Laue gamma-ray lenses for space astrophysics: status and prospects Filippo Frontera1 and Peter von Ballmoos2, 1University of Ferrara, Physics Department, Via Saragat 1, 44100 Ferrara, Italy; 2Centre d’Etude Spatiale des Rayonnements, 9, Avenue du Colonel Roche, 31028 Toulouse, France ABSTRACT Wereviewfeasibilitystudies,technologicaldevelopmentsandtheastrophysicalprospectsforLauelensesdevoted 1 to hard X-/gamma-ray astronomy observations. 1 0 1. INTRODUCTION 2 n Hard X-/soft gamma-ray astronomy is a crucial window for the study of the most energetic and violent events a in the Universe. With the ESA INTEGRAL observatory,1 and the NASA Swift satellite,2 unprecedented sky J surveysinthe bandbeyond20keVarebeingperformed.3,4 Asaconsequence,hundredsofcelestialsourceshave 6 already been discovered, new classes of Galactic sources are being identified, an overview of the extragalactic sky is available, while evidence of extended matter-antimatter annihilation emission from our Galactic center5 ] M and of Galactic nucleosynthesis processes have been also reported.5,6 However, in order to take full advantage of the extraordinary potential of soft gamma–ray astronomy, a new generation of telescopes is needed. The I . current instrumentation has relied on the use of direct–viewing detectors with mechanical collimators (e.g., h BeppoSAX/PDS, Ref. 7) and, in some cases, with modulating aperture systems, such as coded masks (e.g., p - INTEGRAL/IBIS, Ref. 8). These telescopes are penalized by their modest sensitivities, that improve at best o as the square root of the detector surface. The only solution to the limitations of the current generation of r t gamma–rayinstrumentsis the useoffocusingoptics. Tostudy eitherthe continuumemissionorthe nuclearline s a emission from celestial sources, Laue lenses, based on diffraction from crystals in a transmission configuration, [ are particularly suited to focus photons in the hard X–/soft gamma–ray (< 1 MeV) domain. As we will show, they show imaging capabilities for on-axis sources. 3 v With these lenses, we expect a big leap in both flux sensitivity and angular resolution. As far as the 8 sensitivity is concerned, the expected increase is by a factor of at least 10–100 with respect to the best non- 0 focusinginstrumentsofthecurrentgeneration,withorwithoutcodedmasks. Concerningtheangularresolution, 3 4 the increase is expected to be more than a factor 10 (from 15 arcmin of the mask telescopes like INTEGRAL ∼ . IBIS to less than 1 arcmin). 7 0 The astrophysical issues that are expected to be solved with the advent of these telescopes are many and 0 of fundamental importance. A thorough discussion of the science case has been carried out in the context of 1 the mission proposal Gamma Ray Imager (GRI), submitted to ESA in response to the first AO of the ’Cosmic : v Vision 2015–2025’plan9 (but see also Refs. 10–12). We summarize here some of these issues. i X r Deep study of high energy emission physics in the presence of super-strong magnetic fields (magnetars) a • The XMM and INTEGRAL observed spectra of Soft Gamma Ray Repeaters13 and Anomalous X–ray pulsars14 leave unsolved the question of the physical origin of the high energy component (>100 keV). A better sensitivity at E>100 keV is needed. Furtherauthor information: (Sendcorrespondence toF.F.) F.F: E-mail: [email protected], Telephone: +39 0532 974 254 Deep study of high energy emission physics in compact Galactic objects and AGNs • A clue to the emission region and mechanism, along with the properties of the hidden black hole, can be obtained with the measurement of the high energy cutoff and its relation with the power–law energy spectrumofthe compactobjects. The currentobservationalstatus is farfromclear(see,e.g., Ref.15,16). Much more sensitive observations are needed, for both AGNs and compact Galactic sources. In the case ofblazars,the gamma–rayobservationsarecrucialforthe determinationoftheir emissionpropertiesgiven that their energy emission peaks at hundreds of keV.17 Establishing the precise role of non-thermal mechanisms in extended objects like Galaxy Clusters • TheexistenceofhardtailsfromGalaxyClusters(GC)isstillmatterofdiscussion,18 