Capacitor plague

Failed aluminum electrolytic capacitors with open vents in the top of the can, and visible dried electrolyte residue (reddish-brown color)

The capacitor plague was a problem related to a higher-than-expected failure rate of liquid electrolyte aluminum electrolytic capacitors between 1999 and 2007, especially brands from some Taiwanese manufacturers.^[1][2] The capacitors failed prematurely due to an electrolyte with incorrect composition which caused corrosion accompanied by gas generation, often rupturing the case of the capacitor from the build up of pressure.

High failure rates occurred in many well-known brands of electronics, and was particularly evident in motherboards, video cards, and power supplies of personal computers, leading to premature failure of these devices.

History

First announcements

Faulty capacitors have been a problem since their development, but the first flawed capacitors linked to Taiwanese raw material problems were reported by an industry publication in September 2002.^[1] Shortly thereafter, two mainstream electronics journals began to report widespread prematurely defective capacitors from Taiwanese manufacturers in motherboards.^[3][4]

These publications informed engineers and other technically interested specialists without much public exposure. However, that changed when Carey Holzman published his experiences about "leaking capacitors" in the Overclock performance community.^[5]

Visual capacitor symptoms

Closeup of a broken electrolytic capacitor vent and dried electrolyte residue

When opening a broken electronic device, the failed capacitors can easily be recognized with clearly visible fault symptoms which include the following:^[6]

• Bulging of the vent on top of the capacitor. (The "vent" is shaped by an impression stamped into the top of the can, forming the seams of the vent. It is designed so that if the capacitor becomes pressurized it will split at the vent's seams, relieving the pressure rather than exploding.)
• Broken or cracked vent on top of the capacitor, often accompanied with visible crusty rust-like brown or red dried electrolyte deposits.
• Capacitor casing sitting crooked on the circuit board, as the bottom rubber plug is pushed out. In this case sometimes electrolyte may leaked onto the motherboard from the base of the capacitor, visible as dark brown or black surface deposits on the PCB.^[7] However, the leaked electrolyte can be confused with thick elastic glue sometimes used to secure the capacitors to the board if vibration or shock occurs during transportation. A dark brown or black crust up the side of a capacitor is invariably glue, not electrolyte. The glue is itself harmless.

Visible symptoms of failed electrolytic capacitors

Failed Chhsi capacitor with crusty electrolyte buildup on the top  Failed capacitors next to CPU motherboard socket  Failed Tayeh capacitors which have vented subtly through their aluminum tops  Failed electrolytic capacitors with swollen can tops and expelled rubber seals, dates of manufacture "0106" and "0206" (June 2001 and June 2002)  Failed capacitor has exploded and exposed internal elements, and another has partially blown off its casing  Failed Choyo capacitors (black color) which have leaked brownish electrolyte onto the motherboard

Public attention

Results of fire on a printed circuit board, caused by leaked electrolyte which short-circuited conductors carrying power

The news from the Holzman publication spread quickly on the Internet and in public newspapers, partially due to the spectacular nature of the images of the failures — bulged or burst cans, expelled sealing rubber and leaking electrolyte on countless circuit boards. A lot of PC users were affected, and caused an avalanche of reports and comments on thousands of blogs and other website communities.^[4][8][9]

The massive spread of this incident also resulted in many misinformed users and blogs posting pictures of capacitors that had failed due to reasons other than the electrolyte-based capacitor plague.^[10]

Prevalence

Appearance

Most of the affected capacitors failed in the early to middle years of the first decade of the 2000s, from 2002 to 2005. They were produced in the year 1999 to 2003. Problems with capacitors produced with an incorrectly formulated electrolyte have affected equipment manufactured up to at least 2007.^[2]

Major vendors of motherboards such as Abit,^[11] IBM,^[1] and Dell,^[12] Apple, HP, and Intel^[13] are affected by capacitors with faulty electrolyte.

In 2005 Dell spent some US$420 million replacing motherboards outright and on the logistics of determining whether a system was in need of replacement. ^[14][15]

Many other equipment manufacturers have unknowingly assembled and sold boards with faulty capacitors. This caused the "capacitor plague" to appear around the world in all kinds of devices. Because not all manufactures had offered recalls or repairs, detailed repair instructions for self-help was established and can be found on the Internet.^[16][17]

Common characteristics

The non-solid aluminum electrolytic capacitors involved in the case of using an improperly formulated electrolyte mostly belong to the so-called "low ESR", "low impedance" or "high ripple current" e-cap series. The advantages of e-caps using an electrolyte composed of 70% water or more, are in particular a lower ESR, which allows a higher ripple current and results in a lower price.^[18] Water is the least costly material in a capacitor.

