[math-fun] Question about base 3/2

Is it possible to write 1/3 as an infinite sum of the form a_1 (2/3)^1 + a_2 (2/3)^2 + ... where the sequence of nonnegative integer coefficients a_1, a_2, ... is eventually periodic? Jim Propp