Page 469 - Numerical Methods for Chemical Engineering
P. 469

458     9 Fourier analysis




                   Now, if q · (l 1 a + l 2 b + l 3 c) = m2π, m =±1, ±2,... , then
                                      n 0  	             n 0  	       n 0 N cells
                                               −im2π
                          C ρ,ρ (q ) =        e     =            (1) =               (9.124)
                                    (2π) d/2           (2π) d/2       (2π) d/2
                                          l 1 ,l 2 ,l 3      l 1 ,l 2 ,l 3


                   Thus,  C ρ,ρ (q)  has sharp peaks at all q such that q · (l 1 a + l 2 b + l 3 c) is a multiple of
                   2π. If we define the reciprocal lattice vectors α, β, γ to satisfy
                                      α · a = 2π   α · b = 0  α · c = 0
                                      β · a = 0    β · b = 2π  β · c = 0             (9.125)
                                      γ · a = 0    γ · b = 0  γ · c = 2π

                   these peaks occur in q-space at

                                             q = n 1 α + n 2 β + n 3 γ               (9.126)

                   since

                             q · (l 1 a + l 2 b + l 3 c) = (n 1 α + n 2 β + n 3 γ) · (l 1 a + l 2 b + l 3 c)

                                               = (n 1 l 1 + n 2 l 2 + n 3 l 3 )(2π)  (9.127)

                   Above, we have assumed that the periodicity ρ(r + l 1 a + l 2 b + l 3 c) = ρ(r) extends
                   throughout the entire sample; however, often (unless we go to great lengths to anneal
                   the sample or crystallize it slowly from a single nucleation site) the sample contains many
                   different crystal domains, with the crystal lattices in each domain isotropically oriented at
                   random. In such a case, we measure a powder spectrum, and obtain not the full structure
                   factor S(q) but rather merely an isotropically-averaged structure factor S(q). When S(q)is
                   nonzero (it usually appears as a circular “halo” at some scattering angle θ), this signifies
                   that the material has structural periodicity on a length scale

                                                        2π
                                                  σ len ≈                            (9.128)
                                                         q
                   Smaller q-values denote structures on larger length scales. Hence, when X-ray scattering
                   (for a Cu-Kα 1 source, λ is 1.54 Å) is used to measure the ˚angstrom-scale periodicity of
                   crystals, we measure S(q) at large values of q, and thus also at large scattering angles θ
                   (Figure 9.10). By contrast, when we attempt to probe longer nanometer-scale O(10 −9  m)
                   structure, we must measure the scattering at small angles. Thus, when X-ray scattering is
                   performed to probe the atomic-scale structure of crystals, it is known as WAXS (wide angle
                   X-ray scattering). When it is used to probe the structure on the order of tens of nanometers
                   of materials such as self-assembled block copolymers and micellar emulsions, it is known
                   as SAXS (small angle X-ray scattering). At very small θ, the scattered intensity is buried
                   within the intensity of the incident beam, resulting in an effective upper limit of resolution
                   for SAXS of O(100 nm). With special facilities, this technique can be extended to longer
                   length scales, as at the USAXS (ultra small angle X-ray scattering) facility at Argonne
                   National Laboratories.
   464   465   466   467   468   469   470   471   472   473   474