Borosilicate vs Fused Silica Glass for TGV Substrates

Choosing the glass for a through-glass-via (TGV) substrate isn’t just a materials-sourcing decision — it sets the ceiling on electrical performance, thermal reliability, and cost for everything built on top of it. Borosilicate glass and fused silica are the two most common candidates, and they sit at nearly opposite ends of the trade-off space: one is more affordable and easier to process at scale, the other delivers markedly better electrical and thermal performance at a higher cost. This guide compares the two so you can match the substrate to the application rather than defaulting to whichever one a given supplier already stocks.

Key Takeaways

• Borosilicate glass and fused silica differ fundamentally in structure: borosilicate is a phase-separated silica-boron oxide composition, while fused silica is a single-phase, ultra-pure form of amorphous SiO₂.
• Fused silica has a coefficient of thermal expansion roughly six times lower than typical borosilicate glass, and it holds that value more consistently across temperature.
• Fused silica also has meaningfully lower dielectric loss than borosilicate, making it the substrate of choice for the highest-frequency RF and mmWave applications.
• Borosilicate’s phase-separated internal structure contributes to both its higher dielectric loss and, in some cases, stress concentration around TGVs during thermal cycling.
• Borosilicate’s CTE can be tuned across a range of compositions, which is useful for matching it more closely to silicon or to organic substrate materials depending on the application.
• Fused silica costs more to produce, since it requires higher-purity raw material fused at very high temperatures, while borosilicate benefits from a more established, lower-cost supply chain.
• The lowest-CTE glass isn’t automatically the most reliable choice — some studies show that glasses with a higher CTE actually suppress redistribution-layer edge cracking better than ultra-low-CTE options like fused silica, because the mismatch that matters most is between the glass and the build-up layers on top of it, not just the glass’s absolute CTE.
• Even with a well-matched bulk glass CTE, the copper inside a TGV has its own CTE mismatch with the surrounding glass, so via-level thermal stress remains a design consideration regardless of which substrate glass is chosen.

What Is Borosilicate Glass?

Composition and Key Properties

Borosilicate glass is made by adding boron oxide (B₂O₃) to a silica (SiO₂) base, typically resulting in a composition around 80% silica. The addition of boron reduces the glass’s thermal expansion and improves chemical resistance compared to ordinary soda-lime glass, while remaining considerably easier and less expensive to produce than fused silica. Structurally, borosilicate glass tends to be phase-separated at a microscopic level, with boron-rich and silica-rich regions coexisting within the same material — a structural feature that shapes both its dielectric behavior and, in some TGV applications, its response to thermal stress.

Borosilicate’s CTE is commonly cited around 3.3 ppm/°C for standard compositions (such as Borofloat- or Pyrex-type glass), though it can be adjusted across a broader range through composition changes, which is one reason it has remained popular for applications that need a CTE closer to silicon’s (roughly 2.6 ppm/°C) or, alternatively, closer to organic PCB materials on the higher end.

Common Uses

Borosilicate is the default substrate for a large share of current TGV and glass-core packaging work, including 5G mmWave modules, automotive radar packages, and general-purpose glass interposers. Its combination of moderate cost, reasonably low dielectric loss, and an established manufacturing base — including laser drilling and etch recipes that have been refined over years of use — makes it the practical starting point for most new TGV programs.

What Is Fused Silica?

Composition and Key Properties

Fused silica is produced by fusing extremely high-purity silica sand at temperatures above 1,800°C, yielding a non-crystalline material that is over 99.9% pure SiO₂. Unlike borosilicate’s phase-separated structure, fused silica is structurally homogeneous — a single, continuous amorphous silica network without a secondary boron oxide phase. That homogeneity is the root cause of most of its performance advantages: it’s what gives fused silica its very low dielectric loss and its exceptionally low, stable coefficient of thermal expansion, typically around 0.5-0.55 ppm/°C.