and,shouldtheyexist, their originis also an open issue. Are they the result of a diffuse emission or are they due to AGNs in the GC? In the former case, what is the emission mechanism? What is their contribution to CXB? To answer these questions will require much more sensitive observations, like those achievable with broad band Laue lenses. Origin of Cosmic hard X/soft gamma-ray diffuse background • Currently, a combination of unobscured, Compton thin and Compton thick radio-quiet AGN populations withdifferentphotonindexdistributionsandfixedhighenergyspectralcutoff(E )areassumedinsynthesis c models of the Cosmic X–ray background(CXB).19 Is it reasonableto assume a fixed E for these sources? c Aphoton-energydependent contributionfromradio-loudAGNto CXB,like blazars,is generallyassumed. Buttheirrealcontributionisstillmatterofdiscussion. Deepspectralmeasurementsofasignificantsample of AGNs beyond 100 keV is needed to solve these issues. Positron astrophysics • Positronproductionoccursinavarietyofcosmicexplosionandaccelerationsites,andtheobservationofthe characteristic511keVannihilationlineprovidesapowerfultooltoprobeplasmacomposition,temperature, density and ionization degree. The positron annihilation signature is readily observed from the galactic bulge region yet the origin of the positrons remains mysterious. Compact objects - both galactic and extragalactic - are believed to release significant numbers of positrons leading to 511 keV gamma-ray line emissionintheinevitableprocessofannihilation. ArecentSPI/INTEGRALall-skymap5 ofgalactice−e+ annihilation radiation shows an asymmetric distribution of 511 keV emission that has been interpreted as a signature of low mass X-ray binaries with strong emission at photon energies >20 keV (hard LMXBs). A claim for an annihilation line from a compact source (Nova Muscae) was reported in the 90s20 but was neverconfirmed. Muchmoresensitiveobservationsareneededtostudytheannihilationlineorigin,sources and their nature. Physics of supernova explosions • Type Ia supernovae (SNe Ia) are major contributors to the production of heavy elements and hence a criticalcomponentfortheunderstandingoflifecyclesofmatterintheUniverseandthechemicalevolution of galaxies. Because Laue lens telescopes allow the direct observation of radioactive isotopes that power the observable light curves and spectra, gamma-ray observations of SNe Ia that can be performed with this type of instrument are in a position to allow a breakthrough on the detailed physical understanding of SNe Ia. This is important for its own sake, but it is also necessary to constrain systematic errors when using high-z SNe Ia to determine cosmologicalparameters. High resolution gamma-ray spectroscopy provides a key route to answering these questions by studying the conditions in which the thermonuclear explosion starts and propagates. A sensitivity of 10−6 photons cm−2 s−1 tobroadenedgamma-raylinesallowsobservationsofsupernovaeouttodistancesof50–100Mpc. Within this distance it is expected that there will always be a type Ia SN in the phase of gamma-ray line emission, starting shortly after explosion, and lasting several months. In this paper, we review the physical principles of Laue lenses, their geometry, their optimization criteria, their optical properties, the current development status and the prospects for future missions for gamma–ray astronomy. Figure 1. The Bragg condition for constructive interference of a gamma-ray photon beam with the atoms of a given crystallineplane. Leftpanel: Braggdiffractioninreflectionconfiguration(Bragggeometry). Rightpanel: Braggdiffraction in transmission configuration (Laue geometry). 