MLCC Capacitors: DC Bias, Cracking, and Sourcing During Shortages

Multilayer ceramic capacitors are among the most numerous components on any modern PCB, yet they carry two under-discussed failure risks that catch even experienced engineers off guard: the DC bias effect, which silently erodes effective capacitance under real operating conditions, and mechanical cracking, which can produce intermittent failures that are notoriously difficult to diagnose. Add an ongoing structural tightness in MLCC supply driven by AI server buildouts and electric vehicle production, and the stakes around correct selection, design margin, and procurement strategy are higher than they have been in years.

This article covers all three dimensions in practical terms. The goal is a reliable reference — one that explains what actually happens inside an MLCC under voltage and mechanical stress, what the field consequences look like, and how to think about sourcing when specific specifications become constrained. Whether you are reviewing a new design, evaluating a substitute part, or managing a BOM through a shortage cycle, the same fundamentals apply.

What Is an MLCC and Why Does It Dominate Electronics Design?

A multilayer ceramic capacitor (MLCC) consists of hundreds of thin alternating layers of ceramic dielectric material and metal electrodes, fired together into a monolithic chip. The layered construction allows high capacitance values in very small footprints, which is why MLCCs have displaced most other capacitor technologies in surface-mount designs. They offer low equivalent series resistance (ESR), wide operating temperature ranges, long service life under normal conditions, and low unit cost at volume. A single modern smartphone may contain over 1,000 of them.

The two broad dielectric classes — Class 1 and Class 2 — behave quite differently, and that distinction is at the root of most MLCC-related design problems. Class 1 materials (principally C0G, also written NP0) are based on non-ferroelectric ceramics and deliver excellent capacitance stability across temperature and voltage. Their drawback is limited capacitance density, which caps practical values at a few hundred nanofarads in standard surface-mount packages. Class 2 materials — X7R, X5R, and Y5V being the most common — use barium titanate (BaTiO₃) ferroelectrics to achieve much higher dielectric constants and therefore much higher capacitance in the same physical volume. The trade-off is that those ferroelectric properties introduce sensitivity to temperature, applied voltage, and time. Understanding that trade-off is the foundation of working with MLCCs reliably.

DC Bias: Why Your 10 µF Capacitor May Not Be 10 µF

The DC bias effect is one of the most frequently overlooked phenomena in PCB design. MLCC capacitors marked with a nominal value are tested and specified at zero volts — not at the operating voltage in your circuit. When a DC voltage is applied across a Class 2 ceramic capacitor, the effective capacitance decreases, sometimes dramatically. A component labeled 10 µF may deliver only 3 to 5 µF under real operating conditions, depending on the dielectric, package size, voltage rating, and the fraction of that rating being applied.

The Physics Behind the Drop

Class 2 dielectrics are based on ferroelectric barium titanate. At the crystal level, BaTiO₃ contains domains of aligned electric dipoles that are responsible for the material’s high permittivity. When a DC voltage is applied, those domains begin to align along the electric field. This alignment reduces the material’s ability to respond to additional charge, effectively lowering its permittivity and therefore its capacitance. The effect is not a defect — it is an intrinsic physical property of ferroelectric materials. What makes it problematic in practice is that nominal capacitance values in datasheets are measured at near-zero voltage, so they do not reflect in-circuit behavior.

In laboratory demonstrations, a 1 µF X7R capacitor with a 25 V rating has shown capacitance losses of more than 40% when tested at its full rated voltage. At typical operating voltages that represent 50 to 80 percent of the rating, losses commonly fall in the 30 to 60 percent range, varying by manufacturer, dielectric formulation, and package size. The result in a power supply design can be a filter that no longer sits at the frequency the math predicted, or a decoupling reservoir that is too small to hold the rail stable during load transients.

