---
title: "Which kind of capacitor should feed a load?"
topic: "Capacitors"
answer: "The one whose time scale matches the load’s. On the same 1 A spike, a ceramic MLCC drops least for the first 0.31 µs, a polymer tantalum until 12 µs, a supercap until 0.29 s, and a lithium-ion capacitor after that. On the same 1 mW load, six kinds last from 1.1 s to 40 h."
url: "https://nearfieldnotes.com/notes/capacitor-types/"
site: "Near Field Notes"
author: "Michael Walker"
note: 13
family: "Power"
published: "2026-11-07"
revised: "2026-11-07"
revision: 1
video: "https://nearfieldnotes.com/videos/capacitor-types/"
---

Note 13 · [Power](https://nearfieldnotes.com/families/power/) · Rev 1 · [Watch the video · 2:39](https://nearfieldnotes.com/videos/capacitor-types/)

Capacitors

# Which kind of capacitor should feed a load?

**Short answer:** The one whose time scale matches the load’s. On the same 1 A spike, a ceramic MLCC drops least for the first 0.31 µs, a polymer tantalum until 12 µs, a supercap until 0.29 s, and a lithium-ion capacitor after that. On the same 1 mW load, six kinds last from 1.1 s to 40 h.

Each kind has a catch as well. The figure charges a 22 µF MLCC up to its rated 25 V, along Murata’s own curve. By 25 V it holds 1.8 µF.

In the close-up, the ceramic’s dipoles turn to line up as it charges. Once most have turned, each extra volt stores less.

**Note 13’s opening figure: A ceramic chip, charging.** Murata GRM21BR61E226ME44L, marked 22 µF, rated 25 V

![Note 13’s opening figure: A ceramic chip, charging](https://nearfieldnotes.com/notes/capacitor-types/fig-opening.png)

*What the figure shows:* A Murata 22 µF ceramic chip capacitor, cut open, with a close-up of its nickel electrodes and the ceramic between them, and Murata's chart of its capacitance against the DC voltage across it. It holds 22 µF at 0 V, 14.6 µF at 3.3 V, 10.2 µF at 5 V and 1.8 µF at its rated 25 V. In the close-up the ceramic's dipoles turn to line up as it charges.

Capacitance from Murata’s SimSurfing data, at 120 Hz and 25 °C. The close-up is a sketch, not to scale. Its dipoles turn in step with the charge stored, worked out from the same curve.

## Six ways to store charge

*Cut open, at true size*

Six common parts store charge on a board. Three are capacitors, with a thin insulator between two conductors. They are the ceramic MLCC, the polymer tantalum and the aluminum electrolytic. The [supercap](https://nearfieldnotes.com/glossary/#g-supercapacitor) packs ions against carbon. The lithium-ion capacitor is half supercap, half battery, and the chip battery is a battery built like an MLCC.

These are real parts of each. Their outsides follow the datasheets. Makers don’t publish the insides, so those are drawn from their construction drawings, with far fewer and thicker layers than the real parts have.

**Fig. 13: The six.**

![Fig. 13: The six](https://nearfieldnotes.com/notes/capacitor-types/fig-13.png)

*What the figure shows:* Figure 13: six parts that store charge, side by side at true size, each cut open by a quarter. 1, MLCC, Murata GRM21BR61E226ME44L, 22 µF 25 V · 2.0 by 1.25 by 1.25 mm. 2, Polymer tantalum, KEMET T520B107M006ATE040, 100 µF 6.3 V · 3.5 by 2.8 by 1.9 mm. 3, Aluminum electrolytic, Panasonic EEE-FK1C221P, 220 µF 16 V · Ø8 by 6.2 mm. 4, Supercap, KYOCERA AVX SCCR12B105SRB, 1 F 2.7 V · Ø8 by 12 mm. 5, Lithium-ion capacitor, Eaton HS1016-3R8306-R, 30 F 3.8 V · Ø10 by 16 mm. 6, Chip battery, TDK CeraCharge BCT1812M101AG, 100 µAh 1.5 V · 4.4 by 3.0 by 1.1 mm.

Rendered by one camera at one distance, so they compare at true size.

