R create new variable from several columns if one satisfies condition

sorry for the very unspecific title. I really don't know how to name the problem... So I'm dealing with this data frame :

 C1 <- c("made in Italy", "100% silk", "dry clean only")
 C2 <- c("80% cotton, 20% polyester","made in France", "hand wash")
 C3 <- c("made in Italy", "Designer color : vanilla", " 100% nylon")
 eg <- as.data.frame(rbind(C1,C2,C3))

I'd like to add one veriable "composition" by extracting all the values containing "%" sign. As you can see, the composition values are not in the same column for each observation... I tried several methods but failed to achieve this. for example :

 fcompo <- function(x){
   if (grepl('%',x) = TRUE){eg$composition <- paste(x)}
 else {eg$composition=NA}
 }

then I got lost in the function... being very new to R language... I also tried :

  library(stringr)

  eg$composition <- str_extract(eg[,c(1:3)], "%$" )

thanks a lot for your generous help