Ephemeral Dust: Why Bitcoin Now Allows Zero-Value Outputs (and Why It's Safe)
Bitcoin's relay policy now permits zero-value outputs as ephemeral dust, enabling cleaner fee bumping for Lightning and other protocols.
For over a decade, Bitcoin Core has refused to relay transactions containing outputs below a minimum value. This dust limit exists to prevent the UTXO set from accumulating tiny, unspendable outputs that every full node must store indefinitely. But in 2025, Bitcoin Core introduced a targeted exception: a zero-value output can now pass through the relay network if it is guaranteed to be spent immediately. This exception, called ephemeral dust, is part of a broader set of relay policy upgrades that reshape how Layer 2 protocols handle fees on-chain.
What Is the Dust Limit and Why Does It Exist
Every Bitcoin transaction creates outputs, and every output occupies space in the UTXO set until it is spent. Full nodes keep this entire set in memory for fast validation. If a user creates a 1-satoshi output, the cost of spending it (the transaction fee for the input) will almost certainly exceed the output's value. The rational response is to never spend it, which means every full node carries that output forever.
The dust limit is a mempool policy rule, not a consensus rule. Miners can still include dust outputs in blocks, but Bitcoin Core nodes will not relay transactions that create them. The threshold is calculated by multiplying the output's serialized size plus the estimated size of the input needed to spend it by the dust relay fee rate (default 3 sat/vB).
Dust Thresholds by Output Type
| Output Type | Dust Threshold (sats) | Calculation |
|---|---|---|
| P2PKH | 546 | 3 × (34 + 148) = 546 |
| P2WPKH | 294 | 3 × (31 + 67) = 294 |
| P2WSH / P2TR | 330 | 3 × (43 + 67) = 330 |
| P2A (Pay-to-Anchor) | 240 | 3 × (27 + 53) = 240 |
These thresholds have been part of Bitcoin Core since version 0.8.2 in 2013. The concept was originally proposed to protect the network from UTXO bloat, and it has worked: the UTXO set grows primarily through economically meaningful outputs, not spam.
Policy vs. consensus: The dust limit is enforced only at the relay and mempool level. A miner who includes a 1-satoshi output in a block produces a valid block. The rule shapes what most nodes will propagate, not what the network will accept.
The Problem: Fee Bumping in Layer 2 Protocols
The dust limit creates friction for fee bumping in protocols like Lightning. When a Lightning channel force-closes, the commitment transaction must be broadcast on-chain. But the transaction was signed hours, days, or weeks earlier, and the fee rate it carries may no longer match current mempool conditions.
Lightning's solution has been anchor outputs: small outputs attached to the commitment transaction that either party can spend via CPFP to add fees at broadcast time. Under the current anchor channel format (defined in BOLT 3), each commitment transaction carries two anchor outputs, one per channel participant, each valued at 330 satoshis (the P2WSH dust threshold).
Why Keyed Anchors Are Expensive
The 330-satoshi anchor is the minimum value that Bitcoin Core will relay, but it creates several costs:
- Each commitment transaction locks up 660 satoshis across two anchors that exist solely to enable fee bumping
- Unspent anchors persist in the UTXO set, contributing to the bloat the dust limit was designed to prevent
- The
update_feemessage requires both parties to agree on a feerate at signing time, triggering force-closes when they disagree during fee spikes - Each anchor is keyed to one party for 16 blocks, adding script complexity and limiting who can fee-bump
The irony is clear: the dust limit was designed to keep uneconomical outputs out of the UTXO set, but anchor outputs that go unspent become exactly that. The protocol needs a way to create a fee-bumping handle that costs zero satoshis and never touches the UTXO set.
The Solution: Three Interlocking Policy Changes
Bitcoin Core addressed this with three separate but coordinated policy upgrades, shipped across two major releases. Together, they form a cohesive stack for fee bumping that eliminates the anchor output cost problem.
1. TRUC Transactions (v3): BIP 431
TRUC (Topologically Restricted Until Confirmation) transactions, defined in BIP 431, are an opt-in mempool policy for transactions that set nVersion = 3. Bitcoin Core 28.0 introduced this policy, which constrains the unconfirmed topology to at most one parent and one child.
