When you read '100µF 16V' on a schematic, you are only seeing the headline. The actual behavior, lifespan, and failure mode of the component are dictated entirely by what is inside the capacitor. The dielectric material—the insulator separating the conductive plates—determines the equivalent series resistance (ESR), temperature coefficient, microphonic noise, and DC bias derating. Grabbing the first 100µF capacitor from your bin without checking the dielectric is a fast track to a melted PCB or an oscillating power rail. Here is a bench-level breakdown of capacitor anatomy, how to decode their cryptic markings, and how to substitute them safely when your exact BOM part is out of stock.

The Anatomy: What Actually Lives Inside a Capacitor

At a fundamental level, every capacitor consists of two conductive plates separated by a dielectric. But the physical execution varies wildly depending on the chemistry inside the can or casing.

Aluminum Electrolytic: If you slice open a standard radial can, you will not find two solid plates. Instead, you will find two thin strips of etched aluminum foil separated by a paper spacer, all tightly wound into a cylinder. The dielectric is actually a microscopic layer of aluminum oxide grown on the anode foil. The paper is soaked in a liquid electrolyte that acts as the true second electrode (cathode). This winding creates high inductance (ESL) and the liquid electrolyte gives it a limited lifespan as it slowly evaporates through the rubber bung.

Multilayer Ceramic (MLCC): Inside a tiny 0805 SMD ceramic capacitor is a monolithic block of alternating metal electrode layers and ceramic dielectric (usually barium titanate). Because the layers are stacked in the microscopic range (often less than a micron thick), MLCCs achieve massive capacitance in a tiny footprint with near-zero ESR and ESL. However, the ferroelectric ceramics used for high capacitance (like X5R and X7R) suffer from severe DC bias derating and piezoelectric 'singing' under AC ripple.

Dielectric Showdown: Which Type for Which Job

Selecting the right dielectric is about matching the internal physics to your circuit's stress profile. Use this matrix to choose the right chemistry for the job.

Type / Dielectric Internal Construction Typical Tolerance Tempco / Stability Best Application
C0G / NP0 (Ceramic) Paralectric ceramic stack ±5% (J) 0 ±30 ppm/°C (Ultra-stable) RF filters, precision timing, PLL loops
X7R / X5R (Ceramic) Ferroelectric ceramic stack ±10% (K) / ±20% (M) ±15% over temp (Moderate) Decoupling, bulk bypass, general DC blocking
Aluminum Electrolytic Etched foil, oxide dielectric, liquid electrolyte ±20% (M) High drift, ESR rises as temp drops Bulk input/output filtering, high energy storage
Polymer / Solid Tantalum Sintered tantalum powder, manganese/polymer cathode ±10% to ±20% Stable ESR across temp, no dry-out Space-constrained low-voltage DC rail filtering
Metallized Film (PP/PET) Vacuum-deposited metal on plastic film, wound ±5% to ±10% Excellent, self-healing properties AC line filtering, snubbers, high-current audio

Decoding the Shell: How to Read Capacitor Markings

Capacitor markings are notoriously inconsistent compared to resistors. Through-hole ceramics and film caps often use the EIA 3-digit code, while SMD ceramics are completely unmarked, forcing you to rely on reel labels or precise measurement.

The 3-Digit EIA Code (pF base)

The first two digits are the significant figures, and the third digit is the multiplier (number of zeros) in picofarads (pF).

  • 104 = 10 × 10⁴ pF = 100,000 pF = 100 nF (The most common bypass cap in existence).
  • 473 = 47 × 10³ pF = 47,000 pF = 47 nF.
  • 221 = 22 × 10¹ pF = 220 pF.

Tolerance and Voltage Letter Codes

You will often see a letter trailing the numeric code. This denotes tolerance: J = ±5%, K = ±10%, M = ±20%.
For voltage, many Japanese and European manufacturers (like Murata or TDK) use a 1-letter/1-number code. According to Murata's standard coding, the number represents the first digit and the letter the multiplier:

  • 1H = 50V (1 = 10, H = 5.0 multiplier → 50V)
  • 1E = 25V
  • 1C = 16V
  • 1A = 10V
  • 0J = 6.3V
Bench Warning: Never assume an unmarked SMD ceramic capacitor's voltage rating based on its size. A 0603 10µF MLCC could be rated for 6.3V or 25V. If you salvage SMD caps from scrap boards, test their breakdown voltage or use them only in low-voltage (<3.3V) circuits.

Bench War Story: The 12V Rail Meltdown

To understand why the chemistry inside the capacitor matters more than the capacitance value, let us look at a real bench failure involving a buck converter input stage.

The Setup: I was prototyping a 12V-to-5V buck converter rated for 3A continuous output, switching at 500 kHz. The schematic called for a 22µF input decoupling capacitor. I grabbed a standard, cheap 22µF 25V aluminum electrolytic from the bin and soldered it in.

The Numbers: The input ripple current ($I_{in(rms)}$) for a buck converter is calculated as $I_{out} \times \sqrt{D \times (1-D)}$. With a duty cycle ($D$) of roughly 0.41, the input RMS ripple current was 1.47A. The standard electrolytic I chose had an ESR of 0.5Ω and a maximum ripple current rating of just 200mA.

The Outcome: Four minutes into the full-load burn-in test, a sharp 'pop' echoed across the lab. The capacitor's aluminum vent had ruptured, spraying vaporized electrolyte across the workbench.

What Went Wrong: I looked only at the capacitance (22µF) and voltage (25V > 12V). I ignored the heat generated inside the component. Power dissipation is $I^2 \times ESR$. $1.47A^2 \times 0.5\Omega = \mathbf{1.08W}$ of pure heat. A small radial can has a thermal resistance of roughly 50°C/W. That 1W of internal heat caused a 54°C temperature rise above ambient. The internal liquid electrolyte boiled, generating gas pressure that exceeded the mechanical limit of the vent score. The fix was to replace it with a parallel array of three 10µF X7R MLCCs, which dropped the effective ESR to under 5mΩ, eliminating the heat entirely.

Visual Autopsy: Recognizing Failure Modes

When troubleshooting a dead board, the physical symptoms of a failed capacitor tell you exactly what stress killed it. According to KEMET's failure analysis guides, different dielectrics fail in highly specific ways.

  • Aluminum Electrolytic (Thermal Runaway / Overvoltage): Look for a domed or bulging top aluminum vent. If the pressure was too sudden, the rubber bottom bung will be pushed out, and you will see crusty, dried brown/black electrolyte residue on the PCB pads. This usually indicates excessive ripple current or reverse polarity.
  • MLCC (Mechanical Flex Crack): Ceramic caps fail silently and invisibly. If the PCB was bent during depanelization or connector insertion, the brittle ceramic inside cracks. This creates an internal short. Visually, the cap looks perfect. You must use a multimeter in-circuit (looking for near-zero ohms) or inspect the cap under a 40x microscope to see hairline fractures near the termination edges.
  • Tantalum (Catastrophic Thermal Event): Tantalum capacitors do not have a pressure vent. When subjected to a voltage spike exceeding their rating, or installed with reverse polarity, the manganese dioxide cathode acts as an oxidizer. The result is literal fire. You will find a charred, blackened crater on the PCB where the component used to be.
  • Film Capacitor (Corona / Arcing): In high-voltage AC applications, if the internal metallization degrades, the plastic casing may show melt marks or a distinct smell of ozone and burnt plastic, indicating internal partial discharge.

The Substitution Matrix: Swapping Parts Safely

When your exact BOM part is on a 40-week lead time, you have to substitute. But you cannot just swap based on the microfarad rating. Follow these rules to substitute safely without altering circuit stability.

  1. Voltage Derating (The Golden Rule): You can always substitute a higher voltage rating for a lower one (e.g., using a 25V cap in a 10V circuit). Never go lower. For tantalum and aluminum electrolytics, derate the voltage by at least 50% (use a 10V cap on a 5V rail). For MLCCs, a 20% margin is usually sufficient, provided you check DC bias.
  2. The DC Bias Trap (MLCCs): If you are replacing a 100µF 16V aluminum electrolytic with a 100µF 16V X5R MLCC, your circuit will likely fail. Ferroelectric ceramics lose massive capacitance under DC voltage. That '100µF' MLCC might only provide 25µF at 12V DC bias. Always check the manufacturer's DC bias curve before swapping bulk electrolytics for ceramics.
  3. LDO and Compensation Networks (Do Not Upsize): If a capacitor is on the output of a Linear Dropout Regulator (LDO) or in the compensation loop of an op-amp, do not substitute a larger capacitance value. LDOs rely on a specific ESR and capacitance range for loop stability. Swapping a 10µF cap for a 47µF cap can push the pole frequency and cause the regulator to oscillate violently.
  4. Temperature Coefficient Upgrades: You can freely substitute a C0G/NP0 ceramic for an X7R, or an X7R for a Y5V. Never substitute a stable dielectric with an unstable one (like Y5V) in a timing or filtering circuit, as Y5V can lose 80% of its capacitance just from reaching room temperature.

Understanding what is inside the capacitor transforms it from a generic schematic symbol into a complex electrochemical component. By respecting the dielectric physics, reading the markings correctly, and calculating the hidden thermal stresses, you will build power supplies and signal chains that survive long past the prototype stage.