How to Choose a Glass Substrate for Inertial Sensor Applications

To choose a glass substrate for an inertial sensor, I recommend starting with the bonding process, thermal expansion match, electrical requirements, surface quality, thickness, and wafer-level manufacturing constraints. For many MEMS accelerometer and gyroscope designs, borosilicate glass is a practical starting point because it can provide electrical insulation, optical transparency, and a coefficient of thermal expansion that may be compatible with silicon. However, fused silica, aluminosilicate glass, or another specialty glass may be more suitable when low thermal expansion, high temperature stability, or optical performance is the priority.

For more information, please visit our website.

At Glass Circuit, I evaluate the substrate as part of the complete sensor stack rather than as an isolated sheet. The correct specification depends on the silicon structure, bonding method, electrode layout, operating temperature, package design, and required production volume. The following process helps engineering and purchasing teams define a technically appropriate and commercially realistic glass substrate specification.

1. Define the Inertial Sensor’s Requirements

Before comparing glass types, I first document the sensor architecture and manufacturing flow. A glass substrate may function as a cap wafer, a bonding carrier, an electrical isolation layer, an optical interface, or a structural component in a MEMS package. Each role imposes different requirements for thickness, flatness, dielectric behavior, thermal stability, and surface finish.

I also recommend separating confirmed requirements from early-stage assumptions. For example, a prototype may accept a wider thickness tolerance, while a wafer-level production process may require tighter total thickness variation and edge quality. This distinction prevents the purchasing team from over-specifying the material before the process window is understood.

Questions to Confirm First

• Is the glass used for anodic bonding, fusion bonding, adhesive bonding, or mechanical support?
• Will the glass contact silicon, silicon oxide, metal electrodes, polymers, or a combination of these materials?
• What are the minimum and maximum process temperatures?
• Does the design require electrical isolation, optical transmission, or both?
• What wafer diameter, thickness, edge profile, and quantity are required?
• Will the substrate be processed as a full wafer, diced panel, or custom-cut component?

2. Select the Glass Material Family

The glass composition affects thermal expansion, chemical resistance, electrical behavior, strength, optical transmission, and bonding compatibility. I normally compare borosilicate glass, fused silica, aluminosilicate glass, and soda-lime glass before selecting a final grade. No single material is optimal for every inertial sensor application.

Borosilicate Glass

Borosilicate glass is often considered for silicon MEMS bonding because its thermal expansion can be relatively close to silicon compared with higher-expansion soda-lime compositions. SCHOTT lists a coefficient of thermal expansion of approximately 3.25 × 10^-6/K for BOROFLOAT^® 33 over a defined temperature range, while silicon is commonly represented at approximately 2.6 × 10^-6/K near room temperature. The actual stress after bonding still depends on the temperature profile, substrate dimensions, glass grade, and device geometry.

I would consider borosilicate glass when the design needs electrical insulation, transparent inspection, and a relatively balanced thermal expansion relationship with silicon. It is also commonly available in wafer and sheet formats, although the available thicknesses, diameters, polishing levels, and edge specifications must be confirmed for each supplier.

Fused Silica

Fused silica offers very low thermal expansion, with representative values near 0.55 × 10^-6/K for some commercial grades. This property can reduce dimensional change during temperature variation, but fused silica may require a different bonding strategy and can have higher material or processing costs than standard borosilicate glass.

I consider fused silica when thermal drift, optical stability, or high-temperature dimensional control is more important than straightforward anodic bonding compatibility. I do not recommend selecting it solely because its expansion coefficient is low; the team must also confirm electrode integration, surface activation, machining, and supplier availability.

Aluminosilicate and Soda-Lime Glass

Aluminosilicate glass may be useful when higher mechanical strength, chemical durability, or specialized thermal performance is required. Soda-lime glass can be attractive for cost-sensitive prototypes and simple insulating structures, but its thermal expansion is generally higher and more composition-dependent than that of many borosilicate grades.

For a precision inertial sensor, I treat soda-lime glass as a candidate that requires additional thermal-stress evaluation rather than as an automatic replacement for borosilicate glass. If the substrate will be bonded to silicon or exposed to repeated temperature cycling, I ask for material data across the relevant temperature range and request a process-level compatibility review.

