Inline vs. Cluster PVD for TGV Metallization
Share
Choosing between Cluster PVD and Inline PVD for TGV metallization is not simply a question of which vacuum coating system is faster. The decision depends on through-glass via (TGV) process feasibility, copper seed-layer continuity, glass substrate handling, process flexibility, production capacity and total cost per good substrate.
This guide focuses on the equipment-selection factors that matter to process and manufacturing engineers. It distinguishes equipment architecture from deposition physics: changing from Cluster to Inline does not, by itself, solve the challenge of coating high-aspect-ratio through-glass vias.
1. Cluster and Inline PVD: What Is the Difference?
Both architectures can integrate substrate preparation and metal deposition under vacuum. Their main difference is how substrates move through the process and how process modules are arranged.
Cluster PVD
A Cluster PVD system typically uses a central vacuum transfer module connected to multiple process chambers. A substrate is transferred between chambers for steps such as pre-cleaning, adhesion-layer deposition and copper seed-layer deposition.
- Modular chamber arrangement supports different process sequences and recipe development.
- Independent process chambers can be configured for different materials or treatments.
- Single-substrate or batch processing may be used, depending on the system design.
- Transfer-module capacity, chamber scheduling and chamber utilization can constrain throughput.
Inline PVD
An Inline PVD system transports substrates through a sequence of connected process zones. Loading, surface treatment, deposition and unloading sections can be integrated into a coordinated production flow. Substrate movement may be continuous, intermittent or stepwise, depending on the transport and process design.
- Can be designed around rectangular or large-format glass panels.
- Supports sequential processing and integration with automated material handling.
- Can provide stable, repeatable production flow when the line is properly balanced.
- Changes to process sequence, product mix or process time may affect line layout and takt time.
2. Practical Equipment Comparison
The following comparison describes common architectural tendencies. It is not an absolute rule: either architecture can be customized, and the detailed machine design determines actual performance.
| Selection factor | Cluster PVD | Inline PVD |
|---|---|---|
| Production mode | Single-substrate or batch operation, depending on chamber and transfer design. | Sequential substrate transport; movement may be continuous, intermittent or stepwise. |
| Throughput driver | Process time, chamber count, transfer time, scheduling and load/unload time. | Line takt time, bottleneck zone, transport speed, process-zone length and availability. |
| Product mix | Often advantageous for multiple recipes and frequent process changes. | Often advantageous for stable products and a defined process sequence. |
| Thin-glass handling | Robot, edge-grip, carrier or other transfer design; acceleration and handoff are key risks. | Roller, carrier, support or other transport design; sag, vibration and transfer points require attention. |
| Large panels | Possible, subject to chamber size, robot reach and transfer-module design. | Can be designed around panel dimensions and a dedicated transport path. |
| Pre-clean / anneal integration | Can use dedicated process chambers; module addition may require transfer and capacity review. | Can add process zones; line length, thermal isolation and takt-time balance must be reviewed. |
| Footprint | Driven by chamber count, transfer module, service access and layout. | Driven by line length, process zones, loading and unloading arrangement. |
| Maintenance | Individual chambers may be serviced separately, but shared transfer equipment can remain a dependency. | Zones may be serviced individually, but maintenance can affect line flow or require bypass strategies. |
| Investment structure | Driven by process modules, transfer automation, chamber count and redundancy. | Driven by line length, number of zones, transport automation and production capacity. |
3. How to Compare Throughput and Cost
Do not compare a Cluster tool's batch capacity with an Inline tool's nominal transport speed. Compare accepted substrates produced over the same scheduled period, using the same substrate, film stack and acceptance criteria.
Cluster: cycle-based capacity
A simplified gross-capacity estimate is:
Cycle time should include loading, transfer, processing, unloading and any scheduling constraint that limits the next cycle. If several chambers operate in parallel, calculate the actual bottleneck and chamber utilization rather than multiplying nominal chamber capacity blindly.
Inline: takt-based capacity
A simplified gross-capacity estimate is:
The effective takt time is governed by the slowest process or handling stage. Deposition time, zone length, transport speed, substrate spacing, load/unload and interruptions all affect the result. For a step-and-repeat system, use the complete interval between successive substrate starts, not just the time spent moving the substrate.
