at very small focusing size the diffraction makes the laser pulse to spread transversally in a longitudinal size 2Z R = 2πw0/λ 2 smaller than the electron bunch length σ L, making the queue of the bunch to interact with a poorly intense laser pulse; on the opposite side, a too large focusing size reduces the pulse intensity and thus the scattered radiation yield. Further the pulse duration T is linked to the non linear phenomena that appear at high pulse intensity while the diffraction effect imposes an upper limit to T. The requirements for the maximum energy spread and high order harmonics maximum intensity impose strong constrain on the maximum collecting (or acceptance) angle θ M, Since the energy of a scattered photon is almost completely correlated with the scattering angle, see 2, in a linear or weakly non linear regime it comes out that by collecting the radiation within a cone of half aperture θ M = 1/γ an overall energy spread exceeding 50% is obtained. As result of the optimization process described in detail in [8] a laser pulse of waist size w 0 = 15µm, duration T = 6ps, intensity I = W/cm 2 and amplitude a 0 = 0.33 has been chosen to collide with the electron bunch. The backscattered radiation will be collected within a cone of aperture θ M = 8mrad, yielding a flux of photons/shot with an energy spread of 20% FWHM. In Fig. 3 the spectralangular (integrated in the azimuthal angle φ) distribution of the collected radiation is shown.the fundamental at energy about 20 kev and the second harmonics are clearly visible while the third harmonic is much less intense. Note the dependence of the energy on the scattering angle. Figure 3: Spectral-angular(integrated in the azimuthal angle φ) distribution of the collected radiation for the optimized parameters w 0 = 15µm and duration T = 6ps θ M = 8mrad 6 X-ray imaging simulation The X-ray spectrum produced by the simulation code (T S) 2 is used to generate images of a breast equivalent phantom, in order to evaluate image quality. A set of Monte Carlo simulations have been performed to explore the image quality of a mammographic phantom upon the parameter variation [8]; the code described in [9] has been used to generate the images: in the spectral distribution of the Thomson source, Fig.?? a central area with the mean energy E mean = 20.6keV and standard deviation of 1.7 kev has been selected, the fluence is supposed to be uniform over the phantom. The object to be imaged is a phantom made of 50% adipose and 50% glandular

5 tissue. For elemental composition and density of adipose and glandular tissue values from ICRU Report 44 [10] are used. The thickness of the phantom is 5 cm.tumor-like masses of thickness 1, 2, 5 and 10 mm are simulated. Tumor-like masses are supposed to have the same chemical composition of glandular tissue and a higher density (1.044g/cm3) [11]. The considered detector is a digital flat panel detector based on amorphous selenium (a-se). The absorber is a direct converting a-se of 0.25 mm of thickness, with a density of 4.28g/cm 3. These parameters are typical for mammographica-se flat detectors. Other detector layers and structures are supposed to be negligible in the detection process [11]. Noise is considered to follow Poisson statistics. The pixel pitch is 100µm. The image quality is evaluated in terms of dose efficiency or quality factor Q [11], defined as the ratio of the squared signal-to-noise ratio (SNR) to the mean glandular dose (MGD) [?,?]. Hence: Q = SNR2 (3) MGD The dose efficiency Q is expressed in arbitrary units in order to compare the imaging performances of different spectral distributions and the influence of detector resolution and blurring and the effect of any visual system are neglected [11]. In Fig. 4 the dose efficiency calculated for the TS source is reported as a function of detail thickness. For comparison Q values are also reported for monochromatic sources at optimal energy and for the X-ray tube. Figure 4: Dose efficiency of Thomson scattering source, as a function of detail thickness. For comparison Q is also reported for the optimal monochromatic energy and for the X-ray tube for digital mammography. It can be seen that Q values forts source are 5 6% smaller than maximum Q values obtained by monochromatic beams. On the other hand X-ray tube shows dose efficiencies that are about 40% smaller than optimal values. The percent reduction of Q values (with respect to peak Q values for the same detail) for the TS source and for the X-ray tube depends on the discrepancy between the mean energy of the beams and the optimal energy to image the detail. The mean energy of the TS source (20.6 kev) is very close to the optimal energies to image the details (between 20.3 and 20.7keV) while the mean energy of the X-ray tube is only 17.7 kev. The different performances of the TS source and the X-ray tube are also due to the different energy spreads of the two beams: the TS source presents an energy spread of 1.7 kev, while the spectrum of the polychromatic beam differs significantly from zero in the range 10 30keV.

6 7 The Experimental apparatus In 2011 the BEATS apparatus has been set up and installed, a picture of the X-ray beamline is shown in Fig. 5. Figure 5: X-ray beamline, (a) detail of the apparatus for collimation, monitoring and characterization; (b) a view of the complete x-ray beamline The first stage of the X-ray beamline is for x-rays monitoring and characterization, in particular this system consists of: first wide collimation and lead shutter for beam stopping; a rotating collimator holder that allows to reduce the angular divergence of the beam, equipped with six different collimator corresponding to angular acceptance varying from about 9 mrad to 1 mrad; a free-air ionization chamber, used as a x-ray beam monitoring; two additional filter/collimator holders with six available position each, for beam filtration or further collimation; a removable device based on a silicon PIN diode for X-ray flux measurement. In Fig. 5 is also shown the table that will provide the support for imaging detectors and sample. An additional table will be placed downstream the beam to permit the study of imaging techniques such as free-propagation phase contrast that require longer propagation distances. 7.1 X-ray beam monitor The monitoring of the production of x-ray pulses, in order to verify the correct operation and the pulse-to-pulse intensity repeatability, is provided by a ionization chamber that collects the charge produced in air by radiation, without affecting the beam. This device was designed and assembled by Ferrara research unit and has been tested for stability and linearity at Larix Laboratories of Ferrara University, at University of Pisa and at the ELETTRA synchrotron facility (SYRMEP beamline) both with polychromatic and monochromatic beams and continuous and pulsed irradiation. A picture of the chamber and of the electrometer used for acquisition is shown in Fig.6. The

