Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 50, 3, pp. 797-805, Warsaw 2012 50th Anniversary of JTAM WAVE-QUASI-PARTICLE DUALISM IN THE TRANSMISSION-REFLECTION PROBLEM FOR ELASTIC WAVES Gérard A. Maugin, Martine Rousseau Université Pierre et Marie Curie – Paris 6, Institut Jean Le Rond d’Alembert, Paris, France e-mail: gerard.maugin@upmc.fr Following along the line of recent works in which the notion of quasi-particles associated with surface acoustic waves of different types was introduced via canonical conservation laws, here the emphasis is placed on the possible exploitation of this dualism in the classical problem of the transmission and reflection of waves by a discontinuity between two media (in perfect contact orwith possible delamination) and themore general case of amono-layer or multi-layer sandwiched slab. Key words: elastic waves, quasi-particle, transmission-reflection 1. Introduction In recent works (Maugin and Rousseau, 2010a,b; Rousseau andMaugin, 2011), a kind of wave- -particle dualism has been constructed for elastic waves in acoustic physics. The idea is close to that of phonons, but we only need to know the analytical wavelike solution to the continu- um problem in order to build this dualism. For practical reasons related to a potential use in Non Destructive Evaluation techniques, the emphasis was placed on SAWs (Surface Acoustic Waves) propagating on the top of an elastic half-space with various boundary conditions on the limiting plane and various physico-mechanical couplings (e.g., in electroelasticity). What must be retained from these studies is (i) the exploitation of the notion of pseudo-momentum (in the sense of Peierls (1991, pp. 30-42) orwave momentum in themanner of Brenig (1955) and in the spirit of the dynamical theory of configurational forces (Maugin, 2011) – and the accompanying application of Noether’s theorem for conservation laws, (ii) the fact that the quasi-particle thus associatedwith thewave-likemotionhas amomentumobtainedby integration of thisBrenigmo- mentum (or its canonical generalization in amulti-field theory) over a volume representative of thewavemotion (for a SAW,onewavelength in thepropagation direction, onepenetrationdepth in depth in the substrate, and one unit laterally), and (iii) only wave modes were considered, i.e., independently of any initial-condition or interaction problem. Themotion of quasi-particles thus obtained is “Newtonian” and inertial in the absence of dissipative effects. The associated energy is purely kinetic in form, while in the original continuum problem it could be of mixed origin, including, e.g., kinetic, elastic, electromagnetic, and electro-magneto-elastic interactions. Still the “mass” associated with the said quasi-particle accounts for all these origins as well as for the type of boundary condition and depth behaviour. In the present work, relying on a simple case of the wave – one elastic component of the SH (shear-horizontal) type – and considering the normal incidence of such a wave on the interface between two elastic continua characterised by their own elastic (shear) coefficient and mass density, we revisit the problem of the transmission and reflection of such awave at this interface in terms of the wave properties (this is only a reminder of well known results) or those of the associated quasi-particle properties (this is the new contribution). The situation considered is that which prevails in the potential exploitation in nanoscale systems (for macro-systems this 798 G.A.Maugin, M. Rousseau would be the realm of potentially dangerous seismic waves). In the present case, discarding the variation of amplitude with depth we are satisfied with considering a plane face wave. The representative volume element of the wave modes then is one wavelength in the propagation direction and a unit square in the transverse plane. 