Illustrative calculation: Cluster vs. Inline
The following example uses hypothetical values only. It demonstrates how to compare the architectures using a consistent calculation method; it is not a SIMVACO equipment specification or a typical industry benchmark.
| Input | Illustrative Cluster system | Illustrative Inline system |
|---|---|---|
| Substrates per cycle / effective takt | 4 substrates per completed cycle | 1 substrate every 180 seconds |
| Cycle time | 900 seconds per completed cycle | 180 seconds per substrate |
| Gross throughput | 4 × 3,600 ÷ 900 = 16 substrates/hour | 3,600 ÷ 180 = 20 substrates/hour |
| Availability assumption | 85% | 85% |
| Yield assumption | 95% | 92% |
| Estimated good output | 16 × 0.85 × 0.95 = 12.92 substrates/hour | 20 × 0.85 × 0.92 = 15.64 substrates/hour |
Under these assumptions, the Inline example produces more accepted substrates per scheduled hour. However, this result does not establish that Inline is inherently more productive. A different Cluster chamber configuration, cycle time, availability or process yield could change the comparison.
Compare cost per good substrate
A useful first-pass model is to divide the total cost over a defined operating period by the number of accepted substrates produced during that same period.
Include equipment depreciation or financing, labor, utilities, targets and consumables, maintenance, planned and unplanned downtime, changeover losses and process yield. Use the same accounting period and cost boundaries for both systems.
Illustrative cost comparison
For example, assume each system runs 400 scheduled production hours per month and the illustrative hourly operating costs below apply. These figures are hypothetical and exclude capital cost, financing and depreciation.
| Metric | Cluster example | Inline example |
|---|---|---|
| Good output per hour | 12.92 substrates | 15.64 substrates |
| Scheduled hours per month | 400 | 400 |
| Monthly good output | 5,168 substrates | 6,256 substrates |
| Illustrative operating cost | US$320/hour | US$430/hour |
| Monthly operating cost | US$128,000 | US$172,000 |
| Operating cost per good substrate | US$24.77 | US$27.49 |
In this example, Inline has higher output but also a higher operating cost per accepted substrate. The result could change after including capital cost, labor model, target utilization, maintenance strategy, product mix and actual yield. The purpose of the model is to make assumptions visible—not to declare one architecture cheaper.
Key production metrics
| Metric | What to measure |
|---|---|
| Good substrates per hour | Accepted output, not theoretical movement or deposition rate. |
| Availability | Operating time relative to scheduled production time, using a clearly defined downtime policy. |
| Yield | Percentage meeting agreed film, coverage and electrical criteria. |
| Utilization | Actual productive use relative to the time the equipment is available for production. |
| Changeover | Time and material loss when changing product, recipe or target. |
| Cost per good substrate | Total cost within the defined accounting boundary divided by accepted output. |
Availability, utilization and yield should be defined separately to avoid double-counting downtime or production losses. The example above applies availability and yield to nominal throughput; it does not apply a separate utilization factor. If utilization further reduces the time available for production, include it explicitly and consistently in both models. No universal substrates-per-hour range is stated here because capacity changes substantially with panel size, film thickness, target configuration, process time, handling and acceptance criteria.
4. TGV Sidewall Coverage: Deposition Physics and Engineering Options
TGV is a through-hole in glass. This article uses via entrance, sidewall and via exit to describe the two openings and the internal wall. The aspect ratio (AR) is commonly expressed as glass thickness divided by via opening diameter. For tapered vias, the entrance and exit diameters should be stated separately; a single nominal diameter can conceal meaningful geometric differences. For an initial conservative geometry check, engineers may also calculate thickness divided by the minimum via opening; this is a screening ratio, not a complete description of the via profile or deposition difficulty.
Conventional magnetron sputtering has a directional component. Material arriving at the via entrance may not reach deeper sidewall regions or the opposite opening at the same rate. Coverage depends on via diameter, depth, taper, profile, target-to-substrate geometry, particle angular distribution, pressure, substrate orientation and the required film thickness.
Illustrative TGV geometries
| Via opening diameter | Glass thickness | Aspect ratio | Engineering interpretation |
|---|---|---|---|
| 100 μm | 300 μm | 3:1 | Lower geometric challenge than the examples below; continuity still requires validation. |
| 80 μm | 400 μm | 5:1 | A useful caution point for process evaluation, not a universal PVD limit. |
| 50 μm | 400 μm | 8:1 | More demanding material access; via profile and deposition geometry become increasingly important. |
| 50 μm | 500 μm | 10:1 | High geometric challenge; feasibility should be demonstrated on representative samples. |
These are illustrative geometric examples, not SIMVACO process capability claims or universal PVD thresholds. For tapered vias, identify whether the stated opening diameter is measured at the entrance, exit or narrowest section.