7 minimum number of photons per pulse that produce a readable signal is about 10 6 photons with an average energy of 20 kev. Figure 6: Xray fluence measurement system in detail. 7.2 PIN diode system for flux measurement The current produced in a PIN diode by an x-ray beam is proportional to the rate of energy released in the photodiode active area by the radiation. The rate of energy released in a silicon slab depends on the incident photon energies and the flux. Owen et al. [?] demonstrate that it is possible to measure the flux of a monochromatic x-ray beam impinging on the diode, multiplying the photon induced current by a coefficient calculated from the energy absorbed in the silicon layer. For our measurements a PIN diode HAMAMATSU mod. S operating in photo-voltaic mode (i.e. without applying a reverse polarization) is used. The sensitive area of this detector is 28 x 28 mm 2. The diode is mounted in a metallic box with an entrance window made of an aluminumcoated polymide film to avoid photocurrent production by visible light. The x-ray absorption of the entrance window is negligible in the energy range of interest. A picture of the system is shown in Fig. 7 Figure 7: Silicon PIN diode system for x-ray flux measurement. The photocurrent produced is measured by an electro-meter Keithley mod. 6517B (Keithley Instruments Inc., Ohio, US).

8 Table 1: Diode calibration coefficients K, measured at synchrotron facility and evaluated from theoretical model. E (kev) K (ph/c) K T (ph/c) Ratio ± ± ± ± ± If Q is the charge created by the interaction of a flux ϕ of x-rays with an energy hν ( 35 kev) on a silicon diode, the photocurrent produced is I = ϕq, and the photoconversionratio K can be expressed as: K = ϕ I = ɛ e[1 exp( µ pe t])], (4) where ɛ is the average energy to create an electron-hole pair, e is the electron charge and µ pe is the photoelectric linear attenuation coefficient of silicon. This photoconversion factor K can be evaluated theoretically or measured, for this reason the system was previously tested at the ELETTRA synchrotron facility (Trieste, Italy) with monochromatic x-rays in the energy range between 16 and 24 kev. Diode response has been calibrated comparing its signal to the air-dose signal provided by two suitable free-air ionization chambers. The coefficients K to convert the current produced in the diode to the flux are shown in Table 1 in comparison with the ones predicted by the theoretical model K T, showing good agreement. The minimum number of photons per pulse that produce a readable signal is about photons with an average energy of 20 kev. PIN diode have proved to work properly at high x- ray fluxes (up to ph/s) but in continuous irradiation condition. High instantaneous fluxes of a pulsed source, in the case of BEATS experiment < ph/s, could produce in the diode a charge density extremely high, leading to recombination and partial collection of charge, so to an underestimation of the real number of interacting photons. Sources with an instantaneous x-ray flux as high as needed in order to perform test on our devices are not available, and a preliminary test on a pulsed laser (800 nm) showed the possibility to be in such a regime of partial collection. For this reason is ongoing the measure of diode response coupled to a slow scintillator (CsI) in order to increase the time of charge production in the diode and decrease the istantaneous charge density. 7.3 System for energy distribution evaluation The use of traditional spectroscopic techniques based on single photon detection, either with photomultiplier coupled to scintillator or solid state device, is very difficult because of the extremely high instantaneous flux produced by the source. In fact with this kind of detectors it is not feasible to operate with sub-picosecond data acquisition time. In order to evaluate the energy distribution of the x-rays produced a technique based on the analysis of the diode photocurrent produced by beam filtered with suitable k-edge materials has been implemented [13]. Using filters of Mo, Nb Zr and Al with thicknesses properly selected it is possible to obtain a measure of the energy distribution in an energy range from 16 to 22 kev. The technique was also tested measuring the energy distribution of an x-ray beam having a spectrum similar to the BEATS expected one by using a tungsten anode x-ray tube properly filtered and powered. In Table 2 is shown a comparison of the normalized energy distribution measured with a traditional HPGe detector ϕ(e) and the one measured with k-edge technique ϕ k edge (E). It is possible to notice that the two energy distribution

9 Table 2: Energy distribution of the x-ray tube measured ϕ(e) and obtained from PIN diode current ϕ P D (E) (normalized data). E (kev) ϕ(e) (norm.) ϕ k edge (E) (norm.) < > are in good agreement (maximum discrepancy about 7%). References 1. M. Gambaccini et al, Nucl. Instr. and Meth. in Phys. Res. A 365 (1995) P. Oliva et al, Med. Phys , , P. Tomassini et al, IEEE Trans. on Plasma Science Volume: 36 Iss L.A.Gizzi et al Eur. Phys. J. ST , M. Ferrario et al ICFA Beam Dyn. Newslett , P. Tomassini, A.Giulietti, D.Giulietti, L.A.Gizzi, Appl. Phys. B 80 (2005) R.J.Loewen, SLAC-Report-632,June, P. Oliva et al, Nucl. Instr. and Meth. in Phys. Res. A 615 (2010) P. Oliva et al, Nucl. Instr. and Meth. in Phys. Res. A 608 S106 (2009) 10. ICRU, Tissue substitutes in radiation dosimetry and measurement, Report 44 of the International Commission on Radiation Units and Measurements, Bethesda, MD J.P. Bernhardt, T. Mertelmeier, M. Hohisel, Med. Phys. 33 (11)(2006) R.L. Owen et al, Journal of synchrotron radiation 16 (2009) P. Cardarelli et al, submitted to J App Physics