Electrical symptoms

The electrical characteristics of a failed electrolytic capacitor with an open vent are the following:

• capacitance value decreases to below the rated value
• ESR increases to very high values.

Electrolytic capacitors with an open vent are in the process of drying out, regardless of whether they have good or bad electrolyte. They always show low capacitance values and very high ohmic ESR values. Dry e-caps are therefore electrically useless.

E-caps can fail without any visible symptoms. Since the electrical characteristics of electrolytic capacitors are the reason for their use, these parameters must be tested with instruments to definitively decide if the devices have failed. But even if the electrical parameters are out of their specifications, the assignment of failure to the electrolyte-problem is not a certainty.

Non-solid aluminum electrolytic capacitors without visible symptoms, which have improperly formulated electrolyte, typically show two electrical symptoms:

• relatively high and fluctuating leakage current^[19][20]
• increased capacitance value, up to twice the rated value, which fluctuates after heating and cooling of the capacitor body

Premature Failure

All electrolytic capacitors with non-solid electrolyte age over time due to evaporation of the electrolyte. The capacitance usually decreases and the equivalent series resistance (ESR) usually increases. The capacitance may normally degrade to as low as 70% of the rated value and the ESR may increase to twice the rated value over the normal life span of the component before it should be considered as a "degradation failure".^[21][22]

The "normal" lifespan of a non-solid electrolytic capacitor of consumer quality is roughly 6 years for a 2000 h/85 °C specification capacitor continuously operating at 40 °C. It can be more than 10 years for a 1000 h/105 °C capacitor also working at 40 °C. However electrolytic capacitors that operate at a lower continuous temperature can have considerably longer lifespan. The "life" of an electrolytic capacitor with defective electrolyte can be as little as two years. It may fail prematurely after reaching approximately 30% to 50% of the expected lifetime.

Responsibility

In the November/December 2002 issue of the industrial journal PCI, which had already published in its September/October issue the story about defective electrolyte,^[1] reported that some large Taiwanese manufacturers of electrolytic capacitors were denying responsibility for defective products.^[23]

However, while industrial customers had confirmed the failures, they were not able to trace the source of the faulty components. The defective capacitors are marked with previous unknown brands like "Tayeh", "Choyo" or "Chhsi" ^[24] which have no means of identification. The marks had not been linked to a company or product brand. Other failed e-caps with well known brands may have had failures not related to the defective electrolyte.

The motherboard manufacturer ABIT Computer Cor. was the only one that had publicly admitted that defective capacitors obtained from Taiwan capacitor makers were used in its products.^[23] However, the company would not reveal the name of the capacitor maker that supplied the tainted products.

Non-solid aluminum electrolytic capacitors

The first electrolytic capacitor developed was an aluminum electrolytic capacitor with a liquid electrolyte, invented by Charles Pollak in 1896. Modern electrolytic capacitors are based on the same fundamental design. After roughly 120 years of developments this component is used in billions of inexpensive and reliable (within their specified life span) capacitors used for electronic devices.

Basic construction

Basic construction details of non-solid aluminum electrolytic capacitors
Construction of a typical single-ended aluminum electrolytic capacitor with non-solid electrolyte  Closeup cross-section diagram of electrolytic capacitor, showing capacitor foils and oxide layers

Aluminum electrolytic capacitors with non-solid electrolyte are generally called "electrolytic capacitors" or "e-caps". They consist of two strips of aluminum foil, separated mechanically by a paper spacer, which is saturated with a liquid or gel-like electrolyte. One of the aluminum foil strips, called the anode, is chemically etched (roughened) to increase the surface area. Then it's oxidized (formed). The very thin oxide layer on the anode surface is an electrical insulator and serves as the dielectric of the capacitor. The liquid electrolyte, which is the cathode of the electrolytic capacitor, covers the etched roughened surface of the oxide layer on the anode perfectly and makes the increased anode surface effectual. This increases the effective capacitance.

