Plasma Cleaning in Hybrid Microelectronic Assembly

Introduction: Why Surface Contamination Is the #1 Problem

Every hybrid microelectronic assembly confronts the same invisible enemy: surface contamination. Even in cleanroom environments, organic residues from resist strips, flux carriers, mold release agents, and handling oils accumulate on substrate and die surfaces at the molecular level. These contaminants are invisible to optical inspection and typically not detectable without specialized surface analysis—but they have dramatic effects on the reliability of critical interfaces.

The wire bond interface between a gold or aluminum wire and a metallized pad is particularly sensitive. Ultrasonic energy during thermosonic ball bonding must break through surface contaminants to achieve true metallurgical contact. If a hydrocarbon layer separates the wire from the pad, the bond forms but does not adhere—leading to lifting during pull testing, parameter drift during temperature cycling, or field failures months later. Similarly, die attach void formation is driven primarily by surface contamination preventing wetting of the die attach material to the substrate or die back-metallization.

Plasma cleaning solves this problem by using reactive plasma to chemically break down organic contaminants into volatile species that are removed by vacuum pumping. Unlike manual solvent cleaning, plasma cleaning reaches into recesses, under wire bonds already placed, and into via arrays where solvents cannot effectively penetrate. It is the standard surface preparation step prior to die attach, wire bonding, lid sealing, and adhesive application in hybrid assembly.

Why Plasma Cleaning Matters

Surface Energy and Wetting

The quality of an adhesive or solder joint depends on the surface energy of the substrate relative to the contact angle of the applied material. A high-energy surface promotes spreading and intimate contact; a low-energy, contaminated surface causes beading and void formation. Surface energy is measured in dynes/cm (or mN/m), and the contact angle of a liquid droplet on a solid surface provides a direct measure.

Clean alumina and aluminum nitride substrates have surface energies of 40–50 dynes/cm. Organic contamination reduces this to 20–30 dynes/cm, which is below the critical threshold for most conductive epoxies (which require ≥ 35 dynes/cm for wetting) and solder pastes (which require ≥ 38 dynes/cm). After effective plasma treatment, surface energies on ceramic substrates typically increase to 45–55 dynes/cm, with the surface being fully wettable by standard materials.

Sources of Organic Contamination

• Photoresist residues from the substrate fabrication process, particularly after via formation and metal etching steps
• Flux residues from solder paste or preforms used in reflow attachment of surface-mount components
• Hydrocarbon oils from handling—skin oils, cutting fluids, vacuum pump oils, and temporary protective coatings
• Mold release agents from overmolding or casting processes applied to completed modules
• Outgassing products from polymeric materials in the assembly—wire insulation, underfill, glob-top materials—that migrate to surface during elevated temperature processing
• Surface oxides on aluminum and gold wire and die pads, which require a reducing atmosphere to remove

Contact Angle Measurement

Contact angle measurement is the simplest and most widely used in-process verification of surface cleanliness. A small deionized water droplet (typically 2–5 μL) is dispensed onto the surface, and the angle between the droplet edge and the surface is measured. A contact angle below 30° indicates a high-energy, clean surface; above 50° indicates a contaminated surface requiring additional treatment. Most hybrid assembly specifications require contact angles below 20° after plasma cleaning.

IPC Standards for Surface Cleanliness

IPC standards relevant to surface cleanliness in hybrid assembly include:

• IPC-TM-650 Method 2.3.28: Solvent extract conductivity (SEC) for ionic contamination measurement
• IPC-TM-650 Method 2.3.25: Ion exchange chromatography for ionic species identification
• IPC-J-STD-001: Requirements for soldered electrical and electronic assemblies
• MIL-STD-883 Method 3008: Microelectronic denuding (a type of surface treatment)

For military hybrids, surface cleanliness verification is typically required as part of the incoming inspection of substrates and is re-checked after plasma cleaning as part of the process control system.

O₂ vs Ar Plasma: Chemistry and Applications

Not all plasma cleaning gases are equal—and selecting the wrong gas chemistry for a given contamination type is one of the most common plasma cleaning mistakes.

