When/why (not) to use Tantalum capacitors. Pros, cons, alternatives

[Christe4nM]

A few weeks ago I worked through the Altium SMPS design course from FEDEVEL academy and that little project left me with an unresolved question. Somehow I got the idea in the past few years that tantalum capacitors are to be avoided, never really knowing why. Yet try to find a 330uF, 6.3V capacitor with ~4mOhm ESR  and you’ll find that tantalum is basically the only option. So I wanted to start a discussion about tantalum caps with the main questions being:

-   When and why to use tantalum capacitors?
-   Why avoid using tantalum capacitors?
-   Alternatives to tantalum capacitors with pros and cons.


I did some quick research already and here’s summary of what I’ve found so far

Why and when to use tantalum capacitors?
Tantalum is used to create small sized capacitors with ‘large’ capacitance. Compared to other materials the oxide layer can be quite thin. So for all applications where pcb space is limited (e.g. mobile phones) they are the to go to type of capacitor when ceramic doesn’t cut it anymore.

Also tantalum capacitors can be created with quite small ESR. This is why they are used a lot in (local) switched power supplies as bulk capacitor. Ceramics have even lower ESR, but in power supply regulators that might be too low for loop stability.

Why avoid using tantalum capacitors?
From what I can find these are the main reasons to avoid tantalum capacitors in your design:

Financial:
Tantalum capacitors are relatively expensive even when supply and demand are balanced. Not too long ago prices went sky high due to a shortage in raw materials.

Reliability:
There are numerous reports of tantalum capacitors spontaneously combusting.  I haven’t read into this any further, but I take it some forum members can elaborate on this.

Moral:
Since some of the major tantalum ore (columbite-tantalite / coltan) mines are located in conflict zones (e.g. Democratic Republic of Congo) so one can argue that tantalum is a conflict mineral. I’m not sure though to what extent that still counts today as there are major mines elsewhere in the world too.
Tantalum is a heavy metal and as such toxic and not really well for the environment. I did find a paper by AVX that discussed ‘green’ tantalum technology though.

Alternatives to tantalum capacitors and pros and cons?
Niobium caps
Apart from being mentioned I haven’t been able to find useful information yet, apart from that there is more Niobium ore available which should make Niobium caps cheaper. On the other hand Coltan is also used for mining Niobium so maybe the conflict argument could apply here too?

(Wet) Aluminum Electrolytic caps (Elco)
Specific low-ESR Elcos can be used. Larger values, larger ripple current rating, larger voltage ratings are some benefits. I must admit though that I realized that I haven’t seen SMT Elcos a lot outside consumer gear. Did I miss something?

Polymer (Solid) Aluminum caps
Used as in- or output bypass caps in SMPSs. Larger values only available in small voltage ratings. I don’t know by how much these must be derated though. Please tune in if you know more about this kind of caps.

Monolytic Ceramic caps
Larger voltage ratings, smaller derating, and higher ripple current ratings are among its benefits compared to tantalum caps. Since the ESR is much lower a small external resistance may be needed for loop stability in SMPS designs.

Sources
Here is some of the info I found. There is a lot to be found, and you'd have sift to get the info you want too. I included the year of publishing when available, since a few years in our industry means lots of changes.

AVX. (n.d.). Comparison of Multilayer Ceramic and Tantalum Capacitors

AVX. (n.d.). “Green” Environmentally Friendly Technology For Tantalum And Niobium Oxide Capacitors

Digikey. (n.d.). Tantalum Alternative Solutions by Panasonic

EETimes. (2001). Tantalum capacitor options weighed

Kemet. (2008). Comparison of Ceramic and Tantalum Capacitors

NIC Components. (1999). Alternates to surface mount tantalum electrolytic capacitors

Wikipedia. (n.d.) Coltan


Related Threads on the EEVblog
Tantalum caps

why we tend to not use electrolytics for decoupling to gnd?

Output capacitor of a DC/DC converter

(edit: URL layout issues )

[Rerouter]

My opinion on it is there is still a strong stigma from the 1980's - 1990's where endless bits of test gear succumbed to a tantalum shorting out a supply rail or worse, with the last bit of test gear i repaired having over 30 shorted tantalum's, where they used them for decoupling where some values where low enough that ceramic could have been used. (was a current limited switch-mode that would give a negative leading edge if it was driving too big of a load, which likely resulted in blowing one after the other.)

in that case, almost all of them where replaced with an electrolytic in parallel with a ceramic, (high speed logic) or plain ceramic on its own, and though i haven't gone and tested the transient response of the board, its happily working.

