first and second have identical columns transaction_id, date (datetime) and amount. The export overlapped, so a transaction can appear in both.

Stack them, drop rows repeating a transaction_id already seen (keeping the first occurrence), sort by date then transaction_id, and renumber the index from 0.

Input

first =
   transaction_id       date  amount
0               1 2022-01-31    10.0
1               2 2022-01-15    20.0
second =
   transaction_id       date  amount
0               2 2022-01-15    20.0
1               3 2022-02-01    30.0

Output

   transaction_id       date  amount
0               2 2022-01-15    20.0
1               1 2022-01-31    10.0
2               3 2022-02-01    30.0