Observations on cap composition and stability

[Elitist]

There was a time when Nichicons were flavour of the month (not necessarily on this Forum). With several projects in prospect, friend bought a fresh batch of 2200/6.3 whilst in Japan and left them with me. Technology and events evidently moved along. ~10yrs later, rediscovered them still unused. Every one had that familiar bulge and a little brown stain. All goes to show the importance of foil purity and design of electrolyte, along with adequate manufacture 'forming & soak' and QC. Left unpolarised for long periods may not be best practise either, especially if design, manufacture, etc have not been attended to. After all, neither entropy nor the inexorable nature of diffusion across a chemical potential gradient can be denied.

[Wester547]

Re: Observations on cap composition and stability

What are the series and datecodes? Are they the HM, HN, or HZ series and do they have 2001-2005 datecodes?

[Elitist]

Re: Observations on cap composition and stability

Ah ha! An expert. They are marked HN(M), 105^oC, H0453. Does that mean 2004, week '53' - bearing in mind the leap year? Appreciate what extra light you can shed. Are they known offenders?

[Wester547]

Re: Observations on cap composition and stability

I wouldn't call myself an expert. Not even close.

H0453 means they were produced in Ohno, Japan, during the 53rd week of 2004 (late December 2004 to early January 2005). (M) denotes the capacitance tolerance (-+20%, not actually part of the series name). I believe you were active on this forum around the time of the "capacitor plague", well over a decade ago. It was around that time that Nichicon bestowed quite the plethora of bad capacitors upon the world, with their faulty HM and HN series. Their frequency of failure was high enough that Dell had to deal with a $300 million lawsuit for using faulty HN capacitors, particularly in their Optiplex GX270 series (and faulty HM series in the Optiplex GX260 line). Their failure rate was 100% within a year or less (sometimes even six months). I would guess those Nichicons you acquired are genuine and not counterfeit.

I doubt the issue with those Nichicons was foil impurity, as Nichicon are known to use 99.9% pure (ultra pure) cathodic and anodic aluminum foil. They hatched after sitting on the shelf for ~11.5 years because the electrolyte is unstable enough to break down and release hydrogen gas all by its lonesome, without any potential across the leads (outside of the "aging" step of the manufacturing process which forms the anodic oxide layer so as to prevent infant mortality failures) or without having conducted any ripple current. This can happen at room temperature as well, of course, as heat only expedites all chemical reactions (also known as the Arrhenius theory, that with every 10˚C drop in temperature, the lifespan is doubled). If it was purely a physical phenomenon, heat would not be a factor (IE dielectric dissolution or phase change).

QC testing is of course everything (although Nichicon have held claim to a "zero defect policy" since 2002 or thereabouts, but no doubt that is marketing at its finest). The issue with those Nichicons, at least in my mind, is that just like the decade old Taiwanese and Chinese capacitors of the plague, they use a faulty water base electrolyte. Not the same exact formula of course, but electrolyte with a high concentration of water requires extreme purity of all materials, very precise manufacturing, and special oxidizers to rein in the aggressive tendencies of that type of electrolyte (which has a tendency to attack the aluminum oxide and beget a phenomenon known as foil corrosion and hydration). The original "rumor" and "theory" about those Nichicons is that they were "overfilled", but that's not actually true. That came from an Iowa State University notice and advisory that was filed on Dell's behalf, not Nichicon. There was another announcement in 2003 (not 2004) that forewarned not to use certain disinfectants to clean capacitors (in preventing SARS), but Rubycon and Chemi-con also issued similar statements on that matter so it had nothing to do with the HM and HN series (no mention of them was made in that announcement). All three companies also issued notices in 2006 to warn users against counterfeit capacitors.

PCBONEZ claimed that Nichicon held a press conference to concede the problem and pull the faulty capacitors from the supply channels before 2005, but that story is not consistent with reality, as those HM and HN series capacitors manufactured in early-mid 2005 were still known to fail prematurely (and for no reason). It also could not have been around the same time that Dell made their announcement because their announcement surfaced in 2005 and not 2004. So I'm not convinced such an announcement actually exists (until he decides to randomly show up again one day and post it to prove his word, if he can find it ) as I have not found hide nor hair of it on the internet, nor on the web archive of Nichicon's website. What's also interesting is that a 2005 CNET article notes a spokesman from Nichicon declined to comment upon the matter when questions were asked - why would they decline to comment on the issue if they held a press conference to admit the problem and pull the faulty capacitors the year before?