2. LAUE LENS CONCEPT Diffractionlensesusetheinterferencebetweentheperiodicnatureoftheelectromagneticradiationandaperiodic structure such as the matter in a crystal. For a classical textbook on X–ray diffraction see, e.g., Ref. 21. In a Laue lens, the photons pass through the full crystal, using its entire volume for interacting coherently. In order to be diffracted, an incoming gamma-raymust satisfy the Bragg-condition,relating the spacing of lattice planes d with the energy of incident photons E and with the angle of incidence θ with respect to the chosen set of hkl B planes (hkl) ∗: hc 2d sinθ =n (1) hkl B E where d (in ˚A) is the spacing of the lattice planes (hkl), n is the diffraction order, hc = 12.4 keV˚A and hkl · E is the energy (in keV) of the gamma-ray photon. An elementary illustration of the Bragg condition, in two different configurations (reflection and transmission), is given in Fig. 1, where it can be seen that the incident waves are reflected by the parallel planes of the atoms in the crystal. A Laue lens is made of a large number of crystals, in transmission configuration (Laue geometry), that are disposed such that they will concentrate the incident radiation onto a common focal spot. A convenient way to visualize the geometry of a crystallens is to consider it as a sphericalcup coveredwith crystaltiles having their diffracting planes perpendicular to the sphere (see Fig. 2). The focal spot is on the symmetry axis at a distance f =R/2 fromthe cup, with R being the radius of the sphere ofwhich the sphericalcup is a part; f is called the focal length. From the Bragg equation, for the first diffraction order (n =1), it can be seen that the photons incident on a given crystal at distance r (r r r ) from the lens axis can be reflected toward the lens focus if their min max ≤ ≤ energy E is given by hc 1 f hcf E = sin arctan (2) 2d 2 r ≈ d r hkl (cid:20) (cid:18) (cid:19)(cid:21) hkl where the approximated expression is valid for gamma–ray lenses, given the small diffration angles involved. ∗Theindicesh,k,l,knownasMillerindices,aredefinedasthereciprocalsofthefractionalinterceptswhichthelattice planemakeswiththecrystallographicaxes. Forexample,iftheMillerindicesofaplaneare(hkl),writteninparentheses, then the plane makes fractional intercepts of 1/h, 1/k, 1/l with the axes, and, if the axial lengths of the unit cell are a, b, c, the plane makes actual intercepts of a/h, b/k, c/l. If a plane is parallel to a given axis, its fractional intercept on that axis is taken as infinity and the corresponding Miller index is zero. If the Miller indices [hkl] are shown in square brackets, they givethe direction of the planewith the same indices. Figure 2. Geometry of a Laue lens (see text). Conversely,the lens radius r (see Fig. 2) at which the photon energy E is reflected in the focus is given by hcf r=ftan[2θ ] (3) B ≈ d E hkl Rotation around the lens optical axis at constant r results in concentric rings of crystals (see left panel of Fig.3),whileauniformlychangingvalueofr givesrisetoanArchimedesspiral(rightpanelofFig.3). Assuming that the chosen diffracting planes (hkl) of all the lens crystals are the same, in the first case (constant r) the energy of the diffracted photon will be centered on E for all the crystals in the ring, while in the second case (Archimedes spiral), the reflected energy E will continuously vary from one crystal to the other, as shown in Fig. 4. 2.1 Energy passband Any Laue lens will diffract photons over a certain energy passband (E ,E ). From eq. 2, at first order min max diffraction (the most efficient), it results that hcf E (4) min ≈ d r hkl max hcf E . (5) max ≈ d r hkl min Giventhat,forastronomicalapplications,thelenspassbandisdesiredtobecoveredwiththehighesteffective area † and in a smooth manner as a function of energy, the energy bands of the photons reflected by contiguous crystal rings or, in the case of the Archimedes structure of a lens, by contiguous crystals, have to overlap each †The effective area at energy E is defined as the geometrical area of the lens projected in the focal plane times the total reflection efficiency at energy E. Figure 3. The basic design of a crystal diffraction lens in Laue geometry. Flat crystal tiles are assumed. Left: concentric rings of a given radius r concentrating a constant energy E. Right: crystal tiles disposed along an Archimedes’ spiral resultinacontinuouslyvaryingenergyE. Giventhefootprintofthecrystals,theimageinthefocalplanehasaminimum size equalto that of thecrystal size. 0.35 0.3 0.25 y vit 0.2 cti e efl 0.15 R 0.1 0.05 0 280 290 300 310 320 330 Energy - keV Figure4.Anexampleoftheexpectedreflectivityprofileofthreecontiguouscrystalswithamosaicity of1.5arcmin along an Archimedes’spiral. Reprinted