The second aluminum foil strip, called the "cathode foil", serves to make electrical contact with the electrolyte. The spacer separates the foil strips to avoid direct metallic contact which would produce a short circuit. Lead wires are attached to both foils which are then rolled with the spacer into a wound cylinder which will fit inside an aluminum case or "can". The winding is impregnated with liquid electrolyte. This provides a reservoir of electrolyte to extend the lifetime of the capacitor. After electrolyte impregnation, the assembly is inserted into an aluminum can and sealed with a plug. Aluminum electrolytic capacitors with non-solid electrolyte have grooves in the top of the case which are designed to split open in the event of excessive gas pressure caused by heat, short circuit or failing electrolyte.

Forming, aluminum oxide dielectric

View onto the structures of a low-voltage anode foil
Cross-section side view of etched 10 V low voltage anode foil  SEM image of the rough anode surface of an unused electrolytic capacitor, showing the openings of pores in the anode  Ultra-thin-cross-section of a etched pore in a low-voltage anode foil, 100,000-fold magnification, light grey: aluminum, dark grey: amorphous aluminum oxide, light: pore, in which the electrolyte is active

The aluminum foil used in non-solid aluminum electrolytic capacitors must have a purity of 99.99%. The foil is roughened by electrochemical etching to enlarge the effective capacitive surface. This etched anode aluminum foil has to be oxidized (called "forming"). Forming creates a very thin oxide barrier layer on the anode surface. This oxide layer is electrically insulating and serves as the dielectric of the capacitor. The forming takes place whenever a positive voltage is applied to the anode, and generates an oxide layer whose thickness varies according to the applied voltage. This electrochemical behavior explains the self-healing mechanism of non-solid electrolytic capacitors.

The normal process of oxide formation or self-healing is carried out in two reaction steps.^[25] First, a strongly exothermic reaction transforms metallic aluminum (Al) into aluminum hydroxide, Al(OH)₃:

2 Al + 6 H₂O → 2 Al(OH)₃ + 3 H₂ ↑

This reaction is accelerated by a high electric field and by high temperatures, and is accompanied by a pressure buildup in the capacitor housing, caused by the released hydrogen gas. The gel-like aluminum hydroxide Al(OH)₃ (also called alumina trihydrate (ATH), aluminic hydroxide, aluminum(III) hydroxide, or hydrated alumina) is converted, via a second reaction step (usually slowly over a few hours at room temperature, more rapidly in a few minutes at higher temperatures), into the amorphous or crystalline form of aluminum oxide, Al₂O₃:

2 Al(OH)₃ → 2 AlO(OH) + 2 H₂O → Al₂O₃ + 3 H₂O

This oxide serves as dielectric and also protects the capacitor from the aggressive reactions of metallic aluminum in the presence aggressive parts of the electrolyte. The problem for forming or self-healing processes in non-solid aluminum electrolytics is that on the one hand, that the electrolyte has to deliver enough oxygen to generate the oxide layer, and water in this case is the best to do that; on the other hand, water is very aggressively corrosive against unprotected aluminum.

Electrolytes

The name "electrolytic capacitor" derives from the electrolyte, the conductive liquid inside the capacitor. As a liquid it can conform to the etched and porous structure of the anode and the grown oxide layer, and form a "tailor-made" cathode.

From an electrical point of view the electrolyte in an electrolytic capacitor is the actual cathode of the capacitor and must have good electrical conductivity, which is actually ion-conductivity in liquids. But it is also a chemical mixture of solvents with acid or alkali additives,^[26] which must be non-corrosive (chemically inert) so that the capacitor, whose inner components are made of aluminum, remains stable over its expected lifetime. In addition to the good conductivity of operating electrolytes, there are other requirements, including chemical stability, chemical compatibility with aluminum, and low cost. The electrolyte should also provide oxygen for the forming processes and self-healing. This diversity of requirements for the liquid electrolyte results in a broad variety of proprietary solutions, with thousands of patented electrolytes.