[dannyf]

never really knowing why.

(Earlier liquid / paste-based) tantalum capacitors tend to fail short and regularly. Newer solid state tantalum capacitors don't suffer that particular failure mode and in my view should be used.

~4mOhm ESR

Low ESR isn't the only design criteria and in some cases may not be desirable. Great pieces of equipment were produced before tantalum capacitors and can be continued produced without them - you just need to know how to engineer them.

[calexanian]

Tantalams are great. We have used them in our products for over 30 years and to date have NEVER had a field failure of one without an extreme fault otherwise happening to the unit. Yes, its a left over stigma from the days of old. When people would accidental install them backwards in low impedance circuits they will explode with a small fire ball. They actually make great electric matches. I have also NEVER seen one leak when designed with properly. Ye use about 6 on every single one of our boards and have sold from 1500 to 3000 a year of them for most of the last 30 years. For a matter of cost however we limit our use of them to applications under 35 volts and 10 uF. Above that their attributes are a law of diminishing returns anyways.

[nctnico]

Tantalums are basically bombs waiting to go off. Oxidizer + fuel. You'd need to control the rate of the current flowing into them or they may ignite. Reverse mounting is also asking for trouble. They will burn right through the board or explode in your face (both happened to me and it was not my fault). Besides that the raw material is found in a place where nature is to be preserved.

Nowadays there really is no reason to use tantalum caps.

[Rufus]

All electrolytic capacitors are have self forming and self repairing dielectric layers.

The problem with tantalums is if too much current is allowed to flow through a fault in the dielectric the localised heating can cause more damage not repair. The damage can cause thermal runaway and rapid catastrophic failure.

For that reason it is unwise to use tantalums in low impedance circuits which can source a lot of current and wise to substantially underrate their voltage if you must. Some manufacturers recommend adding series resistors to limit current which makes their low ESR rather pointless.

[SeanB]

Tantalum is really only used where you need ultra small, and the voltage rating is a thing that never ever will be exceeded, and where you never apply a reverse voltage to them, no matter how briefly. You do need to limit power put in them, but if you derate voltage enough they have a very high current capacity.

I replaced a 1000uF 16V electrolytic on my PC power supply once with a 10uF 25V wet slug tantalum unit, as I needed a capacitor and this was the only one to hand. Worked well for years afterwards, even though the current in the cap was large. the ripple was lower than what the original capacitor had. This was in a LC output filter on the 5V rail, where the original popped it's top.

[CSmith]

The biggest risk with tantalum/nobium capacitors are surges and any reverse polarity at all. These risks can be largely mitigated by generously overrating their voltage. Doubling is a good start. AVX has some good white papers on this stuff.
AVX Surge in Solid Tantalum Capacitors
AVX Voltage Derating Rules for Solid Tantalum and Niobium Capacitors explores when not quite doubling the voltage rating may not be necessary

[calexanian]

You still have to design the circuit right. Other than that they are fine.

[IanJ]

My opinion on it is there is still a strong stigma from the 1980's - 1990's where endless bits of test gear succumbed to a tantalum shorting out a supply rail or worse

Jings, that takes me back!.............I can remember that burnt tantalum smell as I type.
We used to use them all the time for supply decoupling. It was by far the most common reason for electronics failure.

Ian.

[Niklas]

One drawback with tantalum caps is the significant voltage derating at high temperatures. In an automotive application I recently worked with, parts of the potted PCB was a bit over 100 degrees C, resulting in a 70% derating. The 35 V rated capacitor was not good for more than 12 Volts over the complete range of operating temperature.

[G0HZU]

I'm happy to use them although the inrush and discharge current needs to be managed.

The worst tantalum fireball I ever saw was an old school 'blob' type that decided to fail in a big way. It actually detached from the PCB and landed on a work bench and proceeded to eat a decent sized crater into the bench as it crazily fizzed and circled around like a white hot bumble bee that was high on fly spray.

[calexanian]

Once again. You have to design the circuit right. They have no business in power supplies. They are really for RC circuits!

[G0HZU]

I can't remember for sure but the fireball tant was a very early example of a tant cap in a card from an old radar based product. I guess in those days they didn't realise all the risks from these devices. I'd be surprised if a modern tant could expire in quite the same spectacular manner but I know they can get extremely hot and go pop. You also get the classic smell of a dead tant cap filling the lab...