Common engineering approaches and their limitations
Several equipment and process approaches may influence the distribution of deposited material inside a TGV. They are not interchangeable, and none should be treated as a guaranteed solution for high-aspect-ratio coverage. Their usefulness depends on the via profile, film stack, material, equipment geometry and electrical continuity requirement.
| Engineering approach | Potential effect | Key limitations and validation points |
|---|---|---|
| Target angle and source positioning | Changes the incident-flux direction and the regions of the via exposed to sputtered material. | An angle that improves access to one sidewall region may reduce deposition elsewhere. Validate the full cross-section and both via openings. |
| Long-throw sputtering | Uses a larger target-to-substrate distance to modify the arriving flux distribution; in some configurations it can reduce the contribution of highly oblique trajectories. | May reduce deposition rate and require more source power, target area or process time. It does not inherently provide conformal coverage inside deep vias; geometry and pressure remain important. |
| Specialized magnetron / magnetic-field design | Changes plasma confinement, discharge distribution and, depending on source design, the spatial distribution of sputtered material. | Magnetic-field design primarily affects plasma and source behavior; it does not directly guarantee a favorable flux path into every via. Confirm plasma stability, erosion profile, uniformity and deposition distribution. |
| Substrate tilt, rotation or reorientation | Changes the substrate orientation relative to the arriving flux and can average directional non-uniformity. | Rotation does not automatically provide conformal coverage deep inside a high-AR via. Mechanical clearance, carrier design, process time and repeatability must be considered. |
| Double-sided deposition | Introduces material from both via openings and may improve access to regions shadowed during single-sided processing. | The two deposition profiles must overlap sufficiently to form a continuous conductive path. Flipping, alignment, handling, contamination control and added cycle time require evaluation. |
| Ionized PVD (I-PVD) | Increases the ionized fraction of sputtered material. Ion transport and energy can be influenced by plasma conditions and suitable electric-field configurations. | Ionization does not automatically ensure deep sidewall coverage. Ion energy, plasma distribution, charging and the electrical behavior of insulating glass must be considered. Substrate-bias assumptions used for conductive wafers cannot be transferred directly to glass. |
| Higher working pressure | Increases gas-phase collisions and can broaden the angular distribution of arriving sputtered species under suitable conditions. | More scattering may reduce directionality, but can also lower deposition rate and change film density, energy, uniformity and properties. It is not a universal route to improved via coverage. |
Single-sided vs. double-sided deposition
With single-sided deposition, the flux enters from one opening. The entrance region may receive more material while deeper sidewall regions and the via exit receive less, depending on geometry and process conditions. Double-sided deposition introduces flux from the opposite opening and may improve access to regions that are shadowed during single-sided processing.
However, double-sided deposition does not guarantee uniform thickness. The two deposition profiles must overlap sufficiently to provide a continuous conductive path. Substrate flipping, alignment, edge handling, contamination control and additional cycle time also affect equipment design and production economics.
Seed-layer continuity is the critical acceptance question
For a TGV copper metallization flow, the objective of the PVD seed layer is not simply to achieve a specified thickness on the planar glass surface. It is to establish a sufficiently continuous conductive path along the required via surfaces so that the subsequent plating process can proceed reliably.
A planar thickness measurement alone cannot prove continuity inside the via. Local thin regions, discontinuities or poor connection between deposition fronts may cause plating defects even when the planar film thickness appears acceptable. Acceptance criteria should therefore define the required coverage and continuity at the entrance, sidewall and exit.
Cluster vs. Inline: what actually changes?
Neither Cluster nor Inline inherently produces better sidewall coverage. Either architecture may incorporate angled sources, substrate rotation or tilt, double-sided processing, or specialized ionized-PVD modules if the mechanical and vacuum design supports them.
The architecture affects how these features are integrated and how substrates are handled. The deposition result is determined by the source configuration, process conditions and via geometry. Confirm coverage using cross-sectional SEM at defined positions, suitable elemental analysis where needed, electrical continuity checks and downstream plating trials.
Recommended process-validation approach
- Characterize the via: Record entrance and exit diameters, minimum opening, glass thickness, taper and sidewall profile.
- Define the film stack: Specify adhesion, barrier and seed-layer materials and target thicknesses.
- Set acceptance criteria: Define measurement locations and minimum coverage or continuity requirements before testing.