from Ref. 22. other, like in Fig. 4. Since the full width at half maximum (fwhm) of the acceptance angle δ (known as the Darwin width) of perfect crystals is extremely narrow (fractions of an arcsec to a few arcsec, see Ref. 21), such materials are not suitable for astrophysical Laue lenses. In order to increase the energy passband of individual crystalsoneuses mosaiccrystalsorcurvedcrystals(see followingsection). The mosaicity ofmosaiccrystals(see Sect. 3.1) and the total bending angle of curved crystals (see Sect. 3.2) govern the flux throughput, the angular resolution and the energy passband of the Laue lenses. The diffracted flux from a continuum source increases with increasing the mosaicity of mosaic crystals or the total bending angle of curved crystals. For a crystal lens telescope, crystals with mosaicities or total bending angles ranging from a few tens of arcseconds to a few arcminutes are of interest. The bandwidth of a lens for an on–axis source is determined by the mosaicity or total bending angle of the individual crystals and by the accuracy of the alignment of the single crystals. By forming the derivative of the Bragg relation in the small angle approximation (2d θ nhc/E), we get hkl B ≈ ∆θ /θ =∆E/E . (6) B B If ∆θ is the mosaicity of the mosaic crystalor the total bending angle of the curvedcrystal,the corresponding B energy passband ∆E of the crystal becomes 2d E2 ∆θ hkl B ∆E = · · . (7) nhc It is worth pointing out that, whereas the energy passband of a crystal lens grows with the square of energy, the Doppler broadening of the astrophysical lines (e.g. in SN ejecta) increases linearly with energy for a given expansion velocity. 3. CRYSTAL REFLECTIVITY Both mosaic crystals and curved crystals are suitable to be used for a Laue lens. We discuss the properties of both of them and their reflectivity. 3.1 Mosaic crystals Mosaic crystals are made of many microscopic perfect crystals (crystallites) with their lattice planes slightly misaligned with each other around a mean direction, corresponding to the mean lattice planes (hkl) chosen for diffraction. In the lens configuration assumed, the mean lattice plane is normal to the surface of the crystals. The distribution function of the crystallite misalignments from the mean direction can be approximated by a Gaussian function: 1 ∆2 W(∆)= exp , (8) √2πη −2η2 (cid:18) (cid:19) where ∆ is the magnitude of the angular deviation from the mean, while β =2.35η is the fwhm of the mosaic m spread (called mosaicity). For the Laue geometry and diffracting planes perpendicular to the cross sectionof the crystaltile (see, e. g., Fig. 2), the crystal reflectivity R(∆,E) is given by Ref. 21: R(∆,E)= Id(∆,E) =sinh(σT)exp (µ+γ0σ) T = 1(1 e−2σT)e−µγT0 , (9) I − γ 2 − 0 (cid:20) 0(cid:21) where I is the intensity of the incident beam, µ is the absorption coefficient corresponding to that energy, γ 0 0 is the cosine of the angle between the direction of the photons and the normal to the crystal surface, T is the thickness of the mosaic crystal and σ is: σ =σ(E,∆)=W(∆)Q(E)f(A), (10) where F 2 1+cos2(2θ ) Q(E)=r2 hkl λ3 B , (11) e V 2sin2θ (cid:12) (cid:12) B (cid:12) (cid:12) inwhichre istheclassicalelectronradius,Fhkl is(cid:12)thest(cid:12)ructurefactor,inclusiveofthetemperatureeffect(Debye- (cid:12) (cid:12) Waller’sfactor), V is the volume ofthe crystalunit cell, λ isthe radiationwavelengthandθ is the Braggangle B for that particular energy, while f(A) in Eq. 10 is well approximated by: B (2A)+ cos2θ B (2A cos2θ ) 0 B 0 B f(A)= | | | | . (12) 2A(1+cos2θ ) B Here B is the Bessel function of zero order integrated between 0 and 2A, with A defined as follows: 0 πt 0 A= , (13) Λ cosθ 0 B in which t is the crystallite thickness, and Λ (extinction length) is defined for the symmetrical Laue case (see 0 0 e.g. ref. 23) as: πV cosθ B Λ = , (14) 0 r λ F (1+ cos2θ ) e hkl B | | | | In general f(A)<1 and converges to 1 if t Λ . In this case we get the highest reflectivity. 0 0 ≪ Thequantityγ σ isknownassecondary extinctioncoefficientandT/γ isthe distancetravelledbythedirect 0 0 beam inside the crystal. 