The restrictions are designed to prevent transaction pinning attacks, where an adversary attaches a large, low-feerate descendant to a victim's transaction, making it prohibitively expensive to replace. Under TRUC rules, the child transaction is capped at approximately 1,000 virtual bytes, which bounds the cost of replacement. If a conflicting child arrives with better incentive compatibility, Bitcoin Core evicts the existing child (sibling eviction).
2. Pay-to-Anchor (P2A): BIP 433
Pay-to-Anchor is a new standard output type defined in BIP 433. The script is OP_1 <0x4e73>, a SegWit v1 output that anyone can spend with an empty witness. Bitcoin Core 28.0 made spending P2A outputs standard.
Because P2A is keyless, any party that sees the commitment transaction can create a child that spends the anchor and attaches fees. This eliminates the 16-block delay and per-party keying of the old anchor format. The mainnet address for the P2A script is bc1pfeessrawgf.
3. Ephemeral Dust: Bitcoin Core PR #30239
The final piece arrived in Bitcoin Core 29.0, released in April 2025. PR #30239 introduced the ephemeral dust policy: a transaction may create a single dust-value output (including zero-value) if the transaction itself pays zero fees and a child transaction in the same package spends the dust output.
The zero-fee requirement is the key safety mechanism. Because the parent pays no fee, no miner has an incentive to include it without its child. The child must spend the dust output, so by the time the transactions are confirmed, the dust output has already been consumed. It never persists in the UTXO set.
Why zero fee matters: If a dust-creating transaction paid its own fee, a miner could profitably include it alone, leaving the dust output stranded in the UTXO set. The zero-fee requirement ensures the parent is worthless without its child, so the dust is always ephemeral.
How the Pieces Fit Together
These three policies combine into what is commonly called an ephemeral anchor. The pattern works as follows:
- A commitment transaction is constructed as a TRUC (v3) transaction with zero fees
- The transaction includes a single P2A output with zero value
- At broadcast time, whoever needs to push the transaction on-chain creates a child that spends the P2A output and sets the fee rate appropriate for current mempool conditions
- Bitcoin Core's 1P1C package relay (shipped in Core 28.0 via PR #28970) delivers both transactions to the mempool as a unit
- TRUC rules prevent pinning by limiting the child size and enabling sibling eviction
- The ephemeral dust exception allows the zero-value P2A output to pass relay despite being below the dust threshold
The result: a commitment transaction that carries no fees, creates no persistent UTXO bloat, and can be fee-bumped by anyone at any time.
Before and After: Anchor Output Comparison
The transition from keyed anchors to ephemeral anchors changes several properties of Lightning commitment transactions. This table compares the two approaches:
| Property | Keyed Anchors (Current) | Ephemeral Anchors (New) |
|---|---|---|
| Number of anchor outputs | 2 (one per party) | 1 (shared) |
| Anchor value | 330 sats each (660 total) | 0 sats |
| Who can fee-bump | Keyed party for 16 blocks, then anyone | Anyone, immediately |
| Commitment transaction fee | Non-zero (agreed via update_fee) | Zero |
| Feerate negotiation | Required (update_fee message) | Eliminated |
| UTXO set impact | Unspent anchors persist | Anchor never enters UTXO set |
| Pinning resistance | CPFP carve-out (limited) | TRUC topology limits + sibling eviction |
| Transaction version | v2 | v3 (TRUC) |
| Required Bitcoin Core version | Any | 29.0+ |
The elimination of update_fee is arguably the most impactful change for Lightning operators. During mempool congestion spikes, disagreements over the commitment feerate have historically triggered unintended force-closes. With zero-fee commitments, that failure mode disappears entirely.
Why This Is Safe: The UTXO Set Argument
The obvious concern with allowing zero-value outputs is UTXO set bloat. If anyone could create zero-value outputs freely, the UTXO set would grow without bound. The ephemeral dust policy avoids this through two constraints:
- The parent transaction must pay zero fees, so no miner has an incentive to mine it alone
- The child transaction must spend the dust output, so the dust is consumed before confirmation
In practice, the dust output exists only in the mempool, only briefly, and only as part of a package. It never appears in the UTXO set after block confirmation. The safety argument rests on economic incentives: a zero-fee transaction is worthless to mine without its fee-paying child, and the child is required to clean up the dust.