The other two issues with faulty Taiwanese and Chinese capacitors are foil impurities and highly basic (high pH values) electrolyte. The first issue is known to cause galvanic couples to form with the impurities in the aluminum foil (too much copper, iron, magnesium, zinc...) and the hydrogen ions in the electrolyte, producing hydrogen gas once again. The second issue is what happens when the electrolyte proves to be basic rather than acidic, and it's well known that aluminum will easily and readily dissolve into electrolyte that is highly basic (especially without the proper protection barriers, E.G phosphate compounds, in place to protect the electrolyte). Strong alkalis (basic electrolyte) are also known to accelerate the degradation of the rubber seal.

With all this said, the HM and HN series with 2006+ datecodes have been known to be reliable, so I do think Nichicon quietly took care of business. Sorry for the long winded nature of this post, I just thought I'd cover every talking point.

[Elitist]

Re: Observations on cap composition and stability

Impressive - thanks! Your memory serves you well, too. Not so sure about the causes and the chemistry, though. Al purity might well be an issue for any manufacturer at any time, especially a rogue batch. Its production is a miracle of technology in its own right.Sold in giant bars (at least, prior to rolling) and so little in each cap, a dose of stringers at one end of a single billet might suffice to manufacture several million (billion?) caps. The role of water may be a red herring, too, as nearly all the early ones used an aqueous electrolyte and I've got some that are ~50yrs old, maybe more, still in good working order! Now when they started to used exotic organic additives - that immediately triggers attention. Svante's little catch22 certainly applies to most of them, esp. in solution, esp. in mixed aqueous one and esp. much above room temperature. Of course, the miracle of the barrier film is that without it the metal could spontaneously ignite, possibly explode, in air. A colleague in the capacitor industry once described the whole technology as more a black art than a science...

[Wester547]

Re: Observations on cap composition and stability

Just as a postscript and a correction, I meant to say that impurities in the aluminum can react with the ions in the electrolyte.

Impressive - thanks! Your memory serves you well, too. Not so sure about the causes and the chemistry, though. Al purity might well be an issue for any manufacturer at any time, especially a rogue batch. Its production is a miracle of technology in its own right.Sold in giant bars (at least, prior to rolling) and so little in each cap, a dose of stringers at one end of a single billet might suffice to manufacture several million (billion?) caps.

I understand that the minimum level of purity specified by DAPO is 98%. Many of the Taiwanese and Korean brands have actually started sourcing their aluminum foil (and rubber bungs) from a Japanese joint venture known as K-JCC and Taiwan JCC (Korea and Taiwan Japanese Capacitor Foil Corporation - the former was a joint venture started with Samwha, the latter with Taicon; Samwha is Nichicon's Korean subsidiary and Taicon Nichicon's Taiwanese subsidiary) over the past several years or so (the purity of their foils would be 98%-99.4%). Samyoung (Korea Chemi-con, or Chemi-con's Korean Subsidiary, which they own 33.4% of) also have their own factories for producing aluminum foil, and on their website, they claim the purity is 99.9%. So I think the issue lies in the electrolytic composition. There was a study conducted back in 2004, in Maryland, in which they noted they did not think the impurities in the aluminum foil could completely account for the rapid generation of hydrogen gas (although they certainly admitted the possibility that said impurities could cause hydrogen gas to form at the cathode, especially without depolarizers to absorb any excessive hydrogen or gas). It's true that cracked aluminum or aluminum foil that isn't etched properly could certainly result in catastrophic failure.

The role of water may be a red herring, too, as nearly all the early ones used an aqueous electrolyte and I've got some that are ~50yrs old, maybe more, still in good working order! Now when they started to used exotic organic additives - that immediately triggers attention. Svante's little catch22 certainly applies to most of them, esp. in solution, esp. in mixed aqueous one and esp. much above room temperature. Of course, the miracle of the barrier film is that without it the metal could spontaneously ignite, possibly explode, in air. A colleague in the capacitor industry once described the whole technology as more a black art than a science...