Package Size Makes It Worse

The DC bias effect scales inversely with package size. Smaller packages accommodate the same nominal capacitance by using thinner dielectric layers and more of them. Thinner dielectric layers mean a stronger electric field per layer at the same applied voltage, which more aggressively aligns the ferroelectric domains and produces a steeper capacitance drop. A 22 µF X5R in a 0603 package operating at half its rated voltage may retain only 40 percent of its nominal value, while the same nominal specification in a 1206 package may retain around 80 percent under identical bias conditions. When a design is miniaturized by substituting a larger package with a smaller one, the effective capacitance at operating voltage can change substantially even if the nominal value and voltage rating on the reel are identical.

Voltage Derating as a Practical Counter-Measure

The standard engineering response to the DC bias effect is voltage derating: selecting an MLCC with a rated voltage meaningfully higher than the operating voltage. Using a capacitor at a lower fraction of its rated voltage means the electric field per dielectric layer is weaker, the domain alignment is less severe, and effective capacitance is closer to the nominal value. A common industry practice is to operate Class 2 MLCCs at no more than 50 percent of their rated voltage. On a 3.3 V rail, that means using parts rated at 6.3 V or higher; on a 5 V rail, 10 V or higher; on a 12 V rail, 25 V or higher. For high-reliability applications — automotive, medical, aerospace — programs often derate further still, specifying operation at well under half the rated voltage.

Derating has a second benefit beyond DC bias: it reduces dielectric stress and slows long-term degradation. Operating a capacitor well below its maximum rated voltage significantly extends its effective service life. The practical cost is a slightly larger or more expensive component, which is a straightforward trade-off in any design where reliability matters more than minimum BOM cost.

Aging: The Third Compounding Factor

DC bias does not operate in isolation. Class 2 ceramics also age — they lose capacitance gradually over time even at room temperature and at zero applied voltage. This occurs because the ferroelectric grains slowly relax toward more stable crystal domain configurations, reducing the material’s permittivity. The standard aging model describes roughly 2 to 5 percent capacitance loss per decade of time in hours, depending on the dielectric: after 10 hours, after 100 hours, after 1,000 hours, and so on. Under sustained DC bias, the rate of aging accelerates. Measurements under bias have shown aging running several times faster than the zero-voltage model would suggest in some parts. Combined with the initial DC bias drop and any temperature-related variation, the effective capacitance in a fielded product can be substantially less than the value printed on the component.

The practical implication for design is to check DC bias curves from the specific manufacturer’s datasheet — not just distributor catalog data — at the actual operating voltage. Many manufacturers offer online simulation tools that return the expected capacitance under specified voltage and temperature conditions. If that data is not available for a candidate part, that is reason enough to question it before approving it for production.

MLCC Cracking: The Failure Mode That Hides in Plain Sight

Cracking is the other major MLCC failure mode, and it is responsible for a significant share of field returns in electronics manufacturing. Unlike DC bias degradation, which produces a gradual, predictable decline, cracking tends to produce intermittent or latent failures that are difficult to reproduce on the bench and expensive to diagnose. The ceramic body of an MLCC is strong in compression but brittle in tension — it handles being squeezed well but fractures under bending stress. Once a crack forms, the consequences range from an open circuit to a leakage path to a board-level short circuit, depending on how the fracture runs through the electrode stack.

Flex Cracks from PCB Bending

Flex cracking is the most common mechanical failure mode for MLCCs. It occurs when the PCB bends and the solder joints transmit tensile stress into the ceramic body. Because the MLCC is rigidly bonded to the board at both terminations, any flex in the board creates a moment that the brittle ceramic cannot absorb without fracturing. Flex cracks typically start at the solder fillet and propagate laterally into the capacitor body. The fracture may not be visible at initial inspection, and the component may pass functional testing immediately after assembly — the crack only becomes problematic when thermal cycling, vibration, or further handling causes the crack faces to move relative to each other.

Common triggers include PCB depanelization (particularly V-score breaking, which induces a sharp bending event), screw-mounting deformation when enclosures are assembled, manual handling of bare boards, and in-circuit testing with poorly fixtured probes. Components placed near board edges, near depanelization score lines, or near mounting holes are at elevated risk because those areas experience the greatest deflection during mechanical events.

Thermal Cracks from Reflow and CTE Mismatch

Thermal cracking arises from the mismatch in coefficients of thermal expansion (CTE) between the ceramic capacitor body and the surrounding materials — the solder, the PCB laminate, and any substrate it is mounted on. During reflow soldering, the assembly heats to peak temperatures above 200°C and then cools. If the cooling profile is too rapid, or if the MLCC is placed on a substrate with a significantly different CTE, differential contraction generates enough stress to fracture the ceramic. Cracks from thermal shock tend to originate at the surface and propagate inward, running toward the termination ends. They can also be introduced by hand soldering, where localized heating from an iron tip creates steep thermal gradients across a component a few millimeters long.

Larger case sizes (1206, 1210, and larger) are more susceptible to thermal cracking than small ones because the thermal gradient is greater across a longer body. For large-case, high-capacitance MLCCs — exactly the specifications that are also under supply pressure today — thermal process control during assembly is particularly important.

What Happens When an MLCC Cracks

The electrical consequence of a crack depends on its path through the component’s internal structure. A crack that runs between electrode layers typically produces an open circuit or increased capacitance loss — the capacitor simply stops working or works at reduced value. A crack that intersects multiple electrode layers and creates conductive paths between them produces a short circuit or leakage current. The leakage scenario is the more dangerous one in powered circuits: it can cause localized heating, accelerate dielectric breakdown, and in high-voltage applications, lead to thermal runaway. Cracked MLCCs frequently cause intermittent failures rather than immediate hard faults, which makes them particularly costly to diagnose and rework in field returns.

Prevention Strategies

Several design and process measures reduce cracking risk substantially. On the layout side, avoiding placement of large-case MLCCs near depanelization score lines, board edges, and screw holes removes the highest-risk locations. Routing-based depanelization induces significantly less board flex than V-score breaking, and manufacturers’ application notes consistently report far lower flex-crack rates when routing replaces shearing. Fixtures that support PCB edges during depanelization prevent center deflection from transmitting stress to components.

For the component itself, flexible-termination MLCCs — in which the standard rigid metallic termination is replaced with one that incorporates a conductive epoxy layer — absorb a portion of the mechanical stress before it reaches the ceramic body. They are more expensive than standard parts, but they are a well-established solution for assemblies that cannot avoid flex events, such as boards that are mounted by screwing into enclosures or assemblies that will experience vibration in service. Specifying a controlled reflow profile with a gradual cooling ramp reduces thermally induced stress during the soldering process. These measures are not mutually exclusive — the highest-reliability designs apply several of them together.

Dielectric Selection: C0G, X7R, X5R, and Y5V Compared

Dielectric selection sits at the intersection of performance requirements, sourcing flexibility, and cost. The EIA temperature characteristic code gives the most important information in three characters: the lower temperature limit, the upper temperature limit, and the maximum capacitance change across that range. X7R operates from -55°C to +125°C with a capacitance change tolerance of ±15%, making it the standard choice for power supply decoupling and bypass in most general-purpose, industrial, and automotive-rated designs. X5R covers -55°C to +85°C at the same ±15% tolerance and is well-suited to consumer electronics where the thermal envelope is reliably controlled. Substituting X5R for X7R is only appropriate when the maximum operating temperature including self-heating from adjacent components has been thoroughly validated to stay below 85°C. For AEC-Q200 qualified automotive designs, X7R is typically the minimum acceptable specification.

C0G (NP0) is the right choice when capacitance stability matters more than density. It shows virtually no DC bias effect, no meaningful aging, and a temperature coefficient measured in parts per million per degree Celsius rather than percent. Timing circuits, oscillator load capacitors, RF filters, and precision analog signal paths all benefit from C0G. Y5V, at the other end of the spectrum, can swing from +22% to -82% of nominal capacitance across its rated temperature range and degrades severely under DC bias. It has a place in non-critical, room-temperature bypass applications where the exact capacitance value does not affect circuit function, but it should never appear in timing, filtering, or any node where value stability influences performance.