## The two tests

*Energy, then speed*

Each part takes the same two simulated tests. In the first, an ideal converter draws a steady 1 mW from it, from fully charged down to 1 V, or to 2.2 V for the lithium-ion capacitor, its floor. That asks how much energy it holds.

In the second, its load steps from nothing to 1 A in 10 ns, and its voltage is followed from 100 ns to 1 s after. That asks how fast it gives its charge up. The chip battery sits this one out. TDK rates it for 5 mA at most.

**Fig. 13b: The tests.**

![Fig. 13b: The tests](https://nearfieldnotes.com/notes/capacitor-types/fig-13b.png)

*What the figure shows:* Figure 13b: the two tests. In the first, an ideal converter draws a steady 1 mW from the part, from charged down to its lowest working voltage. In the second, a current source steps from 0 to 1 A in 10 ns, and the drop in the part's voltage is followed from 100 ns to 1 s.

The spike test uses each maker’s model where there is one, Murata’s for the MLCC and KEMET’s for the tantalum. The others are fitted to their datasheets.

## MLCC

*Murata GRM21BR61E226ME44L*

A multilayer ceramic capacitor stacks hundreds of thin nickel electrodes in ceramic, joined to each end in turn. Charge sits on the electrodes, and the ceramic between them stores more by turning its dipoles to line up with the field.

**Fig. 13c: MLCC, cut open.**

![Fig. 13c: MLCC, cut open](https://nearfieldnotes.com/notes/capacitor-types/fig-13c.png)

*What the figure shows:* Figure 13c: the MLCC, Murata GRM21BR61E226ME44L, 22 µF 25 V · 2.0 by 1.25 by 1.25 mm, cut open by a quarter, with a close-up of where it keeps its charge.

Charged, the dipoles between the electrodes all point the same way. The render has 24 electrodes, where the real part has hundreds.

MLCC

### Fast and tiny, until the volts take its capacitance

**Field marks.** A plain ceramic block with metal ends and no marking. Cut open, thin nickel electrodes reach toward each end in turn. The GRM21 is 2.0 × 1.25 × 1.25 mm.

**Look-alike.** The chip battery, built the same way in a larger case. Multilayer chip inductors share the shape too.

**Range.** Runs 1 mW for 1.1 s, and drops least on a spike for the first 0.31 µs. About 2.5 mΩ, the least here, with no polarity. It holds 14.6 µF at 3.3 V and 1.8 µF at its rated 25 V. A crack from a flexing board fails it short. It costs $0.36.

**Habitat.** Beside every chip’s supply pins, and on regulators’ inputs and outputs.

## Polymer tantalum

*KEMET T520B107M006ATE040*

A polymer tantalum starts as a pellet of tantalum grains fused into a sponge. An oxide grown on every grain is its insulator, and a conducting polymer fills the pores as the other side. The sponge’s huge surface gives it its capacitance.

**Fig. 13d: Polymer tantalum, cut open.**

![Fig. 13d: Polymer tantalum, cut open](https://nearfieldnotes.com/notes/capacitor-types/fig-13d.png)

*What the figure shows:* Figure 13d: the polymer tantalum, KEMET T520B107M006ATE040, 100 µF 6.3 V · 3.5 by 2.8 by 1.9 mm, cut open by a quarter, with a close-up of where it keeps its charge.

Charge sits on each grain, and the opposite charge in the polymer, across the oxide skin.

Polymer tantalum

### All its capacitance at any voltage, and the most leakage

**Field marks.** A black molded block, its leads folded under, with a band across the + end and its value on top: 107 for 100 µF. The T520 is 3.5 × 2.8 × 1.9 mm.

**Look-alike.** Older manganese dioxide tantalums wear the same black block and band.

**Range.** Runs 1 mW for 1.6 s from 5.67 V, the 90% of its rating KEMET recommends, and drops least from 0.31 µs to 12 µs. It keeps all 100 µF at any voltage. At most 40 mΩ, and up to 63 µA of leakage, the most of the six. Polarized. It costs $1.07.

**Habitat.** Bulk on rails under about 9 V, at the inputs and outputs of DC/DC converters.

## Aluminum electrolytic

*Panasonic EEE-FK1C221P*

An aluminum electrolytic rolls up a strip of etched aluminum foil with paper soaked in a liquid. An oxide grown on the foil is its insulator, and the liquid reaches into every etched tunnel to carry charge right up to it.

**Fig. 13e: Aluminum electrolytic, cut open.**

![Fig. 13e: Aluminum electrolytic, cut open](https://nearfieldnotes.com/notes/capacitor-types/fig-13e.png)

*What the figure shows:* Figure 13e: the aluminum electrolytic, Panasonic EEE-FK1C221P, 220 µF 16 V · Ø8 by 6.2 mm, cut open by a quarter, with a close-up of where it keeps its charge.

Charge sits on the foil, and the liquid’s ions pack against its oxide.

Aluminum electrolytic

### Cheap capacitance at high voltage, until it dries out

**Field marks.** A bare aluminum can on a square black plastic seat, its lid marked with a black half-moon over the − side. Cut open, a roll of foil and paper. The FK is Ø8 × 6.2 mm.

**Look-alike.** Polymer aluminum capacitors, in the same cans.

**Range.** Runs 1 mW for 28 s from 16 V, but never drops least on a spike: its 0.26 Ω takes its share at once. Leaks up to 35 µA. Rated for 2000 h at 105 °C, and every 10 °C cooler doubles that. Polarized. It costs $0.88.

**Habitat.** Bulk smoothing in converters, away from anything hot. Panasonic’s advice is to run it as cool as possible.

## Supercap

*KYOCERA AVX SCCR12B105SRB*

A supercap stores charge without an insulating layer. Two sheets of porous carbon sit in a liquid full of ions. Charged, the ions pack against the carbon’s surface, an area so large that a small can holds a whole farad.

**Fig. 13f: Supercap, cut open.**

![Fig. 13f: Supercap, cut open](https://nearfieldnotes.com/notes/capacitor-types/fig-13f.png)

*What the figure shows:* Figure 13f: the supercap, KYOCERA AVX SCCR12B105SRB, 1 F 2.7 V · Ø8 by 12 mm, cut open by a quarter, with a close-up of where it keeps its charge.

Positive ions gather at the negative carbon, and negative ions at the positive one.

Supercap

### A farad in a small can, at 2.7 volts

**Field marks.** A can on two leads in a royal-blue sleeve, a stripe down the − side and a scored Y in its bare lid. Cut open, carbon sheets wound in a roll. The SCC is Ø8 × 12 mm.

**Look-alike.** The lithium-ion capacitor, in a similar sleeved can.

**Range.** Runs 1 mW for 52 min, and drops least on a spike from 12 µs to 0.29 s. 150 mΩ at 1 kHz, growing to 500 mΩ over a 5 s discharge. Rated for 500,000 cycles. A cell stops at 2.7 V, so higher voltages take cells in series, with [balancing](https://nearfieldnotes.com/glossary/#g-cell-balancing). It costs $1.76.

**Habitat.** Holding a rail up through a dropout, backing up memory, and taking pulses off a battery.

## Lithium-ion capacitor

*Eaton HS1016-3R8306-R*

A lithium-ion capacitor is half supercap, half battery. Its positive side is carbon, holding ions on its surface like a supercap’s. Its negative side is graphite, with lithium slid in between its layers like a battery’s.

**Fig. 13g: Lithium-ion capacitor, cut open.**

![Fig. 13g: Lithium-ion capacitor, cut open](https://nearfieldnotes.com/notes/capacitor-types/fig-13g.png)

*What the figure shows:* Figure 13g: the lithium-ion capacitor, Eaton HS1016-3R8306-R, 30 F 3.8 V · Ø10 by 16 mm, cut open by a quarter, with a close-up of where it keeps its charge.

Negative ions on the carbon, lithium between the graphite’s layers.

Lithium-ion capacitor

### Battery-like energy that must never run empty

**Field marks.** A can on two leads like the supercap’s, in a slate-gray sleeve, its longer lead the +. Cut open, carbon and graphite sheets wound together. The HS1016 is Ø10 × 16 mm.

**Look-alike.** The supercap.

**Range.** Runs 1 mW for 40 h, the longest here, and drops least on a spike from 0.29 s on. Between 3.8 and 2.2 V it holds 144 J, 19 times the supercap’s energy for its size, and it leaks 3 µA at most. It must never go below 2.2 V, and can only be soldered by hand. It costs $4.14.

**Habitat.** Backup with bursts of amps. Eaton lists meters, storage servers, alarms and asset trackers.

## Chip battery

*TDK CeraCharge BCT1812M101AG*

The chip battery is a rechargeable lithium battery built like an MLCC. Thin electrode layers alternate with a solid ceramic electrolyte, and lithium ions cross the ceramic from one electrode to the next as it charges.

**Fig. 13h: Chip battery, cut open.**

![Fig. 13h: Chip battery, cut open](https://nearfieldnotes.com/notes/capacitor-types/fig-13h.png)

*What the figure shows:* Figure 13h: the chip battery, TDK CeraCharge BCT1812M101AG, 100 µAh 1.5 V · 4.4 by 3.0 by 1.1 mm, cut open by a quarter, with a close-up of where it keeps its charge.

Charged, the lithium has crossed into the next electrode down.

Chip battery

### A battery in an MLCC’s case

**Field marks.** A gray ceramic chip with metal ends, 4.4 × 3.0 × 1.1 mm, the 1812 size. Cut open, dark electrode layers between pale electrolyte.

**Look-alike.** An MLCC in the same 1812 size.

**Range.** Runs 1 mW for 75 s, sliding from 1.44 V to 1 V with only 17% of its charge used. That load is near its limit of 1 mA for 60 s. Its pulses top out at 5 mA, so it can’t take the spike at all. It reflows with the board, once, and ships uncharged. It costs $10.88, the most here.

**Habitat.** Keeping a real-time clock alive at microamps, with a capacitor beside it for pulses, as TDK advises.

## Running 1 mW

*The energy each holds*

How long a capacitor runs a steady load comes from the energy it gives up between its starting voltage and its floor:

`t = (C (V_0^2 − V_f^2))/(2 P)`

where t: time it runs the load (s); C: capacitance (F); V_0: voltage it starts from (V); V_f: voltage it stops at (V); P: power the load draws (W).

`= 1.6 s for the tantalum, 40 h for the lithium-ion capacitor`

The voltage counts twice. Charged to 16 V, the electrolytic outlasts the tantalum 18 times over, with about twice its capacitance.

The MLCC breaks the formula. At its marked 22 µF it would last 6.9 s, but its [capacitance falls](https://nearfieldnotes.com/glossary/#g-dc-bias) as its voltage rises, and it lasts 1.1 s.

The supercap and the lithium-ion capacitor run for 52 min and 40 h. The chip battery holds more than any of the first three, but at this load its resistance cuts it off early.

**Fig. 13i: How long each runs 1 mW.**

![Fig. 13i: How long each runs 1 mW](https://nearfieldnotes.com/notes/capacitor-types/fig-13i.png)

*What the figure shows:* Figure 13i: how long each part runs a 1 mW load through an ideal converter. MLCC: 1.1 s. Polymer tantalum: 1.6 s. Aluminum electrolytic: 28 s. Supercap: 52 min. Lithium-ion capacitor: 40 h. Chip battery: 75 s.

Each part charged to its working maximum and run through an ideal converter to 1 V, the lithium-ion capacitor to 2.2 V. The hollow dot is the MLCC as marked, as if it held 22 µF at every voltage.

## A 1 A spike

*How fast each gives charge up*

On a sudden step, two things pull a part’s voltage down. Its resistance takes a share at once, and the charge it gives up takes more the longer the step goes on:

`ΔV = I R + (I t)/C`

where ΔV: drop in the part’s voltage (V); I: step in current (A); R: the part’s resistance (Ω); t: time since the step (s); C: capacitance (F).

`≈ 3 mV + 78 mV per µs for the MLCC, 150 mV + 1 mV per ms for the supercap`

A small capacitor with almost no resistance drops least at first, then runs short of charge. A big one drops more at once, then barely moves.

The lowest drop passes from part to part. The MLCC holds it until 0.31 µs, the tantalum until 12 µs, the supercap until 0.29 s, and the lithium-ion capacitor after that. The electrolytic is never lowest. Its 0.26 Ω drops more at once than the supercap, and its charge runs out first.

**Fig. 13j: The drop on a 1 A spike.**

![Fig. 13j: The drop on a 1 A spike](https://nearfieldnotes.com/notes/capacitor-types/fig-13j.png)

*What the figure shows:* Figure 13j: the drop in each part's voltage after a 1 A step, from 100 ns to 1 s, on a log time scale. MLCC: 90 mV at 1 µs, past 1 V by 12 µs. Polymer tantalum: 38 mV at 1 µs, past 1 V by 97 µs. Aluminum electrolytic: 265 mV at 1 µs, past 1 V by 0.16 ms. Supercap: 150 mV at 1 µs, 165 mV at 1 ms. Lithium-ion capacitor: 250 mV at 1 µs, 279 mV at 1 ms. The MLCC drops least until 0.31 µs, the tantalum until 12 µs, the supercap until 0.29 s, then the lithium-ion capacitor.

Each part alone on the step, from 100 ns after its edge. The band follows whichever part drops least. The models are linear, so each drop is the same from any starting voltage.

## Side by side

*The six in numbers*

**Figure: The six.**

| Kind | Part | Rating | Size, mm | Runs 1 mW | Resistance | Leakage | Price |
|---|---|---|---|---|---|---|---|
| 1 MLCC | Murata GRM21BR61E226ME44L | 22 µF 25 V | 2.0 × 1.25 × 1.25 | 1.1 s | 2.5 mΩ typ | 11 µA max | $0.36 |
| 2 Polymer tantalum | KEMET T520B107M006ATE040 | 100 µF 6.3 V | 3.5 × 2.8 × 1.9 | 1.6 s | 40 mΩ max | 63 µA max | $1.07 |
| 3 Aluminum electrolytic | Panasonic EEE-FK1C221P | 220 µF 16 V | Ø8 × 6.2 | 28 s | 0.26 Ω max | 35 µA max | $0.88 |
| 4 Supercap | KYOCERA AVX SCCR12B105SRB | 1 F 2.7 V | Ø8 × 12 | 52 min | 0.2 Ω | 6 µA max | $1.76 |
| 5 Lithium-ion capacitor | Eaton HS1016-3R8306-R | 30 F 3.8 V | Ø10 × 16 | 40 h | 0.55 Ω max | 3 µA max | $4.14 |
| 6 Chip battery | TDK CeraCharge BCT1812M101AG | 100 µAh 1.5 V | 4.4 × 3.0 × 1.1 | 75 s | under 200 Ω | 0.1 µA | $10.88 |

Resistance and leakage as each maker gives them. The MLCC’s resistance is typical, the supercap’s is its DC value over 10 ms, and the MLCC’s leakage follows from its insulation resistance at 25 V. Prices for one part on 3 November 2026, from Digi-Key or Mouser.

Pick by how long the load lasts. An MLCC covers nanoseconds, a polymer tantalum microseconds, a supercap up to a fraction of a second, and a lithium-ion capacitor anything longer. The electrolytic’s place is smoothing at higher voltage, and the chip battery’s is a clock at microamps.

#### Conditions

- **Parts:** Murata GRM21BR61E226ME44L, KEMET T520B107M006ATE040, Panasonic EEE-FK1C221P, KYOCERA AVX SCCR12B105SRB, Eaton HS1016-3R8306-R and TDK CeraCharge BCT1812M101AG
- **Run test:** An ideal converter draws 1.000 mW at the part’s terminals, from its working maximum down to 1 V, or 2.2 V for the lithium-ion capacitor. Each part’s resistance is counted, and so is the leakage of the two that run for minutes or hours, at its datasheet maximum.
- **MLCC:** Its capacitance at each voltage from Murata’s DC-bias curve, at 120 Hz and 0.5 V rms
- **Chip battery:** TDK’s own discharge curves at 0.5 and 1 mA, between which 1 mW falls
- **Spike test:** A 1 A step with a 10 ns edge into each part alone, in ngspice, followed from 100 ns to 1 s
- **Models:** Murata’s SimSurfing netlist for the MLCC at 3.3 V, and KEMET’s K-SIM ladder for the tantalum. The electrolytic is a resistor, inductor and capacitor with its datasheet’s 0.26 Ω. The supercap and the lithium-ion capacitor are fitted to their datasheets’ resistances at 1 kHz and DC.
- **Renders:** Blender, one camera. Outsides from the datasheets, insides estimated from the makers’ construction drawings.

