Free spherical waves

Let us first take a closer look at free spherical waves.  The Hamiltonian of a free particle is H0 = P2/(2m).  Since [Pi,Pj] = 0, (i,j = x,y,z), we have [H0,P] = 0 and we can find common eigenfunctions of H0, Px, Py, and Pz.  The observables Px, Py, and Pz form a C.S.C.O. for a spinless particle.  The common eigenstates form a basis {|p>} with normalization and closure relations  <p|p'> = δ(p - p'), ∫d3p|p><p| = I.
P|p> = p|p>,   H0|p> = [p2/(2m)]|p>.
The wave functions associated with the kets |p> are the plane waves
<r|p> = (2πħ)-3/2exp(i(p/ħ)∙r).
Introducing the wave vector k = p/ħ we define the ket |k> = ħ3/2|p>.
Since |k> is proportional to |p> we have P|k> = k|k>,  H0|k> = [ħ2k2/(2m)]|k>.

The eigenstates of H0 are infinitely degenerate.  The normalization and closure relations for {|k>} are written <k|k'> = δ(k - k'), ∫d3k|k><k| = I.  The Hamiltonian of a free particle also commutes with L2 and Lz.  (We can view a free particle as a particle in a central potential of zero strength.)  The observables H0, L2 and Lz form a C.S.C.O. for a spinless particle.  The common eigenstates form a basis {|k,l,m>}.
H|k,l,m> = [ħ2k2/(2m)]|k,l,m>,  L2|k,l,m> = ħ2l(l + 1)|k,l,m>,  Lz|k,l,m> = mħ|k,l,m>.

The wave functions associated with the kets |k,l,m> are the free spherical waves.
<r|k,l,m> = fkl(r)Ylm(θ,φ).
For a free particle the radial functions fkl(r) are solutions to the differential equation
[-(ħ2/(2m))(1/r)(∂2/∂r2)r + ħ2l(l + 1)/(2mr2)]fkl(r) = Eklfkl(r).

With ρ = kr, and Ekl = Ek = ħ2k2/(2m) this equation becomes
[(1/ρ)(∂2/∂ρ2)ρ + 1 - l(l + 1)/ρ2]fkl(ρ) = 0.
This is the spherical Bessel equation.  Its solutions are the spherical Bessel functions jl(ρ) and nl(ρ) and linear combinations of both.

jl(ρ) = (-1)lρl[(1/ρ)(∂/∂ρ]l )sin(ρ)/ρ,  or  2lρlΣs=0(-1)s(s + l)! ρ2s/[s!(2s + 2l + 1)!].
nl(ρ) = -(-1)lρl[(1/ρ)(∂/∂ρ)]l cos(ρ)/ρ, or  2-l (-1/ρ)l+1Σs=0(-1)s(s - l)! ρ2s/[s!(2s - 2l)!].

As ρ --> 0, jl(ρ) ∝ ρl,  nl(ρ) ∝ ρ-(l+1).
Therefore only jl(ρ) is an acceptable wave function for a free particle.  Using the orthonormality of the spherical Bessel functions we have
fkl(r)Ylm(θ,φ) =  (2k2/π)½jl(kr)Ylm(θ,φ) = <r|k,l,m>.
The free spherical waves are orthogonal and complete. Let us take a closer look at jl(ρ).

jl(ρ) = 2lρlΣs=0(-1)s(s + l)! ρ2s/[s!(2s + 2l + 1)!].
As r --> 0, the lowest power in ρ dominates.
jl(ρ --> 0) = 2lρl (l)! /(2l + 1)! =  ρl/(2l + 1)!!.

Double factorial notation:
(2l)!! = 2l*(2l - 2)*(2l - 4)* ... *6*4*2 = 2l*(l)!,
(2l + 1)!! = (2l + 1)*(2l - 1)*(2l - 3)* ... *5*3*1 = (2l + 1)!/(2l*(l)!).

As r --> ∞ the highest power in ρ dominates.
jl(ρ) = (-1)lρl[(1/ρ)(∂/∂ρ)]l sin(ρ)/ρ = (-1)lρl[(1/ρ)(∂/∂ρ)]l-1 (cos(ρ)/ρ2 - sin(ρ)/ρ2).
As we keep on differentiating, the highest power will always come from the derivative of the first term.  We therefore have
jl(ρ-->∞) = (-1)lρl(1/ρ)l+1(∂/∂ρ)l sin(ρ).
(∂/∂ρ) sin(ρ) = cos(ρ) = -sin(ρ - π/2).
(∂/∂ρ)l sin(ρ) = (-1)l sin(ρ - lπ/2).
jl(ρ-->∞) = (1/ρ) sin(ρ - lπ/2).

The probability of finding a particle in a free spherical wave eigenstate of the Hamiltonian H0 near the origin, within a solid angle dΩ0 and between r and r + dr is proportional to
 k2r2|jl(kr)|2|Ylm(θ,φ)|2drdΩ0.
jl(kr) = (kr)l/(2l + 1)!! as kr --> 0.
 k2r2|jl(kr)|2 = (kr)2l+2/[(2l + 1)!!]2 as kr --> 0.
The probability of finding the particle near the origin decreases as l increases.  If r < (l(l + 1))½/k, then the probability of finding the particle between zero and r is negligible.

Note:  To show this, you cannot just require that  k2r2|jl(kr)|2 = 0 in the region  r < (l(l + 1))½/k, because the wave function is a continuum wave function and cannot be normalized to one.  You have to build a wave packet and show that the wave packet does not penetrate the region r < (l(l + 1))½/k.  Because the wave packet does not penetrate, a particle in an eigenstate |k,l,m> of H0 is practically unaffected by what happens inside a sphere centered at the origin of radius  bl(r) < (l(l + 1))½/k.

Let us use classical physics as a guide.  In classical mechanics, a free particle with angular momentum L about the origin and momentum p moves in a straight line with closest distance to the origin b = L/p;  b is called the impact parameter relative to the origin. 
If L = ħ(l(l + 1))½ and p = ħk, then bl(k) = (1/k)(l(l + 1))½.
bl(k) can thus be interpreted semi classically.  bl(k) is the radius where the angular momentum barrier potential, Ueff(r) = l(l + 1)ħ2/(2mr2) is equal to the total energy of the free particle.
l(l + 1)ħ2/(2m(bl(k))2) = ħ2k2/(2m),  bl(k) = (1/k)(l(l + 1))½.
In Quantum Mechanics, for r < bl(k), in the classically forbidden region, the eigenfunctions |k,l,m> can only have an approximately exponentially decaying tail.