O₂ Plasma

Oxygen plasma is the workhorse of plasma cleaning for hybrid microelectronics. The O₂ plasma generates reactive oxygen species (O, O₂⁺, O₃) that react with organic hydrocarbons to form CO, CO₂, and H₂O—volatile products that are pumped away. Key characteristics:

• Effective against: organic residues, photoresist, hydrocarbon oils, flux residues, polymer outgassing products
• Generates polar functional groups (hydroxyl, -OH) on treated surfaces, improving surface adhesion
• Suitable for: alumina, aluminum nitride, LTCC, printed circuit boards, metal leadframes
• May cause slight oxidation of sensitive metal surfaces (copper, silver) if process is too aggressive
• Not suitable for: gold wire already bonded (excessive oxidation risk to bond interface), temperature-sensitive polymers

Ar Plasma

Argon plasma operates by physical sputtering rather than chemical reaction. Argon ions (Ar⁺) are accelerated toward the substrate by the RF field and physically knock contaminant atoms from the surface (sputter etching). Key characteristics:

• Effective against: inorganic particles, metal oxides, particulate contamination embedded in recesses, dust
• Does not chemically react with the surface (no addition of oxygen or other species)
• Safer for temperature-sensitive components—Ar plasma generates less thermal load than O₂ plasma at equivalent power
• Does not remove organic contamination as effectively as O₂ plasma
• Useful as a pre-treatment to remove oxide layers from aluminum wire before wire bonding

N₂/H₂ Forming Gas Plasma

Forming gas (typically 5–10% H₂ in N₂) plasma provides a reducing atmosphere that removes oxide layers from metal surfaces. This is particularly useful for aluminum wire, which naturally forms an Al₂O₃ oxide layer that interferes with wire bonding. The H₂ in the plasma reduces Al₂O₃ to Al metal, exposing a fresh, clean surface for bonding. Key characteristics:

• Effective against: metal oxides (Al₂O₃, CuO, NiO), reducing surface oxides for improved wire bond adhesion
• Not effective against organic contamination (use O₂ for that first)
• Requires careful safety controls due to H₂ content (explosivity risk in large-scale systems)
• Standard for aluminum wire cleaning prior to thermosonic bonding

Process Parameters

Plasma cleaning outcomes are highly sensitive to the interaction of multiple process parameters. Getting the balance right—and controlling it from lot to lot—is the core challenge of plasma process engineering.

RF Power

RF power in plasma cleaning systems typically ranges from 100W to 500W for batch systems processing hybrid modules. Higher power increases plasma density and cleaning rate, but also increases substrate temperature. For assembled modules with temperature-sensitive components (polymeric coatings, adhesives that have not been fully cured), power should be limited to avoid exceeding 80–100°C at the substrate surface. Typical settings:

• 100–200W: Sensitive assemblies, low thermal budget; longer process time required
• 200–350W: Standard hybrid modules; good balance of cleaning rate and thermal load
• 350–500W: Bare substrates, thick residues; risk of thermal damage if not carefully controlled

Chamber Pressure

Operating pressure determines the mean free path of ions and the plasma density profile. Plasma cleaning systems typically operate between 100 mTorr and 500 mTorr (13–67 Pa). Lower pressure (100–200 mTorr) produces more anisotropic plasma—ions have longer mean free paths and arrive with more directional energy, making it more effective at cleaning deep features and recesses. Higher pressure (300–500 mTorr) produces more isotropic plasma with higher radical density, which is better for removing thick organic contamination from open surfaces.

Gas Flow Rates

Treatment Time

Treatment time ranges from 30 seconds for lightly contaminated surfaces to 5 minutes for heavy organic contamination. Over-cleaning is a real risk—extended O₂ plasma treatment (beyond 5–8 minutes) can begin to degrade aluminum metallization, reduce the adhesion of thin-film resistors, and cause corrosion of copper traces. Process validation should include verification of surface energy at multiple time points to determine the optimal treatment time for each substrate type and contamination level.

Temperature Rise Concerns

Plasma treatment generates heat in the substrate. In batch systems, the temperature rise is typically 20–50°C above ambient depending on power, time, and substrate thermal mass. Assembled modules with partially cured adhesives, polymeric conformal coatings, or plastic connector bodies may be sensitive to elevated temperature. For these assemblies:

• Use lower power (100–200W) and longer time to achieve cleaning without overheating
• Pulse the plasma (e.g., 30 seconds on, 30 seconds off) to allow thermal dissipation between cycles
• Implement substrate temperature monitoring with a thermocouple or infrared pyrometer if available
• Set maximum allowable temperature (typically 85–100°C for assemblies with polymeric components)

Benefits for Wire Bonding

Wire bonding is one of the most contamination-sensitive operations in hybrid assembly, and plasma cleaning has a direct, measurable impact on bond quality and reliability.

Gold Wire Bonding Improvements

Gold ball bonding onto aluminum pads (Au on Al) is a common hybrid configuration. The primary contamination concern is the organic interface between the ball bond and the pad that causes bond lift during pull testing. Plasma cleaning removes these hydrocarbons before bonding, improving initial bond adhesion and reducing the risk of stress-induced lifting during temperature cycling.