I've seen nothing like it since and I've seen plenty of dead tants. Not quite the same as 'Alien' acid blood eating through a bench but it did make quite a decent crater 

[calexanian]

I went back and re read the question. So in short, to get esr down you can parallel smaller caps. Electrolytics and add a Mylar cap in there for higher freq stuff. So tantalams as said before have an explosive failure condition, with a side of fire ball. The key is to use them in applications where they will be safe. For power supply filtering they do little. Yes low esr, but you can parallel other caps to the same effect. Also there are high ripple low esr electrolytic caps. Typically orange outer jacket.

[dannyf]

Also there are high ripple low esr electrolytic caps.

Low ESR in some cases can be evil.

I do agree with your other point: it is far more important to know what you are doing than to just drop in a low ESR over-spec'd part.

[Christe4nM]

OK, so far the consensus considering reliability seems to be that tantalum caps can be used without reliability problems if you take care to design the circuit right.

Nowadays there really is no reason to use tantalum caps.

Could you elaborate on that apart from their failure mode?

Tantalum is really only used where you need ultra small, (…)

So basically “you’ll know when you need ‘em” applies here? In other cases use alternatives?

Once again. You have to design the circuit right. They have no business in power supplies. They are really for RC circuits!

This is new to me. The only places where I’ve seen tantalum caps are SMPSs. With RC circuits I immediately think “film caps”, although realizing this is another “I don’t really know why, probably I’ve just been told so once”

[calexanian]

Classically rc timing circuits were a lot longer I. Duration than today. Delays of up to many seconds were common with 555 timers and tantalum caps were the bees knees for that. Also for delayed rise time there are propably billions of them out there. The mechanism of bang on them is durring over current/voltage heat and gas are generated and they have no vent like an electrolytic so bang is the result and the remains actually are combustible.

[LukeW]

Since some of the major tantalum ore (columbite-tantalite / coltan) mines are located in conflict zones (e.g. Democratic Republic of Congo) so one can argue that tantalum is a conflict mineral.

I reckon this is not as big a deal as some people think it is.
Do you know which country supplies most of the world's tantalum? Hint: A country known for digging stuff up, venomous animals, and eevblog.

Personally, I find myself using tantalums less and less today because electrolytics are getting better and better (better reliability and better ESR, and smaller) and SMD multilayer ceramics are getting better and better (higher capacitance and higher capacitance density, at far lower costs than they used to be for the large ones, and lower ESR than Al or Ta.)

Another factor to consider for a "where would I choose tantalum" question might be "where would I not choose wet aluminium electrolytics" where large capacitance is needed - for example in high temperature environments, or very low temperature (below freezing) or vacuum environments (eg aerospace) you probably do not want a conventional aluminium electrolytic.

[calexanian]

That is true. Smd multilayer ceramic will over take tantalum very soon and are far better characterized.

[Kjelt]

How about the reflow "survival" resist, I personally have a lot of bad experience with SMD electrolytics after a reflow process. The small ones are ok but larger ones really need to be glued down which to me indicates a lot of stress from the process. Any one have experience with this?

[nctnico]

OK, so far the consensus considering reliability seems to be that tantalum caps can be used without reliability problems if you take care to design the circuit right.

Nowadays there really is no reason to use tantalum caps.

Could you elaborate on that apart from their failure mode?

Ceramics with similar capacities are available and cheap (especially when combined with high frequency SMPS chips).

[3roomlab]

sorry for reopening an old thread. but i think my question is similar related.

having no usage experience with tantalums, i thought i try to ask if anyone could enlighten me. i did read up a few articles about pros cons etc, however i am puzzled by what i find in E14.

the part i am considering to replace (from a used keithley 2015) is this (see picture). how do you determine its working voltage rating? marking 22uF "10B42" = 10volt, B42 = batch?

assuming it is 10v. i had a look at E14, but then i found out that the ESR rating listed are mostly over 0.5ohm and then right down the list this 22u appears and it costs about USD12 a piece  .  (http://sg.element14.com/avx/trme226k035r0060/cap-tant-22uf-35v-case-e/dp/2353302) do you guys normally buy this?

[Kjelt]

You can find on the same site tantalums of 22uF for $0,10 a piece.
If you can't find the voltage you have a few options:
- measure the voltage across it
- reverse engineer (draw) that part of the circuit
- choose the highest voltage possible, one of a kind anyway.

[3roomlab]

thanks for the tip on AVX. found it inside a AVX catalogue (page 147). E is 25v, the cap is rated for 85^oC
the black SMD above it is a inductor - L105 marking

**edit based on selection criteria of 25v 22uF++. there is no 10cent tantalum in sight. is that even possible for single qty? (the situation is worse on RS website)