It wasn't until the late 90s that electrolytics using a concentration of 40% water or more were actually in production. Rubycon was a leader in developing that electrolyte technology. They were the first brand to find the proper mixture of suppressors and inhibitors to avert corrosion and hydration (with the release of the ZL and ZA series in Fall 1998). Then came Rubycon MBZ in Spring 2001, the very first ultra low ESR electrolytic to be produced. As I recall, when KZGs (by Chemi-con) and HMs (by Nichicon) were failing left and right, the MBZs were holding up just fine. Clearly, Rubycon allotted the proper resources, finances, and a dedicated staff of QC researchers and chemists to make sure that they were developing reliable capacitors at all and any cost. Furthermore, Rubycon's main factory (in Nishikoma, Japan) borders on a spring of incredible purity. Even slightly contaminated water-base electrolyte can spell a certain doom for any electrolytic employing such a solution. Panasonic also had extra incentive to develop reliable capacitors, being one of their own biggest customers. Of course at least a marginal presence of water is necessary even in non-aqueous solvents and solutions as it is needed to provide the right amount of oxygen to reform the oxide film (to correct or repair defects in the oxide layer).

What I also understand about "aqueous" electrolytes as they have been an issue for many decades. Even the first supposedly "water free" glycol-based electrolytes (those rated from -25˚C to +85˚C) yielded an unwanted esterification reaction that would generate up to 20% water in the electrolyte. Water is one of the cheaper solutions used to increase the ionic conductivity (not an easy thing to do!) in the electrolyte (deionized water is often used because of that). The higher presence of water also lowers the boiling or decomposing point of the electrolyte (water boils at 100˚C and gamma-Butyrolactone at 204˚C - quite the difference, although I'm referencing pure water which freezes at about -40˚C and not the water that freezes at 0˚C to be found in faucets). Using exotic additives could be a hazard, especially if they are chemically incompatible with the rest of the materials.

Those 50 year old capacitors must be sealed very well to still measure to spec. It is very true that developing electrolytics is a black art - a black art only a select few companies actually understand. When you are dealing with the volatile nature of chemicals, problems are bound to arise when you start to cut any corners. With the faulty HM and HN series (and the faulty KZG and KZJ series by Chemi-con), some corners were definitely cut if not fully rounded.

[mockingbird]

Re: Observations on cap composition and stability

Bookmarked this thread for future reference.

Fantastic, brilliant, amazing post as always Wester547.

[Wester547]

Re: Observations on cap composition and stability

Haha, thanks. I can't take all the credit, though. Elitist brought up some very valid talking points. Just to be clear, I'm not trying to show off with these long and fastidiously detailed posts (should anyone get that impression), I just find the topic fascinating. It is very possible for the metallic aluminum to combust without the proper oxide barriers in place or intact, especially aluminum of lower purity (which is why high purity aluminum is so critical in the development of aluminum electrolytic capacitors, and why some companies insist on using aluminum of that grade at every and any cost). Hydrogen gas isn't the only way to create immense pressure inside the can. This is another reason why the "aging" part of the manufacturing process is so imperative (and why using the right additives and compounds in the electrolyte is equally vital). When the oxide layer fails, either by way of direct electrolyte attacks or just dissolving into the solution (or containing indelible defects), all sorts of issues can arise. Better seals will make this less a likelihood as well of course.

I'm not trying to knock Nichicon (I brought up the possibility of the HM and HN series containing faulty water base electrolyte because they fail with high leakage current and read very high capacitance way too often) and Chemi-con either if I impressed upon anyone as such. Rubycon's MCZ (and MFZ) series is known to be almost as thermally sensitive as Chemi-con's equivalents (although more chemically stable for sure), but I can't lay blame to Rubycon for that either as it's Dell, Intel, Foxconn, IBM, Asus, HP, etc, who came up with the brilliant idea of stuffing these ultra low ESR and highly aqueous electrolytics right next to burning hot MOSFETs, heatsinks, and coils in poorly ventilated small form factor machines with very limited airflow (although all other factors being equal, increased airflow could actually accelerate evaporation). This is the design component that compromised the longevity of said electrolytics (or to cause them to reach liquidus temperatures). And it seems Chemi-con was the only company of the "good brands" to admit to the possibility of defective products:

https://www.chemi-con.co.jp/e/ir/man_risk.html

^ Risk factor #5. They also admit to the chance that large-scale defects could occur (a la KZG and KZJ). I wonder how those series financially influenced their position?

