- #1
mshr
- 5
- 0
I'm having some trouble finding consistent results for the derivation of the 1D phonon density of state. I'm applying periodic boundary conditions to a 1D monatomic chain.
I can work through and find that D(K)=L/(2π). This is the same result as given by Myers (1990, p. 127). Myers uses only periodic boundary conditions.
Both Kittel (8th ed, pp. 109-110) and Blakemore (1974, pp. 99-100) calculate D(K) using both fixed and periodic boundaries.
Kittel says that the number of modes per unit K-range is L/π in the fixed case, and L/(2π) in the periodic case, and then goes on to use only L/π.
Blakemore says that the number of modes per unit K-range is L/π in the fixed case, and then asserts that it is L/π in the periodic case, even though he shows that K=+/- 2nπ/L.
Di Bartolo & Powell (p. 239) use periodic boundary conditions and say that D(K) = 2 * L/(2π), as there are "two values of corresponding to each angular frequency".
What's going on?Using L/(2π) makes sense for higher dimensions, as you can then say N = (L/(2π))3*(4πK3/3) for 3D, and find D(K) as dN/dK, and from there D(ω) as dN/dK * dK/dω. (as per Kittel (pp. 111-112))
How can I maintain consistency between the approaches used in the 1D and 3D (and 2D) cases?Di Bartolo & Powell (1976) Phonons and Resonances in Solids
I can work through and find that D(K)=L/(2π). This is the same result as given by Myers (1990, p. 127). Myers uses only periodic boundary conditions.
Both Kittel (8th ed, pp. 109-110) and Blakemore (1974, pp. 99-100) calculate D(K) using both fixed and periodic boundaries.
Kittel says that the number of modes per unit K-range is L/π in the fixed case, and L/(2π) in the periodic case, and then goes on to use only L/π.
Blakemore says that the number of modes per unit K-range is L/π in the fixed case, and then asserts that it is L/π in the periodic case, even though he shows that K=+/- 2nπ/L.
Di Bartolo & Powell (p. 239) use periodic boundary conditions and say that D(K) = 2 * L/(2π), as there are "two values of corresponding to each angular frequency".
What's going on?Using L/(2π) makes sense for higher dimensions, as you can then say N = (L/(2π))3*(4πK3/3) for 3D, and find D(K) as dN/dK, and from there D(ω) as dN/dK * dK/dω. (as per Kittel (pp. 111-112))
How can I maintain consistency between the approaches used in the 1D and 3D (and 2D) cases?Di Bartolo & Powell (1976) Phonons and Resonances in Solids