You can also try this:
Concept Set: The ingredients of interest
Cohort Entry Criteria: First exposure of 'ingredients of interest'
Cohort Exit Criteria: Drug Era using 'ingredients of interest' with specified gap days
Censoring Events: Drug Era using 'ingredients of interest' having exactly 2 distinct drug eras of 'ingredients of interest' starting all days before and 0 days after event index.
The bolded part is the key: it says that you kick the person out of the cohort as soon as you see 2 distinct drug eras (2 different drug era concept IDs). The nice thing about this is that instead of finding only people that have exactly one ingredient, you let people into the cohort who might have multiple future-ingredients, but take them out as soon as they have a second.
Some limitations: 1) you have to work at the ingredient-level from the drug-era table because all the drug_concept_ids there are at the ingredient level. There's no notion of 'augmentation' in this definition: if a new drug appears, from 'ingredients of interest' we assume that's not mono anymore, so you have to carefully consider the 'ingredients of interest' wouldn't be those that might be used in combination with each other.
For your reference: