Supplemental Material for “Plasmon Hybridization Model Generalized to Conductively Bridged Nanoparticle Dimers” Lifei Liu1,2, Yumin Wang1,3, Zheyu Fang1,3,4, and Ke Zhao1,3* 1. Laboratory for Nanophotonics, Rice University, Houston, Texas 77005, United States 2. Department of Physics and Astronomy, M.S. 61, Rice University, Houston, Texas 770051892, United States 3. Department of Electrical and Computer Engineering, M.S. 366, Rice University, Houston, Texas 77251-1892, United States 4. School of Physics, State Key Lab for Mesoscopic Physics, Peking University, Beijing 100871, China *Corresponding author: Ke Zhao Email: kz4@rice.edu In the spherical coordinate systems of nanoparticle i (i =1, 2) with the origin at its center Oi, we use ri, θi, and φi to refer to the radial distance, polar angle, and azimuthal angle of a point in space. In addition, when θi and θj (i ≠ j) refer to the same point P on the surface of nanoparticle j, θi is a function of θj and written as θi = Θi(θj), and Xi(θj) is used to denote the distance between Oi and point P. I. Self-interacting Energy of Surface Charge For small dimers with a negligible retardation effect, the interaction energy between surface charges Vint obeys the instantaneous Coulomb law: i (r1 ) i (r2 ) 1 1 2 2 Vint a d1 a d 2 | r r | 2 i 1, 2 4 0 1 2 , 1 1 (r1 ) 2 (r2 ) 2 2 a d1 a d 2 | r r | 4 0 1 2 (S2) where ε0 is the vacuum dielectric constant. The explicit form of Vint is: (n0 e) 2 a 5 l2 2l 2 Vint { [ S l( i ) ( yl( i ) Vl( i ) ) T S l( i ) ] i 1, 2 l 2 0 2l 1 2l 1 1 4ll ' 2 ( V ) T Vll ' S l(1) S l('2 ) } ll ' 2l '1 2 , (S3a) 1 where yl( i ) 2 0 K iYl 0 cos i sin i d i a l 1 cos jYl 0 (i ) Kj sin j d j (i j ) , X il 1 ( j ) Vl 2 0 (i ) 2 0 0 2 0 '0 0 2 V 2 0 (S3b) 2 cos cos ' sin sin ' dd ' d 2 2 sin sin ' cos 2 cos cos ' (S3c) (S3d) 0 d d d 0 1 0 1 2 cos1 cos 2 sin 1 sin 2 2 2 sin 1 sin 2 cos 2 cos1 cos 2 2(2 D / a )(cos1 cos 2 ) (2 D / a ) 2 (S3e) a l 1 (S3f) Vll ' 0 Yl 0 (1 ) l 1 Yl '0 ( 2 ) sin1d1 X 2 (1 ) b K1 , K 2 1 when 0 1 0 ( 0 arcsin ) and 0 2 , and K1 , K 2 0 a anywhere else. II. Equations for Plasmon Modes Derivation Defining Lmax as the maximum order of multipoles included in the calculation, the equations of motion for plasmon modes can be derived from lagrangian d 2S I 2 V S 0, dt where I is the (2Lmax 1) (2Lmax 1) unit matrix, V matrix with the following elements ( l , l ' Lmax ): Vl ,l ' Vl L Vl ,l ' L max max , l ' Lmax ll ' 2l ' Vll ' p2 2l '1 l p2 2l 1 L T Vint : (S6a) is a (2Lmax 1) (2Lmax 1) (S6b) (S6c) 2 Vl L max , l ' Vl , 2 L Vl L V2 L V2 L V2 L and max max max max max 1 2l ' Vl 'l p2 2l '1 yl(1) Vl(1) 2 p 2l 1 , 2 Lmax 1 1, l yl( 2 ) Vl( 2 ) 2 p 2l 1 a 3 ( yl(1) Vl(1) )l 2 2 b D (2l 1) p 1, l Lmax 1, l Lmax 1 (S6f) (S6g) (S6h) a 3 ( V ) 2 p 2 2b 2 D (S6i) max max (S6e) a 3 ( yl( 2 ) Vl( 2 ) )l 2 b 2 D (2l 1) p S is a 2Lmax 1 column array with elements: S l S l(1) Sl L Sl( 2 ) S2 L (S6d) T (S6j) (S6k) (S6l) The corresponding eigenvalue problem is: det[V I 2 ] 0 , (S6m) where the square root of eigenvalues are the plasmon frequencies and the eigenvectors composed of Sl(1) , S l( 2 ) and T give the charge distribution of plasmon modes through Eq. (8) in the main paper. III. Equations for Optical Response L T (Vint Vext ) and in addition, the equation of motion includes the dissipation term derived from F (see Eqs. (13), (14) in the main paper), resulting in To calculate the optical response, the linear response equations: 3 where d 2 S dS I 2 V S , dt dt S , I , and V are the same as those in section II, (S7a) is a (2Lmax 1) (2Lmax 1) matrix with nonzero elements only on the diagonal: l ,l l L max , l Lmax 2L and max 1, 2 Lmax 1 0 gT 0 g0 p2 (S7b) p2 (S7c) is a 2Lmax 1 column array with elements: l l L 2L max 1 max l1 e 4 E0 e it me a 3 e E0 e it me a Solving Eq. (S7a) for Sl(1) , S l( 2 ) (S8a) (S8b) and , the polarizability can be obtained through Eq. (16) in T the main paper. In both the plasmon modes and the optical response calculations, the most time-consuming part is to calculate the different components of dimensional integrals. V , because the calculation involves multi- 4