3.2 Curved crystals Similarly to mosaic crystals, curved crystals have an angular dispersion of the lattice planes and thus a much larger energy passband (see Eq. 7) than perfect crystals. The properties of these crystals and the methods to get them are discussed in Ref. 24. Here we summarize their reflection properties. The most recent theory of the radiationdiffraction in transmissiongeometry for such crystals, in the case of a largeand homogeneouscurvature, is now well fixed and has been comparedwith the experimental results (see Refs. 25,26). In this theory, the distortion of diffracting planes is described by the strain gradient β , that, in s the case of a uniform curvature, is given by: Ω β = (15) s T (δ/2) 0 where Ω is the total bending angle and correspondsto the mosaicityofthe mosaic crystals, T is the thickness of 0 the crystal and δ is the Darwin width. When the strain gradient β becomes larger than a critical value β =π/(2Λ ), it has been shown that, for s c 0 | | a uniform curvature of planes, the peak reflectivity Rmax of a curved crystal is given by: 2 Rpeak(cp,E)= Irpeak(cp,E) = 1 e−πcpdΛhk20l e−cpµcoΩsθB. (16) I0 − ! whereIpeak isthereflectedpeakintensity,c =Ω/T isthecurvatureofthelatticeplanesassumedtobeuniform r p 0 across the crystal thickness, and the extinction length Λ = Λ (E) is given by Eq. 14. The reflected intensity 0 0 profile I (c ,E) is that of a perfect crystal with the Darwin width replaced with Ω. This profile is shown in r p Fig. 5. From the last equation, it can be shown that the highest peak reflectivity is obtained for a curvature of the lattice planes given by: M copt = . (17) p ln 1+ M N where M = π2dhkl and N = µΩ . (cid:0) (cid:1) Λ20 cosθB 3.3 Mosaic crystals vs. curved crystals Both mosaic crystals and curved crystals can be used for a Laue lens, if they can be produced with the needed angularspread. However,inprinciple,curvedcrystalscanreachahigherefficiencythanmosaiccrystals. Indeed, while the diffraction efficiency of mosaic crystals is limited to 50%, that of curved crystals can reach 100%. Another advantage of curved crystals is that the diffraction profile of a curved crystal is rectangular with width relatedto Ω, while that of mosaic crystalsis Gaussian with fwhm equal to the mosaicity β . Given the absence m of Gaussian tails, curved crystals concentrate the flux better (see next Section). This better performance of the curved crystals with respect to the mosaic crystals for Laue lenses is discussed in depth in Ref. 27. Curved crystals can be obtained in various ways.24 The most feasible techniques to be used for Laue lenses include the elastic bending of a perfect crystal (the technique commonly adopted in synchrotron radiation facilities), the deposition of a coating on a wafer, the growing of a two-component crystal whose composition variesalongthecrystalgrowthaxis(see,e.g.,Ref.25)ortheindentationofonefaceofawafer. Thelasttechnique is being developed at the University of Ferrara(V. Guidi, private communication) with very satisfactory results (see Fig. 6). Also the deposition of a coating on a wafer is being tested at the same University. Figure 5. Reflectivity profile of a curved crystal as a function of the rocking angle ∆θ. This angle gives the difference betweentheincidenceangleofthemonochromaticphotonbeamandtheBraggangle. I0 representstheincidentintensity, It(∆θ) thetransmitted intensity and Ir(∆θ) thereflected intensity. Reprinted from Ref. 25. Figure 6. Measured rocking curve, in transmission geometry at 150 keV, of a Si(111) crystal curved at the University of Ferrara(seetext). ∆θ givesthedifferencebetweentheincidenceangleofthemonochromaticphotonbeamandtheBragg angle. Open circles: ratio between measured intensity of the diffracted beam and measured intensity of the transmitted beam (also called diffraction efficiency). Filled circles: difference between transmitted and diffracted intensities. Note thattheangle∆θ,throughtheBragglaw,isrelatedtothereflectedphotonenergy. Thusthefigurealsoshowstheenergy bandwidth of the curvedcrystal. Reprinted from Ref. 28, who tested thecrystal sample. Figure 7. Density of a crystal unit cell versus element atomic number. Reprintedfrom Ref. 24. 