I would like to deterministically derive two different 256-bit keys from a single passphrase -- one used as an ECDSA private key, the other as a symmetric key for AES-GCM.
Would a KDF output sufficient bits of entropy, given a 256-bit input? I'm planning to use Argon2, but I'd be happy to use another KDF if necessary.
The passphrase will be generated using a Diceware wordlist (.txt) (7776 possible words), so if I understand it correctly, a 256-bit passphrase will be $ \log _{7776}{(2^{256})} \approx 20$ words long.