Additionally, O₂ plasma treatment of aluminum pads improves the interfacial chemistry between Au and Al, reducing the rate of Au-Al intermetallic compound (IMC) formation. While some IMC growth is inevitable and necessary for bond formation, excessive IMC growth causes bond embrittlement. Plasma treatment creates a more controlled, uniform IMC layer, improving long-term reliability.

Aluminum Wire Bonding

Aluminum wire wedge bonding benefits particularly from forming gas (N₂/H₂) plasma treatment. The native Al₂O₃ oxide layer on aluminum wire and pads is typically 2–5 nm thick and acts as an insulator, preventing true metallurgical contact during ultrasonic bonding. Forming gas plasma reduces this oxide to metallic Al, enabling bonding to proceed normally.

Key improvements documented in the literature and from production experience:

• Pull test strength improvement: Properly plasma-cleaned aluminum wire bonds typically show 15–30% higher pull test loads compared to non-cleaned controls
• Ball shear improvement: Ball shear values increase by 20–40% after plasma treatment, with failure mode shifting from lift (adhesive failure) to shear (cohesive failure), indicating stronger metallurgical bonding
• Reduced parameter drift: Wire bonds subjected to 1000-hour high-temperature storage (150°C) show less degradation in electrical resistance when surfaces are properly cleaned before bonding
• Lower ball bond diameter variability: Contamination causes inconsistent Free Air Ball (FAB) formation in the spark discharge, leading to variable ball sizes and inconsistent bond quality

Typical Pull Test Data Comparison

Note: Data typical for 25 μm gold wire on aluminum pad, 200W O₂ plasma, 300 mTorr. Actual results vary by substrate, metallization, wire type, and equipment.

Benefits for Die Attach

Die attach is the highest-thermal-resistance interface in a hybrid package, and voids in the die attach layer directly increase the junction temperature of the die. Plasma cleaning reduces void formation, which improves thermal performance and mechanical integrity.

Eutectic Die Attach

Gold-tin (AuSn) eutectic die attach is widely used for high-power and high-reliability applications. The eutectic reaction requires intimate contact between the die back-metallization and the substrate metallization—the die and substrate surfaces must be chemically clean and free of oxides and organics. Surface oxide on gold or nickel substrate metallization prevents wetting and causes void formation at the interface. O₂ plasma treatment removes organic contamination; N₂/H₂ plasma can reduce surface oxides. After plasma treatment, eutectic wetting coverage typically improves from 80–85% to 95–98%.

Conductive Epoxy Die Attach

Silver-filled conductive epoxy die attach is more forgiving than eutectic bonding, but surface contamination still affects wetting and adhesion. Contaminated surfaces cause the adhesive to bead, creating large voids that are visible in acoustic microscopy (C-SAM) inspection. Plasma cleaning improves wetting and reduces void area below the die to below 5% (typically required for thermal performance), compared to 10–20% void area on non-cleaned controls.

The contact resistance reduction from improved wetting is significant: properly plasma-cleaned conductive epoxy joints show 20–40% lower contact resistance compared to non-cleaned joints. For high-current power devices, this translates to lower I²R losses and better thermal distribution.

Quality Verification Methods

Plasma cleaning is only as good as the verification methods used to confirm its effectiveness. A plasma process without measurement is a process without control.

Contact Angle Measurement

The water droplet contact angle test is the simplest and most rapid in-process check. Using a goniometer or a calibrated camera system with drop shape analysis software, a 2–5 μL deionized water droplet is placed on the substrate surface and the contact angle is measured. Acceptance criteria are typically:

• Contact angle ≤ 20° for alumina and aluminum nitride substrates
• Contact angle ≤ 30° for LTCC and polymer-coated surfaces
• Repeatability: ±3° across measurement positions on a single substrate

Surface Energy Calculation (Fowkes Method)

The Fowkes method provides a more complete characterization of surface energy by measuring contact angles with multiple test liquids (typically water and diiodomethane). This allows calculation of the dispersive and polar components of surface energy, which correlate with adhesive and solder wetting behavior. Clean hybrid substrates should show:

• Total surface energy: 45–55 mN/m
• Polar component: > 25 mN/m (indicates presence of polar functional groups that improve adhesion)

XPS (X-ray Photoelectron Spectroscopy)

XPS provides elemental composition of the top 5–10 nm of the surface. It can detect:

• Carbon contamination (hydrocarbons from organics)
• Oxide thickness on metal surfaces
• Presence of adventitious contaminants (sodium, potassium, silicon from handling)

XPS is typically used for process development and qualification rather than production lot acceptance, as it requires a vacuum instrument and skilled operator. A cleaned surface should show carbon signals below 10 atomic percent (ideally below 5%).