Simulated only. Nothing on this page has been measured on a board yet.

#### Limits

- The spike models are linear, so nothing runs out of charge partway. The MLCC is Murata’s netlist for 3.3 V, whose main capacitance is 12.9 µF, less than the 14.6 µF of its 120 Hz curve.
- Inductance is left out after the edge. It matters only during the 10 ns edge, and there it depends on the layout.
- The supercap’s and the lithium-ion capacitor’s slower time constants are estimates, fitted to the few resistances their datasheets give.
- Each part is at its nominal capacitance. The lithium-ion capacitor’s 30 F is its minimum, since its tolerance is 0 to +50%.
- The close-ups are sketches, and the renders draw far fewer and thicker layers than the real parts.

#### Files

- sim/parts.py The six parts’ values and spice models
- sim/tests.py The 1 mW run and the 1 A spike
- render/scene.py The renders
- refs/research.md Every datasheet figure and its source

#### Sources

- [Murata, GRM21BR61E226ME44 datasheet](https://search.murata.co.jp/Ceramy/image/img/A01X/G101/ENG/GRM21BR61E226ME44-01A.pdf)
- [Murata, SimSurfing curves and netlists](https://ds.murata.com/simsurfing/mlcc.html?reqtype=open_parts&partnumbers=%5B%22GRM21BR61E226ME44%22%5D&graphop=mainall&lcid=en-us)
- [KEMET, T52X polymer tantalum datasheet](https://search.kemet.com/component-documentation/download/datasheet/T520B107M006ATE040)
- [KEMET, K-SIM model](https://ksim3.kemet.com/?pn=T520B107M006ATE040)
- [Panasonic, FK series datasheet](https://industrial.panasonic.com/cdbs/www-data/pdf/RDE0000/ABA0000C1181.pdf)
- [Rubycon, aluminum electrolytic technical notes](https://www.rubycon.co.jp/wp-content/uploads/products-aluminum/al-technical-note_en.pdf)
- [KYOCERA AVX, SCC supercapacitor datasheet](https://datasheets.kyocera-avx.com/AVX-SCC.pdf)
- [Würth Elektronik, ANP090b, balancing supercapacitors](https://we-online.com/components/media/o671684v410%20ANP090b_Keep%20the%20Balance%20%20Balancing%20of%20Supercapacitors.pdf)
- [Eaton, HS hybrid cell datasheet](https://www.eaton.com/content/dam/eaton/products/electronic-components/resources/data-sheet/eaton-supercapacitor-hybrid-cylindrical-cells-data-sheet.pdf)
- [Eaton, what a hybrid supercapacitor is](https://www.eaton.com/us/en-us/products/electronic-components/faq/what-is-a-hybrid-supercapacitor.html)
- [TDK, CeraCharge datasheet](https://www.tdk-electronics.tdk.com/inf/75/ds/CeraCharge_BCT1812M101AG.pdf)
- [TDK, CeraCharge application note](https://www.tdk-electronics.tdk.com/download/2532510/3794c3c09a5c15b9befc89bba989f38c/06-dl-application-note.pdf)
- [ngspice, the circuit simulator](https://ngspice.sourceforge.io/)

What should the next note be? [Ask for one](https://nearfieldnotes.com/#ask-for-a-note)

---

Source: https://nearfieldnotes.com/notes/capacitor-types/

Cite as: Michael Walker, “Capacitors”, Near Field Notes, Note 13, Rev 1, 7 Nov 2026. https://nearfieldnotes.com/notes/capacitor-types/

Figures: free to reuse under CC BY 4.0 (https://nearfieldnotes.com/license/). Each one's image is linked under its name. Credit Michael Walker, Near Field Notes, and link the figure: this address plus #fig- and its number, like https://nearfieldnotes.com/notes/capacitor-types/#fig-13.