Roughly speaking, complex numbers seem to have become widely used in physics and engineering around 1940, after gradually gaining popularity starting around 1920.
In the late 19th and early 20th centuries, there were three algebraic systems of numbers competing for attention: complex numbers, quaternions, and vectors. Vectors were late to the party. Maxwell used quaternions (1873), and they were enthusiastically promoted by their adherents, especially Hamilton, for purposes that today we would use vectors for. Quaternions can be viewed as a generalization of both vectors and complex numbers. Most physicists considered quaternions to be very abstruse. As late as 1905, Einstein's paper on special relativity states Maxwell's equations without the use of either vector or quaternion notation. The 1895 edition of Kelvin and Tait's massive Treatise on Natural Philosophy has no index entries for "complex," "imaginary," "vector," or "quaternion," and complex numbers are used only infrequently. Tait wrote an 1890 book on quaternions, so it's not as though he didn't know about them. They just weren't the kind of thing that readers could be expected to know about.
Gibbs and Heaviside created the vector system as an easier alternative to quaternions, and it gradually gained popularity. Gibbs described it in an 1881 book, which he promoted vigorously and successfully.
Kennelly introduced the notion of a complex impedance in 1893, and this was recognized as a big selling point of the use of complex numbers. A popular 1941 textbook by Stratton, Electromagnetic Theory, used complex numbers. So I think the widespread use of complex numbers in physics and engineering must date to approximately 1900-1940. To pin this down a little more, I think it's helpful to look at the google ngrams graphs for "complex impedance" and "complex exponential," which both show nonnegligible usage starting around 1920, and an inflection point around 1940.