Carbon-Free and Chlorine-Free: What TSA Brings to ALD Silicon Nitride

Most silicon precursors carry baggage. Chlorosilanes carry chlorine. Organosilanes carry carbon. For decades, process engineers simply absorbed the consequences — extra byproduct management, etch-back steps, carbon contamination concerns — because the alternatives did not exist at production purity or volume. Trisilylamine, usually abbreviated TSA, changes that equation. It is one of the few silicon sources that delivers silicon to a wafer surface without chlorine and without carbon, and that single property is reshaping how advanced ALD silicon nitride processes are designed.

This article looks at what makes trimethylsilylamine (TSA) chemically unusual, where it outperforms the established precursors, and where the honest trade-offs sit.

The Byproduct Problem Nobody Likes Talking About

Consider what happens inside a deposition chamber with a conventional precursor. With dichlorosilane (DCS), the chlorine leaves as HCl and ammonium chloride species. In a nitride LPCVD furnace at 700 °C or above, that is manageable — furnaces are built for it, and the exhaust chemistry is well understood. But drop the temperature into the ALD window, below 400 °C, and chlorine byproducts start to matter in new ways: they can etch growing film, they leave chloride residues in high-aspect-ratio structures, and they corrode everything downstream of the chamber.

Organosilane precursors trade chlorine for carbon. Tetramethylsilane (4MS) and trimethylsilane (3MS) deposit silicon efficiently in plasma-enhanced processes, but every ligand that does not end up in the film leaves as a carbon-containing fragment. In PEALD silicon nitride, residual carbon and hydrogen in the film are the two impurities every film metrology team tracks first, because both degrade wet etch resistance and dielectric reliability.

TSA sidesteps both issues. Its structure is simple to state: three silyl groups attached to a nitrogen center. Every atom in the molecule is either silicon, nitrogen, or hydrogen — the same elements you want in the film. There is no carbon to remove and no chlorine to scrub. The byproduct chemistry in an ALD cycle reduces almost entirely to ammonia and hydrogen, which existing exhaust and abatement systems handle without special measures.

Volatility: A Practical Advantage, Not Just a Datasheet Number

The second thing people notice about TSA is that it boils at a low temperature and delivers high vapor pressure at near-ambient conditions. For a liquid precursor, this is operationally significant. Delivery systems for less volatile liquids — BDEAS, for example — need heated ampoules, heated lines, and careful vapor pressure management to deliver stable dose pulses. Every heated joint is a potential condensation point and a maintenance item.

A highly volatile liquid self-delivers more easily, and stable vapor delivery translates directly into dose repeatability, which is the currency ALD lives and dies by. When your dose varies, your per-cycle growth per cycle varies, and your film thickness uniformity across a batch quietly drifts. The link between precursor delivery stability and final film metrology is direct, and higher volatility removes an entire class of variability.

Where TSA Fits in the Process Landscape

TSA is not a universal replacement, and pretending otherwise would be a disservice. Its clearest fit is thermal or plasma-assisted ALD silicon nitride where three constraints converge: low temperature, high aspect ratio, and film purity.

The conformality argument deserves its own explanation. In structures with aspect ratios above 10:1 — the regime modern 3D NAND and advanced DRAM capacitors live in — the precursor must penetrate the structure and react on the sidewalls before it saturates. TSA is a relatively small, unstrained molecule, which helps its sticking behavior and penetration depth compared with bulkier aminosilanes such as DIPAS. Bulkier ligands can improve thermal stability, but that stability is purchased with slower diffusion into deep features. Where the feature geometry dominates the process window, molecular size becomes a first-order selection criterion.

Against halide routes, TSA's case is purity-driven. For spacers, liners, and seal films where any chloride residue attacks either the film itself or adjacent materials, a chlorine-free process removes the failure mechanism rather than shrinking it. Some fabs have found that switching the nitride liner in a given integration from a DCS-based process to a TSA-based ALD process eliminates an entire wet-clean step designed to strip chloride residue — a process simplification worth more than the precursor price delta.

For iodine chemistry, the comparison with diiodosilane (DIS) is instructive. DIS offers excellent low-temperature deposition and clean byproducts, but iodine is expensive, corrosive, and needs dedicated management. TSA offers a similar "clean" byproduct story at materially lower precursor cost and simpler handling, which is why it increasingly appears in production evaluations rather than just research papers.

The Honest Trade-Offs

Three limitations come up repeatedly in evaluations.

First, thermal ALD reactivity. TSA's Si-N bonds do not react with surface hydroxyl groups as readily as chlorosilanes do, so purely thermal TSA/ammonia cycles benefit from a plasma assist or a catalytic step at the lower end of the temperature window. In PEALD, this is a non-issue; in thermal-only tools, it needs qualification attention.

Second, oxygen sensitivity. Like all silanes, TSA hydrolyzes. It is a pyrophoric-adjacent material that demands inert handling, and its reactivity with moisture exceeds that of conventional aminosilanes. Gas cabinet design, cylinder connection procedures, and delivery line purge practices all need to be specified for it from day one — retrofitting a delivery system designed for BDEAS rarely goes smoothly.

Third, supply maturity. TSA has historically been available in small research quantities from a limited number of producers. Any process decision should include a supply audit: production capacity, lot-to-lot analytical consistency, and cylinder-level packaging that matches your gas panel. A precursor that qualifies beautifully but cannot be sourced at 500 cylinders per year with consistent 8N-plus purity is a research material, not a production material.

How to Run the Evaluation

Fabs that evaluate TSA successfully tend to follow a short, disciplined sequence. Start with a chlorosilane-free baseline: run your current PEALD nitride process, collect film composition, wet etch rate, and conformality data on your worst-case structure. Then run TSA under equivalent dose-normalized conditions, holding temperature and plasma power constant. The comparison metrics that matter most are carbon and chlorine content by SIMS or XPS, wet etch rate in hot phosphoric acid, and step coverage in the deepest structure you actually build. Qualification on real product, not monitor wafers alone, closes the loop — the same principle that governs any precursor change, as covered in our earlier articles on precursor qualification and second sourcing.

For teams also running CVD processes on the same platform, note that silane (SiH4) and disilane (Si2H6) remain the workhorses for higher-temperature nitride and silicon deposition, and monochlorosilane (MCS) continues to serve high-temperature selective applications well. TSA is not displacing them; it is filling the low-temperature, high-purity ALD slot that previously had no clean answer.

If you are evaluating a chlorine-free, carbon-free silicon source for an advanced ALD process, contact us for specifications, purity data, and sample support.