[Elitist]

Re: Observations on cap composition and stability

Not a lot to add to W547's excellent exposition. For those like mbird who follow the subject, perhaps we should make it clear that there are two (entirely different) oxide films on aluminium in caps. It's the inner barrier oxide layer that protects the metal from catastrophe. This has atomic dimensions. It's the outer 'porous ' oxide layer that is critical in capacitor construction as it absorbs the working electrolyte; this must contain water which enables formation, reformation and repair functions. It's this film where the magic occurs - or not! It is vulnerable at all times to 'impurities', breakdown, dissolution, whatever. Notwithstanding, impurities in the metal can migrate/diffuse through both layers under the influence of the chemical concentration gradient. These forces are normally considered 'slow', but when distances are measured in atomic thicknesses, timescales in years, impurities in ppm (ppb?!) and field gradients in MeV the outcomes may not always be calculable, even using the powerful modelling tools now available but not so a decade ago. Much more has been written in the technical Press, but it's what they don't tell us that's important.
Just in case we focus too closely on capacitors, it's worth remembering that all the above considerations are equally important in eg aeroplanes and the plethora of aluminium products we use today.

[Wester547]

Re: Observations on cap composition and stability

Forgot to mention Apple in the listing of those who are guilty of placing wet electrolytics right next to red hot MOSFETs and chipsets. Also forgot to mention Toshin Kogyo (a company headquartered in Japan) for their lucidly and notoriously unreliable ultra low ESR ATWY and ATWB series - those are just as bad as the older, faulty HMs and HNs and I very much suspect they suffer from a faulty (highly) water base electrolyte as well (at least I see the very high capacitance readings and high leakage current as a result of a dramatically thinned or non-existent anode oxide, followed by the formation of hydrides... again, corrison inhibitors would put at least a temporary stop to this). I believe OST is actually only TK's distributor and not the actual manufacturer of their capacitors. TK's other series (besides ATWT maybe) seem to be alright, though. The ATWY/ATWB series seem to be inconsistent - sometimes every single one is bulged, other times they dry up silently.

Reforming the porous anodic oxide layer also reduces the surface area of the aluminum foil (and in the wrong scenario, the oxygen used to reform the porous layer can diffuse into the electrolyte and increase leakage current). I understand that this decrease in surface area results in the reduction of capacitance over time. To add to Elitist's very good points, the cathodic dielectric (the cathode electrode) is also important. When one reverse polarizes a capacitor, the breakdown voltage of the cathode dielectric (usually anywhere from +1 to +1.5V) is exceeded and that causes the generation of hydrogen gas, the vehement mess that follows, etc. Yes, purity of aluminum is arguably far more critical in other applications such as vehicles and transportation where one's life may depend upon just such a quality! And it isn't just the purity of the aluminum that matters, it's the specific impurities that exist in the metal regardless of the purity level.

As for impurities leeching into the electrolyte, this is why it is crucial to perform the endurance tests, shelf life tests, etc, at the factory, to properly understand the chemical behavior of the inner materials (all of them are of importance, from the barrier oxide to the lead wires).

[Elitist]