Fused silica is also highly resistant to chemical attack — it withstands water, salt solutions, and most acids, with hydrofluoric and phosphoric acid being the notable exceptions (which is also why HF-based wet etching works as a via-formation and photosensitive-glass-patterning chemistry).

Common Uses

Fused silica is the substrate of choice wherever electrical or thermal performance requirements exceed what borosilicate can reliably deliver: applications above roughly 60 GHz (E-band, D-band, and emerging THz work), precision RF calibration standards where dielectric constant stability matters more than cost, and high-power laser or optical applications where fused silica’s established use as an optical material carries over naturally into TGV-based photonic packaging.

Key Differences

Electrical and Dielectric Performance

This is where the two materials diverge most clearly. Fused silica’s homogeneous structure gives it a dielectric loss tangent below 0.0002 at 1 GHz, staying under 0.0005 even above 100 GHz — a meaningfully lower-loss substrate than borosilicate’s roughly 0.004 at the same frequency. Borosilicate’s higher loss is generally attributed to its phase-separated structure: the interfaces between boron-rich and silica-rich regions introduce loss mechanisms that a single-phase material like fused silica simply doesn’t have. Fused silica’s dielectric constant is also more stable — its Dk stays close to 3.78 across a range of temperature and humidity conditions, while borosilicate’s Dk, though still reasonably stable compared to non-glass substrates, shows somewhat more panel-to-panel and condition-to-condition variation.

For applications running well into mmWave or higher frequencies, this gap is large enough to be the deciding factor on its own — it’s the reason fused silica remains the substrate of choice for the highest-frequency RF work even though it costs more.

Thermal Expansion and Reliability

Fused silica’s CTE, at around 0.5-0.55 ppm/°C, is roughly six times lower than typical borosilicate’s 3.3 ppm/°C, and fused silica holds that value with less variation across temperature since it doesn’t go through the phase transitions that affect some other glass types. On paper, that makes fused silica the more thermally stable substrate. In practice, TGV reliability is more nuanced than “lowest CTE wins.”

Two complications are worth knowing. First, borosilicate’s CTE can be engineered across a range of compositions, which lets designers deliberately choose a value that’s close to silicon’s CTE (reducing die-to-substrate stress) or closer to an organic build-up material’s CTE (reducing substrate-to-package stress) — a flexibility that fused silica’s near-fixed, very low CTE doesn’t offer. Second, and less intuitively, some reliability studies on redistribution-layer (RDL) cracking have found that higher-CTE glasses actually suppress edge-crack propagation better than lower-CTE options including standard borosilicate and fused silica, because the failure-relevant mismatch is often between the glass and the copper/dielectric build-up layers on top of it, not the glass’s CTE in isolation. And regardless of which glass is used, the copper that fills a TGV has its own CTE that’s very different from either glass, so some via-level thermal stress is unavoidable and has to be managed through via design and metallization process choices rather than substrate selection alone.

Mechanical Strength and Processing

Both materials are brittle, glassy solids with no meaningful plastic deformation range, but they don’t behave identically under stress. Borosilicate and aluminosilicate glasses, which tend to have a higher baseline defect density, have shown larger strength improvements from surface-strengthening treatments in comparative studies, while fused silica — already closer to its low-defect theoretical strength — showed more limited gains from the same treatments. In practice, this means borosilicate’s mechanical behavior is somewhat more tunable through surface processing, while fused silica’s strength is more a function of how well surface defects are avoided in the first place during handling and processing.

On the manufacturing side, borosilicate benefits from a longer track record in TGV specifically — it was the material most early TGV development work was done on, so laser drilling, LIDE, and etch process recipes are generally more mature and widely available for it. Fused silica is fully compatible with the same general process families (laser modification, wet etching, sputtering, and electroplating), but the extra purity and process control needed to work with it, combined with its higher raw material cost, generally makes fused silica processing more expensive per panel.

