Passphrases add a second recovery layer by turning one seed phrase into many separately derived wallets, but they also create a new secret that the wallet cannot recover for you.
In a BIP39 wallet, the passphrase is not a label or PIN. It is an input to seed derivation:
That means every passphrase produces a valid wallet, including a mistyped one. A blank passphrase opens the ordinary wallet; “vault” and “Vault” open different wallets. Someone who finds the seed can open the blank wallet but still needs the additional secret to reach funds held elsewhere.
The recovery layer therefore works only when the two secrets are stored separately. The seed might sit on paper in a safe, while the passphrase is memorised or kept in another secure location. Lose either one and the network has no reset button. The wallet software cannot prove ownership without the complete combination.
Creating a passphrase wallet costs no blockchain fee. The cost is attention: recording the exact spelling, remembering which account uses it, and checking the destination before signing. Moving assets into that wallet can add ordinary source-chain gas, destination-chain gas, or a relayer fee; congestion, liquidity, and the route’s fee schedule determine the price, not the passphrase itself.
The same distinction matters whether the wallet touches Celestia Network or routes value through Celer Network and Symbiosis Finance: protocols see addresses and signatures, not the passphrase that derived them.
Without a passphrase, recovery is simpler and less error-prone. With one, a compromised seed does not automatically expose the protected wallet, but the operational burden becomes part of the security model.
The broader Manta Bridge question starts with how that route works.
My verdict is practical: use a passphrase when you can document the recovery procedure and test it with a small balance. Otherwise, the extra layer is merely an extra way to lock yourself out.