Re: Observations on cap composition and stability

OK - a few additional comments, if W547 will permit.
Any (with special consideration to electrolytic) capacitor is a form of secondary battery i.e. it stores charge. That charge can be discharged and recovered (stating the obvious!). The problem with electrochemistry is that it takes place at an interFACE, rather than an interphase, so it is, essentially a 2D phenomenon. 2D processes require as large as possible surfaces to enable as large as possible charge transfer. Using rolled up foil electrodes is a great way of increasing the available surfaces. However that is not sufficient, not for modern capacitors nor batteries. The partial solution presently deployed in devices is to create extended electrode surface areas. In capacitors this is achieved by forming a porous oxide layer impregnated with electrolyte. In batteries various arrangements of compressed particles comprise the electrode layers. However, there is a downside to this strategy - current collection through an inhomogeneous, maybe only weakly conducting matrix, induces serious iR loss. In batteries, some of these losses can be partly mitigated by incorporating carbon powder. On the one hand this is conductive so can assist electron transfer up to the primary interface. On the other hand it causes parasitic losses before all the charge can arrive safely at the 2D business end. In electrolytic capacitors, one is always obliged to rely on poorly conducting oxide films, albeit wetted by conductive electrolytes, to transfer electrons. Well, that's the way they teach the basics, but as with everything, it is a lot more complicated because electrons in aqueous (or cyclic/aliphatic ethers & co.) solutions tunnel rather than move. Remember, the oxygen lone pairs are engaged in H-bonding. The concept of electron drift in copper may be more familiar to electrical/electronic engineers. This isn't really a direct parallel to electrolytic proton exchange (wrongly termed 'transfer' in most textbooks!), but does remind us that all is not what it sometimes seems...

[Wester547]

Re: Observations on cap composition and stability

OK - a few additional comments, if W547 will permit.

No need for me to permit as it's your thread. All good points.

In capacitors this is achieved by forming a porous oxide layer impregnated with electrolyte.

Yes, the impregnated electrolyte will saturate the paper separators and successfully penetrate the etched tunnels on the aluminum foil.

In electrolytic capacitors, one is always obliged to rely on poorly conducting oxide films, albeit wetted by conductive electrolytes, to transfer electrons.

To add to this, the electrolyte is the true cathode that actually conducts current. The cathode foil only exists to allow for that "connection" with the electrolyte. When the electrolyte gradually dries out by way of rubber seal diffusion, the cathode electrode disappears and thus the connection to the electrolyte vanishes.

Remember, the oxygen lone pairs are engaged in H-bonding.

Yes, this is correct. Particularly in water base electrolyte, as leakage current flows, the H²O is broken up into oxygen and hydrogen by way of hydrolysis, and the oxygen bonded to the anodic foil heals areas of leakage in the dielectric, creating more oxide. Any excess hydrogen should pass through the rubber bung.

The largely (and relatively poor) ionic conductivity of liquid electrolytics is one reason why many series are being obsoleted in favor of solid conductive functional polymers. Most polymers use PEDOT or poly(3,4-ethylenedioxythiophene), which yields many times the conductivity of water. Polymers rely solely upon electronic conductivity. Their conductivity and thermal stability relative to capacitance and ESR is excellent. The catch22 is that polymers are unable to "heal" their dielectric the way electrolytics can (they can only isolate faults in the dielectric, not repair them). So their leakage current is bound to be much higher and sporadic shorts in the oxide are more a likelihood.

[Elitist]

Re: Observations on cap composition and stability

Oh yes- the Group VI chalcogens! Oxygen, the planet's poison payload from the cosmos without which life could not exist. See how toxicity increases with atomic weight. Organo-sulphur compounds are as notoriously toxic as they are complex with their plethora of lone pairs - originally deployed in gas warfare and cancer treatments. Amazing how technology develops.
Just a minor correction - leakage currents cause electrolysis; we usually reserve hydrolysis for homogeneous reactions although one might argue the point in heterogeneous catalysis, for example. Notwithstanding, by definition, all chemical reactions involve electron transfer, tunnelling or rearrangement, so there may be semantics in the argument.
As for the 'cathode' foil, yes - its primary function is as current collector. But readers will be cognisant of the secondary battery conundrum: which is the cathode and which is the anode depends on whether it's being charged or discharged - a constant cause of confusion amongst students! Best to consider what the electrons are doing at the time. From the capacitor viewpoint, choosing aluminium as the cathode proved fortuitous!

[Wester547]

Re: Observations on cap composition and stability

I wasn't saying that leakage current directly causes hydrolysis, but that the event that is reforming the anode porous film involved hydrolysis, or the breakdown of hydrogen and oxygen in their reaction with water. I do understand that electrolysis (or the decomposition of an ionic liquid once a current passes through) is directly responsible for any hydrogen gas (H2) that the depolarizers don't absorb (if present) or neutralize, as a result of the electrolyte conducting ripple current.

[Elitist]

Re: Observations on cap composition and stability

Fair enough.
We didn't get round to ripple, did we?!