Up to the mid 1990s electrolytes could be roughly placed into two main groups:

• electrolytes based on ethylene glycol and boric acid. In these so-called glycol or borax electrolytes, an unwanted chemical crystal water reaction occurs according to the reaction "acid + alcohol gives ester + water". These borax electrolytes have been standard in electrolytic capacitors for a long time, and have a water content between 5 and 20%. They working up to a maximum temperature of 85 °C or 105 °C in the voltage range up to 600 V.^[27]
• almost anhydrous electrolytes based on organic solvents, such as Dimethylformamide (DMF), Dimethylacetamide (DMA), or γ-butyrolactone (GBL). These capacitors with organic solvent electrolytes are suitable for temperature ranges from 105 °C, 125 °C or 150 °C, have low leakage current values and have a very good long-term behavior of the capacitors.

It was known, that water is a very good solvent for low ohmic electrolytes, however the corrosion problems linked to water hinder up to that time the use of it in larger amount than 20% as a part of electrolytes. However, the water driven corrosion in electrolytic capacitors using the above-mentioned electrolytes are kept under control with chemical inhibitors which stabilize the oxide layer.^{[28][29][30][31]}

The water problem in non-solid aluminum electrolytic capacitors

The aluminum oxide layer in the electrolytic capacitor is resistant to chemical attacks, as long as the pH value of the electrolyte is in the range of pH 4.5 to 8.5.^[32] However, the pH value of the electrolyte is ideally about 7 (neutral). Measurements of leakage current, which were carried out as early as the 1970s have shown that the leakage current is increased due to chemically induced defects when the pH value deviates from this ideal value.^[33] It is known, that water is very aggressive against aluminum and can chemically initiate defects. It is further known, that unprotected aluminum oxide dielectrics can be slightly dissolved into alkaline electrolytes^[34] weaken the oxide layer thickness.

The fundamental issue of water-containing electrolyte systems lies in the control of aggressiveness of the water towards metallic aluminum. This issue has dominated the development of electrolytic capacitors over many decades.^[35] The first commercially used electrolytes in the mid-twentieth century were mixtures of ethylene glycol and boric acid. But even these glycol electrolytes had an unwanted chemical water-crystal reaction, according to the scheme: "acid + alcohol" → "ester + water". Thus, even in the first apparently water-free electrolytes, esterification reactions could generate a water content of up to 20 percent. These electrolytes had a voltage-dependent life span, because at higher voltages the leakage current based on the aggressiveness of the water would increase exponentially, and the associated increased consumption of electrolyte would lead to a faster drying out.^[21][22] Otherwise the electrolyte has to deliver the oxygen for self-healing processes, and water is the best chemical substance to do that.^[18]

Water driven corrosion - Aluminum hydroxide

Attempt at a pictorial representation of the formation of aluminum hydroxide in a pore of a roughened electrolytic capacitor anode foil

It is known that the "normal" course of building a stable aluminum oxide layer by the transformation of aluminum through the intermediate step of aluminum hydroxide can be interrupted by an excessively alkaline or basic electrolyte. For example, alkaline disruption to the chemistry of this reaction results instead in the following reaction:

2 Al (s) + 2 NaOH (aq) + 6 H₂O → 2 Na⁺ (aq) + 2[Al(OH)₄]⁻ (s) + 3 H₂ (g)

In this case, it may happen that the hydroxide formed in the first step becomes mechanically detached from the metallic aluminum surface and will be not be transformed into the desired stable form of aluminum oxide.^[36] The initiating cause of the self-healing process for building a new oxide layer, a defect or a weak dielectric point, remains unmodified, and generated hydrogen gas escape into the capacitor. Then at the weak point a further additional formation of aluminum hydroxide started, which also remain without converting into the stable aluminum oxide. The self-healing of the oxide layer inside the electrolytic capacitor no longer takes place. The reactions do not come to a standstill, since more and more hydroxide in the pores of the anode foil grows and the first reaction step produces more and more hydrogen gas in the can, increasing the pressure.

Scanning electron microscope (SEM) images of different forms of aluminum hydroxide out of failed electrolytic capacitors
Anode surface with grown plaques of aluminum hydroxide  Anode surface with grown beads of aluminum hydroxide

Water-based electrolyte

At the end of the 1990s a third class of electrolytes was developed by Japanese researcher.

• Highly water-containing electrolytes or water-based electrolyte with up to 70% water for so-called "low-impedance", "low-ESR" or "high-ripple-current" electrolytic capacitors with rated voltages up to 100 V ^[18] for low-cost mass-market applications. With this type of electrolyte – not with this manufacturer - the capacitor plague case is connected.