2. Reminder on the wavelike picture For the sake of simplicity, we consider a one-dimensional wave problemalong the x-direction for an elastic displacement u that may be a transverse shear-horizontal (SH)mode. The governing wave equation is given by ∂p ∂t − ∂σ ∂x =0 p= ρ ∂u ∂t σ=µ ∂u ∂x (2.1) where ρ and µ are a prescribedmatter density and fixed elasticity (e.g., shear) coefficient µ in a given region of space. Space and time partial derivatives are alternately denoted by a comma followed by x or a t. The first of (2.1) follows in a variational format from the Lagrangian density per unit volume L= 1 2 ρu2,t− 1 2 µu2,x (2.2) We consider a harmonic wave motion u=U cos(kx−ωt) (2.3) so that from (2.1) we have the trivial dispersion relation D(ωk)=ω2−c2k2 =0 c= √ µ ρ (2.4) 2.1. Transmission-reflection problem for a perfect interface Thestandard transmission-reflectionproblemhere consists in consideringanormally incident wave inmedium1(properties ρ1 and µ1) on the interface at x=x0 =0that separatesmedium1 frommedium 2 (properties ρ2 and µ2). Thewave solution inmedium 1 is sought as the sum of an incident component and a reflected component, while the solution inmedium 2 consists in a transmitted component.With an obvious notation, we have thus u1 =uI +uR u2 =uT (2.5) with u1 =U cos(k1x−ωt)+R0U cos(k1x+ωt) u2 =T0U cos(k2x−ωt) (2.6) where R0 and T0 are the reflection and transmission coefficients for this perfect interface case for which we assume the continuity of displacement and stress, i.e. u1 =u2 µ1u1,x =µ2u2,x at x=0 (2.7) The solution of this system is, for any amplitude U, the value of the reflection and transmission coefficients as R0 = z1−z2 z1+z2 T0 = 2z1 z1+z2 (2.8) where zα = ραcα,α=1,2 are impedances.With these we check the conservation of energy flux in the well known form F0 =1−R 2 0− z2 z1 T20 ≡ 0 (2.9) Wave-quasi-particle dualism in the transmission-reflection problem... 799 2.2. Transmission-reflection problem for an interface with delamination In view of applications to non-destructive (NDT) evaluation methods, it is of interest to consider the case of an imperfect interface that admits the possibility of delamination and for which matching conditions (2.7) are replaced by the conditions (known as Jones’ conditions (Jones andWhittier, 1967)) σ1 =σ2 ≡K[[u]] (2.10) where K is a positive coefficient characterising the degree of delamination and the symbol [[·]] means the jumpof its enclosure, i.e., [[u]] =u2−u1 at x=0.Wemust look for complex solutions of the type u=Aexp[i(kx−ωt)]. Conditions (2.10) yield the system z1(1−R)= z2T z2T =−i K ω (1+R−T) (2.11) where R and T are the complex reflection and transmission coefficients corresponding to this imperfect case. The solution to (2.11) reads R= z1z2− i(K/ω)(z1 −z2) z1z2− i(K/ω)(z1 +z2) T = −2i(K/ω)z1 z1z2− i(K/ω)(z1 +z2) (2.12) By computing the squares of themoduli of the complex quantities R and T , we check that (2.9) is replaced by FK =1−|R| 2− z2 z1 |T |2 ≡ 0 (2.13) We note that FK =F0 ( 1− z21z 2 2 z21z 2 2 +(K/ω) 2(z1+z2)2 ) (2.14) The solution of this imperfect interface case is characterised by the parameter K/ωwhich shows the role played by the frequency ω. The limit case K →∞ corresponds to the perfect interface for which (2.9) holds true. The limit case K → 0 corresponds to full delamination (no more transmission and complete reflection: T =0,R=1). Remark 2.1. It is possible to obtain an estimate of the K coefficient bymeasuring the ampli- tude of reflected or transmittedwaves. In particular, whenevermedia 1 and 2 are identical, but K still is not zero, this measure provides a means of determining the presence of an internal delamination in the body. 3. Associated quasi-particle picture With field equation (2.1), there are associated (via the application of Noether’s theorem or by direct manipulation) conservation laws of energy and wave momentum in the local form in a homogeneous medium ∂H ∂t − ∂Q ∂x =0 ∂P ∂t − ∂b ∂x =0 (3.1) where the energy orHamiltonian per unit volume H, the energy flux Q, thewavemomentum P and (here reduced to a scalar) the Eshelby stress b are defined by (see Maugin and Rousseau (2010a) for the canonical definitions in three dimensions) H =E+W = 1 2 ρu2,t+ 1 2 µu2,x Q=σut =µu,xu,t P =−ρu,tu,x