Sourcing MLCCs During Shortages

What Is Driving the Current Tightness

The MLCC market is experiencing a structural shortage in certain specifications, driven by converging demand from AI server infrastructure, electric vehicles, and advanced industrial electronics. A single AI server rack can require hundreds of thousands of MLCCs, and forecasts suggest MLCC shipments into AI servers will continue growing rapidly through the end of the decade. Electric vehicles have also emerged as a structural driver: a modern EV requires substantially more high-voltage, high-capacitance MLCCs than a conventional automobile. These two demand vectors are pulling on the same manufacturing capacity that has historically served consumer electronics, creating uneven lead times across the MLCC product range.

The shortage is not uniform. Standard consumer-grade parts — 100 nF, 10 V, X7R, 0402, for example — remain relatively available. The tightest pressure is on high-capacitance (10 µF and above), high-voltage (25 V and above) X7R parts in small packages (0402 to 0603), and on AEC-Q200 qualified automotive specifications. For affected specifications, lead times from tier-one manufacturers are quoted in months rather than weeks, and the only reliable figure is a live quote for the specific part number. Treating MLCC risk as an isolated BOM line item can be misleading — a build under pressure from MLCC lead times is often also under pressure from adjacent passives and power semiconductors on the same board.

Practical Sourcing Strategies

The most effective response to structural component tightness is moving away from just-in-time purchasing toward a more proactive supply chain posture. For at-risk MLCC specifications, early identification during BOM creation — rather than at the final design review stage — allows more options. A part flagged as long-lead at schematic capture can often be addressed through an approved alternate source, a design change to a more available specification, or a pre-negotiated inventory buffer. Flagged at the purchasing stage during a shortage peak, the same part leaves procurement with far fewer options.

Cross-referencing to alternate brands requires rigor. Matching package size (including thickness, not just footprint), dielectric code, voltage rating, capacitance value, and tolerance is the starting checklist. Verifying the DC bias curve at the actual operating voltage from the alternate manufacturer’s datasheet is the step that separates a validated substitute from one that merely matches the catalog spec. Two 10 µF X7R capacitors from different manufacturers with identical nominal specifications can have meaningfully different effective capacitance at the same operating voltage, particularly when comparing parts optimized for high capacitance density against parts designed with a larger voltage margin. Whichever candidate passes, final approval of the substitute rests with the customer’s engineering team.

Sourcing runs the length of a product’s life — prototype builds, volume production, and the years after a part is withdrawn while a fielded design still has to be supported — and MLCC risk is easiest to manage when availability data reaches the design team early. WIN SOURCE, a global independent distributor of electronic components, stocks ceramic capacitors across current-production and hard-to-find specifications and supports shortage sourcing, lifecycle management and alternative solutions, engaged as a long-term procurement partner rather than an emergency measure. Its WinLink BOM tool and WinConnect API let engineering and procurement teams pull availability data into the design and production process before a shortage becomes a production stop.

Substitution Checklist

When evaluating an MLCC substitute during a shortage, the following parameters need to be verified in sequence — not just the first few:

• Package dimensions: Match footprint length and width, and verify the height (thickness). An MLCC that is taller than the original can interfere with clearance to adjacent components or mechanical enclosures.
• Dielectric class and temperature rating: Confirm the EIA code (C0G, X7R, X5R, etc.) and verify the temperature range is adequate for the application’s worst-case thermal environment, including self-heating from nearby components.
• Voltage rating with derating margin: Confirm the substitute meets the same derating practice as the original. Do not accept a lower voltage rating even if the nominal operating voltage appears to fit.
• DC bias curve at operating voltage: Obtain the substitute manufacturer’s datasheet and check effective capacitance at the actual operating voltage. If the DC bias curve is absent, request it or reject the part.
• ESR and ripple current rating: Relevant for power supply and filter applications. ESR affects power dissipation and ripple performance; ripple current rating sets the safe operating limit in switching power applications.
• Aging rate: Expressed in the datasheet as percent per decade hour. Varies by manufacturer and formulation for the same nominal dielectric class.
• Source traceability: Confirm the substitute comes with documented traceability and counterfeit risk controls — the questions are the same whoever is quoting.