3. Match the Glass to the Bonding Process

The bonding method is usually the most important selection factor because it determines the required surface condition, temperature, voltage, cleanliness, and thermal-expansion relationship. For silicon-to-glass anodic bonding, the glass normally needs mobile alkali ions and a sufficiently clean, smooth surface to support the bonding mechanism. For adhesive bonding, the emphasis may shift toward surface energy, chemical compatibility, moisture control, and dimensional stability.

Anodic Bonding

Anodic bonding commonly uses elevated temperature and an applied electric field to join glass and silicon. Depending on the glass grade and equipment, process temperatures may be in the approximate range of 200–400 °C, with applied voltages that may reach several hundred volts or more. These are planning ranges only; I require the customer’s equipment capability and a controlled process evaluation before recommending a final specification.

For anodic bonding, I review alkali-ion mobility, coefficient of thermal expansion, dielectric strength, surface roughness, flatness, particle control, and wafer warpage. I also check whether the glass chemistry could affect the electrical field or produce unwanted ionic movement during operation. A glass that bonds successfully in a laboratory trial may still require optimization for uniformity across a full wafer.

Fusion and Direct Bonding

Fusion or direct bonding places greater emphasis on molecular-scale surface contact, cleanliness, hydrophilicity, and very low surface topography. I would normally request a defined roughness value, such as an Ra target in nanometers, together with a measurable flatness or total thickness variation requirement. The correct limit depends on the bonding equipment, wafer diameter, device pattern, and cleaning process.

Adhesive and Hybrid Bonding

Adhesive bonding can provide more flexibility when the glass composition is not suitable for anodic bonding, but the adhesive introduces its own thermal, mechanical, and outgassing considerations. I evaluate cure temperature, bond-line thickness, moisture sensitivity, long-term creep, and compatibility with vacuum packaging. For hybrid processes, I recommend that the substrate supplier review the complete stack rather than qualifying the glass alone.

Glass Circuit contains other products and information you need, so please check it out.

4. Set the Key Substrate Specifications

A purchase description such as “glass wafer for MEMS” is not sufficiently precise for repeatable sourcing. I recommend specifying at least the glass family or grade, wafer diameter, nominal thickness, thickness tolerance, total thickness variation, flatness or bow, surface roughness, edge profile, surface cleanliness, and inspection method. The specification should also state whether the two sides require different finishes or coatings.

The example values above are not automatic acceptance criteria. I use them to structure an engineering discussion, then convert them into a controlled drawing or purchase specification. If a supplier provides only nominal thickness and visual inspection, I would treat the material as insufficiently defined for a high-volume inertial sensor program.

5. Evaluate Thermal and Mechanical Compatibility

Thermal mismatch can create wafer bow, bond stress, sensor bias, resonance shifts, or package-level drift. I therefore compare the coefficient of thermal expansion for the glass and the mating material across the complete temperature range, rather than comparing a single room-temperature number. I also consider the thickness ratio because a thin glass layer and a thick glass wafer may produce very different mechanical behavior.

For example, a difference of approximately 0.7 × 10^-6/K between two materials may appear small, but it can become significant over a 200 °C process excursion and across a 100 mm wafer. The final impact depends on elastic modulus, bonding temperature, cooling rate, geometry, and constraint conditions. I recommend finite-element analysis or bonded-coupon testing when thermal drift is a critical performance parameter.

Mechanical and Handling Considerations

Glass thickness must be balanced against stiffness, mass, breakage risk, and equipment handling. A substrate that is too thin may be difficult to process or more vulnerable to chipping, while a thicker wafer may increase package volume or alter the sensor’s mechanical response. I ask suppliers to define edge exclusion, chamfer or notch requirements, handling method, and allowable chips or cracks.

6. Check Electrical, Optical, and Chemical Requirements

Many inertial sensor designs use the glass as an electrical isolation layer or as part of a capacitive structure. I recommend confirming dielectric behavior, leakage requirements, electrode compatibility, and any ionic contamination limits before ordering. If the sensor operates at elevated temperature, electrical properties should be reviewed at that temperature rather than only at 23 °C.

Optical transparency may be important for alignment, inspection, or integrated optical readout, but “transparent glass” is not a complete optical specification. I ask for the required wavelength range, transmission target, surface reflection limits, and coating requirements. For example, transmission at 850 nm and 1,550 nm may involve different material and coating decisions.

Chemical compatibility is equally important during cleaning, etching, and surface activation. I review exposure to acids, bases, solvents, plasma, deionized water, and drying temperatures. The glass supplier should confirm whether the proposed composition and surface finish are compatible with the actual process sequence.