b=−(L+σux) (3.2) 800 G.A.Maugin, M. Rousseau The first of these last two follows Brenig’s definition. It is a remarkable – but sometimes misleading- fact that in this one-dimensional case there hold the following identities Q=−c2P b=−H (3.3) Now the concept of quasi-particle is introduced by integrating conservation equations (3.1) over a volume that is representative of the present wave motion, i.e., over one wavelength in the x propagation direction and a square section of sides equal to unity in the transverse direction. This introduces averages notedwith the symbolism 〈·〉. For solutions of type (2.3) this procedure yields the following results 〈L〉=0 d dt 〈P〉=0 d dt 〈H〉=0 (3.4) where, in a homogeneous medium 〈P〉= ρωπU2 ≡Mc 〈H〉= 1 2 Mc2 (3.5) where the first of these defined the “mass” M = ρkπU2. Simultaneously, equations (3.4)2,3 show that the said quasi-particle has an inertialNewtonianmotion. It also satisfies theLebnizian conservationof kinetic energy (or vis-viva). Indeed,while therearebothkinetic andelastic energy at the continuum level, the energy of the associated quasi-particle is purely kinetic on account of the given definition of “mass”. Note that homogeneous equations (3.4)2,3 hold good because of the periodicity at the two ends of the integration interval that makes the contributions from Q and b to vanish. Equation (3.4)1 holds true because it is shown in computing the average for solutions (2.3) that the result is proportional to “dispersion” relation (2.4) and therefore vanishes identically, a result known in the kinematic-wave mechanics of Lighthill andWhitham (see, e.g., Whitham, 1974). Note finally that all quantities introduced in (3.4) and (3.5) are proportional to the square of the amplitude of the wave, hence are all of energetic nature. 3.1. Transmission-reflection problem in the quasi-particle picture Considering first the case of the perfect interface at x = 0, we can associate one quasi- particle with each wave component of the problem. With an obvious notation, we have the following “masses” MI = ρ1k1πU 2 MR = ρ1k1πR 2 0U 2 MT = ρ2k2πT 2 0U 2 (3.6) The corresponding averaged wave momenta are given by PI ≡〈PI〉= ρ1ωπU 2 (3.7) and PR ≡〈PR〉=−ρ1ωπR 2 0U 2 PT ≡〈PT〉= ρ2ωπT 2 0U 2 (3.8) where we account for the fact that the averaged wave momentum PR is oriented towards nega- tive x’s. We note ∆M and ∆P the possible misfits in mass andmomentum defined by ∆M := (MR+MT)−MI (3.9) and ∆P =(|PR|+ |PT |)−|PI| (3.10) Wave-quasi-particle dualism in the transmission-reflection problem... 801 where the symbolism | · | refers to the absolute value of its enclosure. That is, we are comparing the strengths of the momenta and, therefore, we are not performing a vectorial balance. Similarly, for the kinetic energy of the associated quasi-particles: ∆H =(HR+HT)−HI (3.11) We say that a quantity is conserved during the transmission-reflection problem if the correspon- dingmisfit vanishes. By applying definition (3.5)2, one immediately shows that ∆H = (1 2 z1ωπU 2 ) F0 (3.12) where F0 has been defined in (2.9). But the latter vanishes. Accordingly, ∆H ≡ 0: kinetic energy is conserved in the transmission-reflection problem seen as a quasi-particle process that may be qualified of Leibnizian (conservation of vis-viva). Historically, it was in fact the collision problem between particles that led the late 17th and early 18th century scientists (prominently Leibniz in 1686; see Smith, 2006) to introduce the vis-viva – twice the actual kinetic energy – as the “physical quantity” to be conserved in such interactions and not Descartes’ “quantity of motion” as suggested by a different school (see below for the particle moment in the present problem). Indeed, “mass” is not generally conserved in the present problem as it is immediately shown that ∆M = ρ2k2πT 2 0U 2 (c21− c 2 2 c21 ) (3.13) Similarly, on computing ∆P it is obtained that ∆P = ρ2k2πT 2 0U 2 (c1c2− c 2 2 c1 ) (3.14) This shows that ∆P = ( c1c2 c1+ c2 ) ∆M (3.15) so that ∆P and ∆M always are in the same sign. Inparticular, ∆M > 0 if c1 >c2 and ∆M < 0 if c1