Development of a water-based electrolyte

With the knowledge in mind that water is a very good solvent for electrolytes, some Japanese manufacturers started, beginning of the 1990s, the development of a new, low-ohmic water-based class of electrolytes. The conductivity of water-based electrolytes, compared to electrolytes with organic solvents like GBL, was significantly improved. Water, with its relatively high permittivity of ε = 81, is a powerful solvent for electrolytes. As such, it dissolves salts in high concentration. The high concentration of dissolved salt ions in the electrolyte increases the conductivity. But water will react quite aggressively and even violently with unprotected aluminum. It converts metallic aluminum (Al) via a highly exothermic reaction into aluminum hydroxide ATH (Al(OH)₃). This is accompanied by strong heat and gas development in the capacitor and may even lead to the explosion of the capacitor. Therefore, the main problem in the development of new water-containing electrolytes is to hinder the aggressiveness of the water against aluminum to get capacitors having a sufficiently good long-term stability.

Normally the anode foil is covered with the dielectric aluminum oxide (Al₂O₃) layer which protect the base aluminum metal against the aggressiveness of aqueous alkali solutions. However, some impurities or weak points in the oxide layer always offer the possibility for water driven anodic corrosion reaction forming aluminum hydroxide (ATH). In e-caps using an alkaline electrolyte this ATH will not be transformed into the desired stable form of aluminum oxide. The weak point remains and the anodic corrosion reaction is still ongoing. This process can be interrupted by protective substances in the electrolyte known as inhibitors or passivators.^[31][37] Inhibitors such as chromates, phosphates, silicates, nitrates, fluorides, benzoates, soluble oils and certain other chemicals can reduce the anodic and cathodic corrosion reaction. However, if inhibitors are used in an insufficient amount, they tend to increase pitting.^[38]

The Japanese manufacturer Rubycon was a leader in the development of new water-based electrolyte systems with enhanced conductivity in the late 1990s. After several years of development the researcher around Shigeru Uzawa had found a mixture of inhibitors that suppressed the aluminum hydration. In 1998, Rubycon announced two series, ZL and ZA, of the first capacitors on the market using an electrolyte with a water content of about 40% which were suitable for a wide temperature range from −40 °C (−40 °F; 233 K) to 105 °C (221 °F; 378 K). Later, newer electrolytes were developed to work with water of up to 70% by weight.

The improvement achieved in the conductivity of the new electrolyte can be seen by a comparison of two capacitors, both of which have a nominal capacitance of 1000 µF at 16 V rated voltage in a package with a diameter of 10 mm and a height of 20 mm. The capacitors of the Rubycon YXG series are provided with an electrolyte based on an organic solvent and can attain an impedance of 46 mΩ when loaded with a ripple current of 1400 mA. ZL series capacitors with the new water-based electrolyte can attain an impedance of 23 mΩ with a ripple current of 1820 mA, an overall improvement of 30%.

Other manufacturers, such as NCC,^[39] Nichicon,^[40] and Elna^[41] followed with their own new products a short time later.

The new type of capacitor was called "Low-ESR" or "Low-Impedance", "Ultra-Low-Impedance" or "High-Ripple Current" series in the data sheets. The highly competitive market in digital data technology and high-efficiency power supplies adopted these new components rapidly because of their improved performance. Even better, by improving the conductivity of the electrolyte, capacitors not only can withstand a higher ripple current rating, they are even cheaper to produce since water is very low cost compared to other solvents. Better performance and low cost drove widespread adoption of the new capacitors for high volume products like PCs, LCD screens and power supplies.

Industrial espionage implicated

A major cause of the plague of faulty capacitors was industrial espionage in connection with the theft of an electrolyte formula. A materials scientist working for Rubycon in Japan left the company with the secret electrolyte formula for the ZA and ZL series of Rubycon and began working for a Chinese company. The scientist then developed a copy of this water-based electrolyte. After that, some staff members who defected from the company copied an incomplete version of the formula and began to undersell the pricing of the Japanese manufacturers with this electrolyte to many of the aluminium electrolytic manufacturers in Taiwan.^[1][42] The subsequent electrolyte produced lacked important proprietary ingredients which were essential to the long-term stability of the capacitors^[4][6] and was unstable when packaged in a finished aluminum capacitor. The bad formulation of electrolyte allowed the unimpeded formation of hydroxide and produced hydrogen gas.^[34]

There are no known public court proceedings related to alleged theft of electrolyte formulas. However, one independent laboratory analysis of defective capacitors has shown that many of the premature failures do appear to be associated with high water content and missing inhibitors in the electrolyte, as described below.

