The statement follows from the following fact.
Lemma. For arbitrary positive integers x and y, the number 4xy−1 divides (4x2−1)2 if and only if x=y.
Proof. If x=y then 4xy−1=4x2−1 obviously divides (4x2−1)2 so it is sufficient to consider the opposite direction.
Call a pair (x,y) of positive integers bad if 4xy−1 divides (4x2−1)2 but x=y. In order to prove that bad pairs do not exist, we present two properties of them which provide an infinite descent.
Property (i). If (x,y) is a bad pair and x<y then there exists a positive integer z<x such that (x,z) is also bad.
Let r=4xy−1(4x2−1)2. Then
r=−r⋅(−1)≡−r(4xy−1)=−(4x2−1)2≡−1(mod4x)
and r=4xz−1 with some positive integer z. From x<y we obtain that
4xz−1=4xy−1(4x2−1)2<4x2−1
and therefore z<x. By the construction, the number 4xz−1 is a divisor of (4x2−1)2 so (x,z) is a bad pair.
Property (ii). If (x,y) is a bad pair then (y,x) is also bad.
Since 1=12≡(4xy)2(mod4xy−1), we have
(4y2−1)2≡(4y2−(4xy)2)2=16y4(4x2−1)2≡0(mod4xy−1)
Hence, the number 4xy−1 divides (4y2−1)2 as well.
Now suppose that there exists at least one bad pair. Take a bad pair (x,y) such that 2x+y attains its smallest possible value. If x<y then property (i) provides a bad pair (x,z) with z<y and thus 2x+z<2x+y. Otherwise, if y<x, property (ii) yields that pair (y,x) is also bad while 2y+x<2x+y. Both cases contradict the assumption that 2x+y is minimal; the Lemma is proved.
To prove the problem statement, apply the Lemma for x=k and y=2n; the number 8kn−1 divides (4k2−1)2 if and only if k=2n. Hence, there is no such n if k is odd and n=k/2 is the only solution if k is even.