Edge Cases and Limitations
The policy does have boundaries. Only one dust output per transaction is permitted. The parent must be a TRUC transaction to benefit from the full anti-pinning protections. And because ephemeral dust is a relay policy, not a consensus rule, miners could theoretically mine a zero-fee transaction without its child. In practice this would mean voluntarily forgoing revenue, which is not an economically rational strategy.
There is also a mempool-level protection: Bitcoin Core's cluster mempool tracks transaction relationships and enforces that the dust-spending child is present before accepting the parent. If the child is evicted (for example, during mempool pressure), the parent is evicted along with it.
Impact on Lightning and Layer 2 Protocols
The BOLT specification update (PR #1228, merged May 2026) defines a new channel type using feature bits 40/41. This channel type uses TRUC commitment transactions with a single zero-value P2A output, replacing the two keyed anchors.
Implementation progress varies across Lightning node software. LDK shipped support for zero-fee commitment channels in version 0.2 (December 2025). Eclair began integrating support in early 2026. Core Lightning has experimental support in development. As nodes upgrade, channels opened with the new format will benefit from cheaper force-closes and the elimination of feerate disagreements.
Beyond Lightning
Ephemeral dust is not Lightning-specific. Any protocol that pre-signs transactions and needs to defer fee selection benefits from this pattern. The Bitcoin Core 29.0 release notes explicitly mention several protocols that can leverage ephemeral dust:
- Ark timeout trees, where pre-signed exit transactions need flexible fee bumping
- ln-symmetry (eltoo), which requires update transactions to carry adjustable fees
- Spacechains and other constructs that anchor data to Bitcoin's chain
- Vault protocols using pre-signed recovery transactions
The Complete Fee-Bumping Stack
It helps to see these policy changes as layers of a single stack, each addressing a different aspect of the transaction pinning and fee-bumping problem:
| Layer | Component | Bitcoin Core Version | What It Solves |
|---|---|---|---|
| Relay | 1P1C package relay (PR #28970) | 28.0 | Delivers parent + child as a unit |
| Topology | TRUC / v3 (BIP 431) | 28.0 | Bounds unconfirmed chain, prevents pinning |
| Output | P2A standardness (BIP 433) | 28.0 | Keyless anyone-can-spend anchor |
| Dust | Ephemeral dust (PR #30239) | 29.0 | Zero-value outputs that never persist |
Each layer depends on the ones below it. Ephemeral dust works because TRUC limits the topology. TRUC is practical because 1P1C relay carries the package. P2A makes the anchor keyless. Together, they replace a system that required 660 satoshis of permanent UTXO bloat and ongoing feerate negotiation with one that costs nothing and delegates fee decisions to broadcast time.
What This Means for the Broader Bitcoin L2 Landscape
These relay policy improvements reduce on-chain costs for Lightning and other channel-based Layer 2 protocols. For the Lightning Network specifically, the elimination of update_fee removes a significant source of unintended force-closes, improving channel stability and reducing the operational burden on routing node operators.
Spark, as a channelless Layer 2, does not use anchor outputs or commitment transactions in the Lightning sense. Its statechain-based architecture handles fee bumping differently: pre-signed exit transactions can be constructed with appropriate fee reserves, and the protocol's leaf structure allows flexible value splitting without on-chain interaction. But a healthier, more cost-efficient Lightning ecosystem benefits Bitcoin's overall Layer 2 landscape. Many wallets and services interoperate across both networks, and improvements to Lightning fee management reduce friction for the entire payment layer.
For developers building on Bitcoin's Layer 2 ecosystem, the Spark SDK documentation covers how Spark handles fee management and exit transactions without anchors. Those interested in seeing these improvements in action through a production wallet can explore General Bread, a Spark-powered wallet that benefits from the broader Layer 2 ecosystem these relay improvements enable. For a deeper look at how fee markets and mempool dynamics affect all Bitcoin protocols, see our research on mempool congestion economics and the Pay-to-Anchor and v3 fee-bumping guide.
This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.

