To answer the question, here quotes from the aforementioned paper. Earliest 1892:

"The Delaware & Hudson established a test section in 1892 consisting of hemlock and yellow pine ties, both untreated and treated two different ways. They were testing different track fastenings— the Davies spike, Servis tie plates, and rail joints—as much as they were the treatments."

"Along with all the attention paid to preventing decay came increased efforts to reduce the mechanical wear of ties. It is pointless to creosote a tie which will be badly rail cut after five years, or if the spikes will wear loose in a short time. More and more railroads were using tie plates, and beginning about 1911 some railroads installed boring and adzing machines in their treating plants. With these machines spike holes were pre-bored, and the seats for the ties plates were adzed to a level, flat surface."

"[Trat II] E. E. Russell Tratman, Report on the Use of Metal Ties and on Preservative Pro- cesses and Metal Tie Plates for Wooden Ties. Bulletin No. 9, U.S. Department of Agricul- ture, Division of Forestry. Washington: Government Printing Office, 1894."