- Run representative trials: Compare selected source geometries and process conditions using the same substrate and via design.
- Inspect cross-sections: Use SEM at defined via locations; use EDS where elemental identification is useful. Recognize that EDS is not automatically a precise thin-film thickness measurement.
- Verify electrical and plating performance: Check continuity and, where appropriate, conduct downstream plating trials and inspect plated cross-sections for defects.
5. Which Architecture Makes Sense?
Use the following as a first-pass selection guide, then confirm the choice with process trials, capacity calculations and a layout review.
A practical decision sequence
- Prove process feasibility: Define via geometry, metal stack, coverage and continuity criteria; evaluate representative samples.
- Define production demand: Specify substrate size, product mix, required good output, shifts and changeover frequency.
- Compare handling and integration: Review carrier or robot strategy, process modules, vacuum interfaces, maintenance and factory layout.
- Calculate ownership economics: Compare CAPEX, OPEX, yield, availability and cost per good substrate using the same assumptions.
- Freeze architecture: Select Cluster, Inline or a customized multi-chamber arrangement only after the preceding requirements are understood.
6. SIMVACO: Advanced PVD Equipment Engineering for Glass Substrate Applications
SIMVACO is an industrial vacuum coating equipment manufacturer with engineering experience in magnetron sputtering, multi-arc PVD, evaporation, PECVD and complex multi-chamber vacuum system integration.
SIMVACO's equipment-engineering scope includes large-area glass substrate handling, chamber and process-zone arrangement, vacuum transfer or isolation, deposition-source integration, thermal management and automation. These capabilities support equipment architecture evaluation; they should not be confused with proof of a specific TGV sidewall-coverage result.
Depending on substrate format, process sequence and production objectives, a project may warrant a Cluster system, an Inline platform or a customized multi-chamber solution. The recommendation should follow the validated process route and production economics rather than a predetermined preference for one architecture.
Information for an initial project review
- Glass dimensions, thickness, material and handling constraints.
- Via entrance and exit diameters, glass thickness, taper and cross-sectional profile.
- Adhesion, barrier and seed-layer materials, with target thicknesses.
- Coverage and continuity requirements at the entrance, sidewall and via exit.
- Downstream plating process and electrical acceptance criteria.
- Required good substrates per hour or shift, product mix and available floor space.
7. Frequently Asked Questions
Is Inline PVD always faster than Cluster PVD?
No. Inline can support a high-output, stable production flow, but actual throughput depends on the bottleneck process zone, takt time, availability and yield. A well-utilized Cluster system can be competitive for certain products and process cycles.
Can Cluster PVD process large glass panels?
It can be engineered for different substrate formats, subject to chamber dimensions, transfer-module design, robot reach and substrate support. Feasibility and cost should be assessed for the actual panel size and thickness.
Does Inline PVD provide better TGV sidewall coverage?
Not inherently. Sidewall coverage depends primarily on via geometry, source arrangement, particle angular distribution, substrate orientation and process conditions. Inline or Cluster describes the equipment architecture, not the deposition physics.
Is an aspect ratio of 5:1 the limit for conventional PVD?
No. It is a useful caution point for closer evaluation, not a universal limit. Feasibility depends on via diameter, depth, taper, source geometry and the required continuity criteria. Representative cross-sectional testing is necessary.
Can long-throw sputtering or higher pressure solve high-aspect-ratio coverage?
Neither is a universal solution. Long-throw geometry and higher working pressure change the arriving particle distribution in different ways, with trade-offs in deposition rate, uniformity and film properties. Their effect must be tested against the actual via geometry and seed-layer continuity requirement.
Does ionized PVD guarantee conformal coverage in TGVs?
No. Ionization can change particle transport and energy, but the result depends on plasma distribution, ion energy, electric-field conditions, via geometry and the electrical behavior of the glass substrate. Process validation is required.
Does double-sided deposition guarantee continuous seed coverage?
No. It may improve access from both openings, but the deposition profiles must connect to form a continuous conductive path. Alignment, via profile, source geometry and process conditions must be validated.
What is the best architecture for a cost-sensitive project?
There is no universal answer. Compare total investment, utilities, labor, consumables, maintenance, yield, utilization, floor space and cost per good substrate using the same production assumptions.
Technical note: This article provides general equipment-selection guidance. Specific process limits, coverage performance, throughput and cost must be confirmed for the actual application and equipment configuration. Numerical throughput and cost examples are hypothetical and are provided only to explain the calculation method.