4. OPTIMIZATION OF A LAUE LENS The free parameters of a Laue lens are the crystal properties (materials, lattice planes for diffraction, micro- crystalsizeandmosaicityinthecaseofmosaiccrystals,totalbendingangleinthecaseofcurvedcrystals,crystal thickness), the lens geometry (ring-likeor Archimedes’spiral), its focallength andits nominalenergy passband. Many optimization studies of these parameters have been performed and tested.24,29–31 4.1 Crystal material Independently of the crystal structure (mosaic or curved), in order to optimize the crystal reflectivity it is important to maximize Q(E) as defined in Eq. 11. This is the integrated crystal reflectivity per unit volume, whosenormalizationis the ratio F /V 2 betweenthe structure factorofthe chosenlattice planes F andthe hkl hkl | | volume V of the unit cell. The inverse ofV is the atomic density N. Thus a small V (or a high N) is important for maximizing Q(E). The value of N as a function the element atomic number Z is shown in Fig. 7. As can be seen, for single–element materials, broad density peaks are apparent in correspondence of the atomic numbers 5, 13, 28, 45, and 78. Common materials like Al (Z = 13), Si (Z = 14), Ni (Z = 28), Cu (Z = 29), Zn (Z = 30), Ge (Z = 32), Mo (Z = 42), Rh (Z = 45), Ag (Z = 47), Ta (Z = 73), W (Z = 74), Au (Z = 42) are good candidates to be used for Laue lenses and should be preferred to other elements if they are availableas crystalswith the requestedproperties. The peak reflectivity versusenergy offew single–element mosaic crystal materials is shown in Fig. 8. Also double–element crystal materials can be used for Laue lenses. Several of them, developed for other applications, are already available, like GaAs, InAs, CdTe, CaF . With some improvements, these crystal 2 materials can be used for Laue lenses (see discussion in Ref. 24). Clearly the best lattice planes are those that optimize the structure factor F , under the condition that hkl | | the corresponding d is consistent with lens constraints, such as energy passband, lens size and focal length hkl (see below). 4.2 Crystallite size and angular distribution in mosaic crystals From the reflectivity equation (Eq. 9), the crystallite thickness t plays an important role in the reflectivity 0 optimization. For fixed values of the mosaicity and crystal thickness, the highest reflectivity is obtained for a crystallitethicknessthatsatisfiestheconditiont Λ . Ingeneral,thisimpliesathicknessoftheorderof1µm. 0 0 ≪ Unfortunatelythis conditionisstillnotalwayssatisfied. FromextendedtestsperformedonCu(111)supplied by ILL,32 it was found that the condition above is satisfied in single points,22 but not when the entire crystal crosssectionisirradiated(valuesevenhigherthan100µmhavebeenfound24). Inaddition,inRef.24itisfound that t is energy dependent, which is a surprising result that requires an interpretation (see discussion therein). 0 0.45 Ag (111) Au (111) 0.4 Cu (111) Ge (111) Mo (110) 0.35 0.3 0 I I/ 0.25 0.2 0.15 0.1 100 200 300 400 500 600 700 800 900 1000 Energy - keV Figure 8. Peak reflectivity of 5 candidate crystal materials. The Miller indices used give the highest reflectivity. A mosaicity of40 arcsec isassumed. Thethicknesshasbeenoptimized. Theproductiontechnology of mosaic crystals with therequired spread is already maturefor Ge and Cu. Figure 9. Normalized effective area for different values of the mosaicity of Cu(111). Left: First diffraction order; Right: second diffraction order. Reprintedfrom Ref. 31. The crystal mosaicity is another crucial parameter for the optimization of the lens performance. It can be seen11,29–31 that, even if a higher mosaicity gives a larger lens effective area (see Fig. 9), a higher spread also produces a larger defocusing of the reflected photons in the focal plane and thus a lower lens sensitivity. This can be seen by introducing the focusing factor G(E) of a Laue lens: A (E) eff G(E)=f (18) ph A d in which A (E) is the effective area of the lens and A is the area of the focal spot which contains a fraction eff d f of photons reflected by the lens. ph Assuming that the detector noise is Poissonian,it canbe easilyshownthat G(E) is inverselyproportionalto the minimum detectable continuum intensity of a lens: n 2B σ I (E)= (19) min η G A ∆T ∆E d r d