[Wester547]

Re: Observations on cap composition and stability

Ripple current would be the most prominent source of internal heating. Ambient temperature of course would serve as the other component of heat. I understand that ripple current is essentially (the RMS value) of the AC current flowing through an electrolytic (or the AC voltage superimposed on the DC bias voltage of the capacitor - ripple voltage is essentially the result of unwanted variations in the DC output derived from AC, or the ever changing waveform). For those electrolytics whose ripple current handling abilities are rated at 105˚C (with the ripple current increasing the core temperature itself 5˚C above the rated temperature), the ripple rating should theoretically increase at lower core and ambient temperatures, all else being equal.

[Elitist]

Re: Observations on cap composition and stability

Not just heating but faradaic rectification, too!

[linuxguru]

Re: Observations on cap composition and stability

Samyoung (Korea Chemi-con, or Chemi-con's Korean Subsidiary, which they own 33.4% of) also have their own factories for producing aluminum foil, and on their website, they claim the purity is 99.9%.

I suppose it's a bit early to take a call on reliability, but Samyoung seem to have some claimed 5000 hour, 105c, moderate/high ripple-current, high-voltage electrolytics. I have some Samyoung NFK 47uf/160V in hand for pre-emptive retrofits on cheap Chinese LED bulbs - the exterior build quality of Samyoung is excellent, comparable to or better than NCC. We'll see how it goes on long-term reliability inside a warm/hot 18W LED bulb in a few months/years - I suspect that the SMD LEDs or the voltage-dropping 400V CBB21/22 film capacitors will fail before the electrolytics.

[Wester547]

Re: Observations on cap composition and stability

Forgot to mention that ripple current (or the RMS value of AC current) also creates a reverse voltage on the cathode foil, or forms the cathode foil if you exceed the rated ripple current by enough of a margin (or eventually does the same thing as reverse polarization, followed by venting and outgassing). This is probably why certain series did well in Pete's torture tests (ZL, KZE, HE, FC, PW, PA, FM, LXZ) and certain series, not so well (YXG, YXH, HD, KZH, KZG). I wonder if Pete actually meant KZJ and not KZG as I'm not aware of any KZG coming in 12.5mm wide diameter cans, only KZJ (Pete stated in another post that his torture tests only applied to 12.5mm wide parts and bigger). I do recall Pete saying a while ago that he torture tested KY as well - I'd be curious to know how they fared (those tests ran the parts at 7x-10x the rated ripple current, but at room temperature - as stated before, at lower temperatures, it's possible to conduct more ripple current in electrolytics).

I suppose it's a bit early to take a call on reliability, but Samyoung seem to have some claimed 5000 hour, 105c, moderate/high ripple-current, high-voltage electrolytics. I have some Samyoung NFK 47uf/160V in hand for pre-emptive retrofits on cheap Chinese LED bulbs - the exterior build quality of Samyoung is excellent, comparable to or better than NCC. We'll see how it goes on long-term reliability inside a warm/hot 18W LED bulb in a few months/years - I suspect that the SMD LEDs or the voltage-dropping 400V CBB21/22 film capacitors will fail before the electrolytics.

There are some reports of failed Samyoungs on this board and the internet. The failed series thus far are LXV, NXB (equal to KZE), KMF, and NXC (equal to KZG). There might have been one more account of a failed GP Samyoung capacitor here, but I can't recall which series at the moment. Samyoung is certainly better than Samwha (which is actually Nichicon's Korean subsidiary, ironically enough), probably about the same as Taicon and possibly better than OST and LTEC, etc.

One thing's for sure, each and every company has had their "large bad batch" of capacitors (even if it's isolated to select series), except maybe Rubycon and Panasonic (unless you count the issue with FMs wherein one in 50,000 might be sleeved wrong from the factory), although Rubycon's obsoleted, ultra low ESR and highly aqueous series such as MCZ and MFZ certainly have a lower phase or dissolution temperature than their other series. Every brand had an issue with electrolytics produced 20-30 years ago, those notorious for leaking from the bottom seal over time as the "quaternary ammonium salt" base electrolyte did not have a stable pH balance, and slowly turned from an acidic to a strong alkali, rapidly expediting the degradation of the rubber bung. This is related to the unstable pH balance of some Taiwanese/Chinese brands stated afore.