Evidence of insufficient composed electrolyte

The situation of unimpeded formation of hydroxide (hydration) and associated hydrogen gas production occurred during "capacitor plague" or "bad capacitors" incidents involving the failure of large numbers of aluminum electrolytic capacitors. This has been demonstrated by two researchers at the University of Maryland who analyzed the failed capacitors.^[34]

These two scientists initially determined by ion chromatography and mass spectrometry, that there really is hydrogen gas present in failing capacitors, which is what leads to bulging of the capacitor's case or bursting the vent. Thus it was proved that the oxidation takes place according to the first step of the formation of aluminum oxide. Because it has been customary in electrolytic capacitors to bind the excess hydrogen with the help of reducing or depolarizing compounds to reduce the resulting pressure, the researchers then searched for compounds of this type. Usually aromatic nitrogen compounds or amines are used for this purpose. Although the above-mentioned analysis methods are very sensitive to detecting such pressure-relieving compounds, no traces of such agents were found within the failed capacitors.

With capacitors in which the internal pressure build-up was so great that the capacitor case was already bulging but the vent had not opened yet, then the pH value of the electrolyte could be measured. The electrolyte of the faulty Taiwanese capacitors was alkaline with pH (7 < pH < 8). Comparable Japanese capacitors on the other hand had an electrolyte with a pH in the acidic range (pH ≈ 4). As it is known that aluminum can dissolve in alkaline liquids, but not in mildly acidic media, then with the electrolyte of the faulty capacitors an energy dispersive X-ray spectroscopy (EDX or EDS) fingerprint analysis was made, which actually detected dissolved aluminum.

To protect the metallic aluminum against the aggressiveness of the water, some phosphate compounds known as inhibitors or passivators can be used to produce long-term stable capacitors with high-aqueous electrolytes. Phosphate compounds are mentioned in patents regarding to electrolytic capacitors with aqueous electrolytic systems.^[43] Since in the investigated Taiwanese electrolytes, phosphate ions were missing and the electrolyte was also alkaline, they evidently lacked any protection against water, and the formation of more-stable alumina oxides was inhibited. Therefore, only aluminum hydroxide was generated.

The results of chemical analysis were reinforced by the measurement of electrical capacitance and leakage current in a long-term test lasting 56 days. Due to the chemical attack, the oxide layer of these capacitors had been weakened, so that after a short time the capacitance and the leakage current increased briefly, before both parameters dropped abruptly upon opening of the vent. The report of Hillman and Helmold proved that the cause of the failed capacitors was a faulty electrolyte mixture used by the Taiwanese manufacturers, which lacked the necessary chemical ingredients to ensure the correct pH of the electrolyte over time, for long-term stability of the electrolytic capacitors. The further conclusion that the electrolyte with its alkaline pH value then had the fatal flaw of continual growth of hydroxide without conversion into the stable oxide, was verified on the surface of the anode foil both photographically and with an EDX-fingerprint analysis of the chemical components.

Literature

• H. Kaesche, Die Korrosion der Metalle - Physikalisch-chemische Prinzipien und aktuelle Probleme, Springer-Verlag, Berlin, 2011, ISBN 978-3-642-18427-7
• C. Vargel, Corrosion of Aluminium, 1st Edition, 02 Oct 2004, Elsevier Science, Print Book ISBN 978-0-08-044495-6, eBook ISBN 978-0-08-047236-2
• W. J. Bernard, J. J. Randall Jr., The Reaction between Anodic Aluminum Oxide and Water, 1961 ECS - The Electrochemical Society
• Ch. Vargel, M. Jacques, M. P. Schmidt, Corrosion of Aluminium, 2004 Elsevier B.V., ISBN 978-0-08-044495-6
• Patnaik, P. (2002). Handbook of Inorganic Chemicals. McGraw-Hill. ISBN 0-07-049439-8.
• Wiberg, E. and Holleman, A. F. (2001). Inorganic Chemistry. Elsevier. ISBN 0-12-352651-5

References