Total Organic Carbon (TOC) Measurement

TOC analysis measures the total amount of organic carbon on a surface by combusting the sample and measuring the CO₂ produced. It is quantitative and reproducible, making it suitable for process control specification. Typical TOC acceptance levels:

• Ceramic substrates: < 1 μg/cm²
• Metal leadframes: < 0.5 μg/cm²

Common Mistakes in Plasma Cleaning

Despite being a mature process, plasma cleaning failures are common. The following mistakes account for the majority of quality issues observed in hybrid assembly plasma operations.

Over-Cleaning

Running plasma for too long or at too high a power causes surface degradation. On aluminum metallization, extended O₂ plasma forms a thick Al₂O₃ layer (visible as a color change from silver to rainbow) that actually worsens wire bondability rather than improving it. On copper, over-oxidation creates a non-conductive oxide layer that affects solderability. On polymer materials in the assembly, over-cleaning can cause embrittlement or delamination at interfaces.

Fix: Establish a process window with a defined minimum and maximum treatment time. Validate using contact angle and wire pull testing. Document the validated time range and do not exceed it.

Incorrect Gas Selection

Using only O₂ plasma on aluminum wire that requires oxide reduction will not achieve the desired result. The O₂ plasma removes organics but does not reduce the Al₂O₃ oxide. Conversely, using Ar plasma alone for heavy organic contamination will leave significant residue. Many assemblies require a two-step process: O₂ plasma for organics, followed by N₂/H₂ plasma for oxide reduction.

Fix: Identify the primary contamination type before selecting gas chemistry. For unknown contamination, use O₂ + Ar mixed plasma as a starting point and validate with contact angle and bond pull testing.

Ignoring Batch-to-Batch Variation

Plasma chamber condition changes with use. Electrode erosion, chamber wall deposits, and gas delivery variations cause the same process parameters to produce different results over time. A process qualified six months ago may not produce the same results today without chamber maintenance.

Fix: Implement statistical process control with regular verification substrates. Track contact angle data over time and set control limits. Perform preventive maintenance on plasma chamber electrodes and seals on a defined schedule (typically every 500–1000 cycles).

Ignoring Fixture Shadowing

Parts placed in the plasma chamber near the edges, against walls, or behind shielding fixtures receive significantly less plasma exposure than parts in the center. The plasma density profile in a batch chamber is not uniform—it is typically higher near the electrodes and lower near chamber walls. Parts in shadow regions may be under-cleaned while parts near the center are over-cleaned.

Fix: Map the chamber with a uniform test substrate (contact angle mapping across all positions) to identify under-cleaned zones. Design fixture layouts to avoid shadow regions. Rotate parts or use multiple positions within a batch for critical production runs.

Not Validating After Process Changes

Any change to the substrate supplier, metallization, cleaning chemistry, or packaging can invalidate the plasma process. Making a process change without re-validation is the most common cause of sudden yield drops in wire bonding and die attach.

Fix: Establish a change control protocol that requires re-validation of plasma process effectiveness (contact angle, wire pull, die attach void %) whenever a new material or supplier is introduced.

Conclusion

Plasma cleaning is an enabling technology for hybrid microelectronics—but only when it is properly qualified, controlled, and verified. The consequences of skipping or incorrectly performing this step are expensive: field failures, customer returns, and the cost of re-building or re-screening assemblies that failed due to contamination-driven defects.

The key principles to follow:

• Match gas chemistry to contamination type — O₂ for organics, Ar for particles, N₂/H₂ for metal oxides
• Validate the process window — Know the minimum time needed for effective cleaning and the maximum time before degradation occurs
• Implement measurement-based process control — Contact angle measurement and statistical process control are not optional; they are the foundation of a reliable plasma process
• Monitor for batch-to-batch variation — Plasma chamber condition changes over time; maintain records and implement trend monitoring
• Control fixture layout — Prevent shadow regions and map the chamber for uniform process coverage

When engaging a contract manufacturer for hybrid assembly, verify their plasma cleaning qualification process, ask for SPC data on contact angle measurements over recent production lots, and confirm that they validate the process after any material or supplier change. A manufacturer who cannot produce this data is running the plasma process without control—and that is a risk that propagates into your final module reliability.