For excess inventory situations or when BOM optimization is part of a broader cost control initiative, the same checklist applies in reverse — validating that a value-engineered substitute meets all electrical and reliability requirements before it reaches production. The engineering decision and the procurement decision need to happen in the same workflow, not in separate silos.

Key Takeaways

MLCC capacitors are reliable, cost-effective, and versatile — but they carry real performance constraints that only become visible when you look past the nominal specifications. The DC bias effect means that the capacitance value on the reel is a measurement taken at zero volts, not at your operating condition. Package size, dielectric choice, voltage rating margin, and manufacturer formulation all influence how much of that nominal value survives in the actual circuit. Aging compounds the effect over the product’s service life. Designing around DC bias requires checking the manufacturer’s simulation tools or datasheet curves at the actual operating voltage, and selecting a voltage rating that provides genuine headroom rather than the narrowest margin that technically fits.

Cracking is a separate failure risk that demands attention during layout, assembly process definition, and component selection — not just during incoming inspection. Flex cracks and thermal cracks are both common, both preventable with the right process controls, and both capable of producing intermittent failures that are expensive to diagnose in the field. Routing-based depanelization, careful placement away from mechanical stress points, and flexible-termination components in high-risk assemblies are the main levers available.

On the supply side, the current structural tightness in high-capacitance and automotive-grade MLCCs is not a temporary blip — it reflects a genuine demand shift driven by AI infrastructure and electrified transportation that will continue to compete for passive component capacity. Early BOM risk identification, cross-referencing with validated alternates, and working with supply chain partners who have proactive shortage and lifecycle management capabilities are the practical responses to that reality.

Frequently Asked Questions

Why does a 10 µF MLCC measure less than 10 µF in the circuit?

Because Class 2 dielectrics such as X7R and X5R lose capacitance under DC voltage. The nominal value is measured at zero volts; at the rail voltage the same part can deliver a fraction of its rating, with small case sizes affected most. Check the manufacturer’s DC bias curve at your actual operating voltage before approving the part.

How much should I derate an MLCC’s voltage rating?

A common practice is to operate Class 2 MLCCs at no more than half their rated voltage — a 6.3 V or higher part on a 3.3 V rail, 10 V or higher on 5 V, 25 V or higher on 12 V. High-reliability programs often derate further. Derating reduces the DC bias loss and slows long-term dielectric aging.

What causes MLCCs to crack?

Mostly board flex — depanelization, screw-mounting, handling and test fixtures bend the PCB and the solder joints transmit tensile stress into the brittle ceramic — and thermal shock from reflow or hand soldering. Cracks often pass initial test and fail intermittently later. Routing-based depanelization, placement away from edges and mounting holes, controlled reflow cooling and flexible-termination parts are the main preventions.

Can I substitute X5R for X7R?

Only after confirming that the maximum operating temperature, including self-heating from neighboring components, stays below 85°C. X7R is rated to 125°C and X5R to 85°C at the same ±15% tolerance; for AEC-Q200 automotive designs X7R is usually the minimum. Any substitution also needs its DC bias curve checked and your engineering team’s approval.

Which MLCC specifications are hardest to source?

High-capacitance (10 µF and above), higher-voltage X7R parts in small packages, and AEC-Q200 automotive grades have been the tightest, because AI server and EV demand competes with consumer electronics for the same production capacity. Standard values such as 100 nF / 10 V / 0402 are generally available. Get current lead times per part number from the manufacturer or distributor.

© 2026 Win Source Electronics. All rights reserved. This content is protected by copyright and may not be reproduced, distributed, transmitted, cached, or otherwise used without the prior written permission of Win Source Electronics.