Cost and Availability

Borosilicate is the more cost-effective option by a meaningful margin, largely because its raw materials and melting process are less demanding than fused silica’s, and because its supply chain for TGV-grade panels is more established. Fused silica’s higher purity requirement (fusing silica sand at temperatures above 1,800°C) and lower defect tolerance push its cost up substantially, which is why it tends to be reserved for applications where its electrical or thermal performance is genuinely required rather than used as a general-purpose default.

Which Should You Choose?

For most current TGV programs — 5G sub-6GHz and mmWave modules, automotive radar, general glass interposers, and MEMS packaging — borosilicate glass offers the best balance of cost, established process support, and adequate electrical performance. Its tunable CTE is also a genuine advantage when the goal is matching a specific silicon die or organic substrate rather than minimizing CTE in absolute terms.

Fused silica earns its higher cost in applications where dielectric loss above 60 GHz, extremely stable dielectric constant, or best-in-class thermal expansion stability are hard requirements rather than nice-to-haves — precision RF references, some photonic packaging, and next-generation mmWave or THz systems where borosilicate’s loss tangent would compromise the signal budget.

Neither material is a universal answer, and the choice shouldn’t be reduced to “fused silica is always better but costs more.” Borosilicate’s tunable CTE and mature process ecosystem make it the more practical substrate for a large share of real applications, while fused silica’s performance ceiling matters mainly when the application’s frequency or precision requirements actually reach it. The right question isn’t which glass has better numbers on a datasheet — it’s which glass’s specific advantages line up with what the application actually needs.

FAQ

What is the main difference between borosilicate glass and fused silica?

Borosilicate glass is a silica-boron oxide composition with a phase-separated internal structure, while fused silica is a single-phase, ultra-pure form of amorphous silica. That structural difference is the root cause of fused silica’s lower dielectric loss and lower, more stable coefficient of thermal expansion.

Which glass has lower dielectric loss?

Fused silica, by a substantial margin — its loss tangent stays below 0.0002 at 1 GHz and under 0.0005 even above 100 GHz, compared to roughly 0.004 for typical borosilicate at 1 GHz.

Which glass is cheaper for TGV substrates?

Borosilicate glass is generally more cost-effective, due to less demanding raw material requirements and a more established manufacturing supply chain compared to fused silica’s higher-purity, higher-temperature production process.

Is a lower CTE always better for TGV reliability?

Not necessarily. While a low, stable CTE generally helps, some studies have found that higher-CTE glasses can actually suppress redistribution-layer edge cracking better than very-low-CTE options, because the relevant mismatch is often between the glass and the build-up layers on top of it rather than the glass’s CTE alone. Via-level stress from the copper fill’s own CTE also remains regardless of which glass is chosen.

Why is borosilicate’s CTE described as “tunable”?

Borosilicate glass compositions can be adjusted to shift the CTE across a range, which lets designers choose a value closer to silicon’s CTE for better die-to-substrate matching, or closer to an organic substrate’s CTE for better package-level matching. Fused silica’s CTE is close to fixed by comparison.

When should fused silica be used instead of borosilicate?

Fused silica is the better choice when the application genuinely needs performance above what borosilicate can deliver — typically above 60 GHz, in precision RF reference applications where dielectric constant stability is critical, or in photonic packaging where fused silica’s established optical properties are an advantage.

Are borosilicate and fused silica processed the same way for TGV formation?

Both are compatible with the same general process families — laser drilling, laser-induced deep etching, wet etching, sputtering, and electroplating — but borosilicate has a longer track record in TGV specifically, so process recipes tend to be more mature and widely available for it. Fused silica requires tighter process control and costs more to process per panel.

Does glass choice eliminate thermal stress inside a TGV?

No. Even with a well-matched glass CTE, the copper that fills the via has a CTE very different from either glass, so some thermal stress concentrates at the via-glass interface regardless of substrate choice. This is why TGV via design and metallization are treated as their own reliability discipline rather than something solved purely by substrate selection.