7. Compare Suppliers and Commercial Conditions

Once the technical requirements are defined, I compare suppliers on more than price per wafer. I review material traceability, inspection reports, sample availability, packaging, process-change notification, custom polishing capability, and experience with the required wafer format. A low unit price may not be advantageous if the supplier cannot maintain thickness, flatness, or surface quality from lot to lot.

Supplier Evaluation Checklist

• Can the supplier provide the exact glass composition or commercial grade?
• Are thickness, TTV, bow, warp, roughness, and edge quality measured and reported?
• Can the supplier support 100 mm, 150 mm, or another required wafer diameter?
• Are prototype quantities and production quantities handled under the same controlled specification?
• Can the supplier provide material certificates and lot traceability without claiming unverified certifications?
• What is the expected sample lead time, production lead time, MOQ, and replacement policy?
• Can the supplier support dicing, custom cutting, coating, cleaning, or special packaging?

MOQ and lead time vary significantly with glass type, thickness, diameter, polish level, and customization. For planning, I separate prototype purchasing from production sourcing and request written quotations for both stages. I also ask whether the quoted lead time includes inspection, packing, and shipping, because these steps can affect the actual project schedule.

8. Avoid Common Selection Mistakes

The first common mistake is choosing a glass type based only on price or transparency. A low-cost substrate may be unsuitable for the bonding chemistry, thermal cycle, or surface requirements. I recommend qualifying the material against the complete process flow before comparing unit costs.

The second mistake is specifying “polished” without defining roughness, flatness, cleaning level, or measurement method. Different suppliers may interpret the same description differently, creating avoidable variation between lots. I use measurable requirements and request sample inspection data before approving production material.

The third mistake is ignoring the substrate’s role in the sensor’s final mechanical behavior. Glass thickness, stiffness, and residual stress can influence package resonance and temperature response. I therefore involve both the MEMS process engineer and the product or package engineer in the selection review.

9. A Practical Selection Workflow

1. Define the function: Identify whether the glass is a cap, carrier, insulator, optical interface, or structural member.
2. Map the process: List bonding, cleaning, etching, coating, annealing, dicing, and packaging conditions.
3. Shortlist materials: Compare borosilicate, fused silica, aluminosilicate, and other suitable grades.
4. Set measurable specifications: Include diameter, thickness, TTV, flatness, roughness, edge condition, and cleanliness.
5. Test representative samples: Use the same bonding and thermal process planned for production.
6. Review supplier capability: Confirm quality documentation, capacity, MOQ, lead time, and change control.
7. Release a controlled specification: Approve the final drawing or purchase document before volume sourcing.

At Glass Circuit, I can support this workflow by reviewing your wafer format, glass material preference, bonding method, thickness range, surface requirements, and delivery plan. Where the final material choice depends on process data that is not yet available, I recommend a sample evaluation instead of making an absolute compatibility claim.

Key Takeaways

• Choose the glass substrate according to the bonding process and sensor architecture, not by material name alone.
• Use thermal-expansion data in ×10^-6/K across the actual process and operating temperature range.
• Define measurable requirements for thickness, TTV, flatness, roughness, edge quality, and cleanliness.
• Borosilicate glass is often a practical starting point for silicon-to-glass MEMS bonding, while fused silica may suit low-expansion or optical applications.
• Validate the selected glass with representative bonded samples before approving production quantities.
• Compare suppliers on traceability, inspection capability, customization, MOQ, lead time, and technical support—not unit price alone.

Conclusion: Choosing the Right Glass Substrate

The best glass substrate for an inertial sensor is the one that satisfies the complete electrical, mechanical, thermal, optical, bonding, and sourcing requirements of the application. I would normally begin with borosilicate glass for a silicon MEMS bonding study, then compare fused silica or specialty glass when thermal expansion, optical behavior, or operating temperature creates a specific need. The final choice should be confirmed through process-representative testing and a controlled supplier specification.

As the next step, prepare your target wafer diameter, thickness, bonding method, temperature range, surface requirements, annual volume, and inspection expectations. Send these details to Glass Circuit for a preliminary material and sourcing review. We can then help determine whether a standard substrate is appropriate or whether custom thickness, polishing, cutting, coating, cleaning, or packaging support should be included in the quotation.

If you are looking for more details, kindly visit glass substrate for inertial sensor.