We will prove that there are only two families of solutions:
(1,1,2x−1),(1,2x−1,2x+1),
where x is a positive integer.
*Step I:* a,b,c are odd and pairwise coprime in pairs.
It is easy to see that
a≡b≡c(mod2).
If a,b,c are all even, denote by d,e,f the exponents of 2 in their prime factorization, respectively, with 1≤d≤e≤f. Then the exponent of 2 in bc+a is d. Since bc+a is a power of 2, we get
bc+a=2d,
which is impossible because 2d≤a. Thus, a,b,c must be odd numbers.
If a,b have a common odd prime factor p, then p divides ac+b, so ac+b is divisible by p and thus cannot be a power of 2. Similarly, one shows (a,c)=1 and (b,c)=1, which completes Step I.
Hence, we can assume a≤b≤c, where a,b,c are odd positive integers pairwise coprime such that
ab+c,ac+b,bc+a
are powers of 2. Since a≤b≤c, we have
ab+c≤ac+b≤bc+a.
*Step II:* a=1. Suppose, by contradiction, that 1<a. Then 1<a<b<c, and
ab+c=2k,ac+b=2m,bc+a=2n,
with 3≤k<m<n (since 2k≥3+5=8).
From the above equalities, we get the congruences:
ab≡−c(mod2k),ac≡−b(mod2k),bc≡−a(mod2k).
Multiplying these congruences and noting that a,b,c are odd, we obtain
abc≡−1(mod2k)
and
a2≡b2≡c2≡1(mod2k)
(by multiplying the first congruence by c, the second by b, and the third by a).
Since for any odd natural number x,
gcd(x−1,x+1)=2,
we deduce that
a≡±1(mod2k−1),b≡±1(mod2k−1),c≡±1(mod2k−1).
From these congruences, it follows that
a,b,c≥2k−1−1.
This inequality implies
2k=ab+c≥(2k−1−1)2+2k−1−1=22k−2−2k−1,
which means 2≥2k−1−1≥3, a contradiction since k>3. Therefore, the assumption is false and a=1.
Step III: the solutions. If b=1, then clearly c=2x−1. Otherwise, assume a=1 and 1<b<c. From Step II, we know that c>b≥2k−1−1. Since b+c=2k and b,c are odd, the only possibility is b=2k−1−1, c=2k−1+1.
Hence, the only solutions are: (1,1,2x−1) and (1,2x−1,2x+1), x∈N∗.