What's creeping around in your data center?
In 1998, the European Union (EU) discovered that alarmingly large amounts of hazardous waste were being dumped into landfill sites. Trends also indicated that the volumes were likely to grow 3-5 times faster than average municipal waste. This highlighted a massive, and growing, source of environmental contamination.
In order to address these issues, the member states of the EU decided to create the Waste Electrical and Electronics Equipment (WEEE, 2002/96/EC) directive, whose purpose was to:
1. Improve manufacturers' designs to reduce the creation of waste,
2. Make manufacturers responsible for certain phases of waste management,
3. Separate collections of electronic waste (from other types of waste), and
4. Create systems to improve treatment, refuse, and recycling of WEEE.
The WEEE directive laid the groundwork for additional legislation and a proposal called EEE (Environment of Electrical & Electronics Equipment) was also introduced along the same lines. However, this policy is generally referred to as the RoHS Directive and is often referred to as "Lead-Free" legislation. This is not a very accurate nickname, because it extends to other pollutants as well.
The European Union (EU) directive 2002/95/EC "on the Restriction of the use of certain Hazardous Substances in electrical and electronic equipment" or RoHS was implemented in July 2006. This directive applies to electrical and electronic equipment designed for use with a voltage rating not exceeding 1,000 volts for alternating current and 1,500 volts for direct current. The requirements of this directive are applicable to the member states of the European Union.
The purpose of the directive is to restrict the use of hazardous substances in electrical and electronic equipment and to contribute to the protection of human health and the environmentally sound recovery and disposal of waste electrical and electronic equipment. The EU's RoHS Directive restricts the use of six substances in electrical and electronic equipment: mercury (Hg), lead (Pb), hexavalent chromium (Cr(VI)), cadmium (Cd), polybrominated biphenyls (PBB) and polybrominated diphenyl ethers (PBDE).
In order to comply with the EU RoHS legislation, all of these substances must either be removed, or must be reduced to within maximum permitted concentrations, in any products containing electrical or electronic components that will be sold within the European Union. Manufacturers have made significant investments in new processes that will eliminate these substances - especially lead.
All applicable products in the EU market must now pass RoHS compliance. In short, RoHS impacts the entire electronics industry and compliance violations are costly - product quarantine, transport, rework, scrap, lost sales and man-hours, legal action, etc. Non-compliance also reflects poorly on brand and image and undercuts ongoing environmental and "due diligence" activities.
WHERE IS ROHS IN EFFECT?
Companies selling a broad range of electrical goods in the EU must now conform to WEEE and those same companies must also conform to RoHS. WEEE and RoHS rules, while laid down at the European level, are put into law at the national level. When exporting to Europe, it is essential to comply with national law in each relevant country. The EU law simply serves as a template for national laws, which may differ considerably. European countries currently requiring conformance with the EU RoHS Directive include: Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, and the United Kingdom. (1)
China RoHS regulations have also been put into effect and many consider them to be considerably more restrictive. As described by a potentially-impacted customer: "Without exemptions, it is impossible to build a compliant board" (Anon 2008)
RoHS regulations are also either in effect or pending in many countries - including the United States. Additional RoHS-like regulations include:
* Argentina has both a WEEE and RoHS bill.
* Japan recently adopted "RoHS" labeling requirements for certain products.
* Taiwan has a voluntary RoHS program.
* Korea is considering RoHS legislation.
* Australia is conducting a survey.
U.S. State bills:
* California SB20 - has the same requirements as EU RoHS except that its scope includes only products with displays larger than 4".
* California AB2202 - includes all EU RoHS products.
* Minnesota - patterned after EU RoHS, however only for consumer products.
* Several other states are looking at RoHS legislation (see Figure 1).
[FIGURE 1 OMITTED]
An aim shared by almost all RoHS legislation is the elimination of lead in electronic products. Thus the main issue for the electronics industry became the use of lead in the manufacture of components and circuit board assemblies.
A printed circuit board, or PCB, is used to mechanically support and electrically connect electronic components using conductive pathways, or traces, laminated onto a non-conductive substrate. Alternative names are printed wiring board (PWB), and etched wiring board. A PCB populated with electronic components is a printed circuit assembly (PCA), also known as a printed circuit board assembly (PCBA).
All PCBs have conducting layers on their surface typically made of thin copper foil. If the copper is left unprotected, it will oxidize and deteriorate. Traditionally, any exposed copper was plated with lead(-based) solder by the hot air solder leveling (HASL) process.
HASL has been working well for many years, is the predominant surface finish used in the industry, and is also the cheapest PCB available. Now RoHS essentially makes PCBs using the HASL process obsolete. Failure modes on other common lead-free PCB finishes such as organic solder preservative (OSP) and electroless-nickel immersion gold (ENIG) make these technologies undesirable. As a result, alternatives such as immersion silver (ImmAg) and organically coated copper (OCC) are currently used as board finishes. Due to inherent processing difficulties with OCC boards, ImmAg boards are quickly becoming the standard PCB finish in the electronics industry (Mazurkiewicz 2006).
Immersion silver would seem to have a bright future under RoHS (Anon B 2008). It is easy to apply to the boards, relatively inexpensive, and usually performs well. While ENIG presently has a larger market share, over the past 12 months more immersion silver process lines have been installed in PCB facilities than any other finish. However, some manufacturers have complained about issues with corrosion. If severe enough, this could lead to shorts and ultimate failure of the board.
The International Society for Automation (ISA) (2) Standard ISA-71.04-1985 (ISA 1985) classifies several levels of environmental severity for electrical and electronic systems: G1, G2, G3 and GX, providing a measure of the corrosion potential of an environment. G1 is benign and GX is open-ended and the most severe.
In a study performed by Rockwell Automation (Anon C 2008) looking at lead-free finishes, four alternate PCB finishes were subjected to an accelerated mixed flowing gas corrosion test. Important findings can be summarized as follows:
1. The immersion gold (ENIG) and immersion silver (ImmAg) surface finishes failed early in the testing. These coatings are the most susceptible to corrosion failures and are expected to be much more susceptible than traditional HASL coatings. The use of these two coatings may make the PCB the weak link with regard to the sensitivities of the electronic devices to corrosion.
2. None of the coatings can be considered immune from failure in an ISA Class G3 environment.
3. The gold and silver coatings could not be expected to survive a mid to high Class G2 environment based on these test results.
A leading world authority on RoHS, ERA Technology, has also reported that, "Recent research has shown that printed circuit boards made using lead-free materials can be more susceptible to corrosion than their tin/lead counterparts" (Anon D 2008). Industry is working diligently to address these concerns but they cannot be addressed overnight.
The Reliability and Failure Analysis group at ERA Technology has diagnosed failures in electronic devices due to interaction with low levels of gaseous sulfides - failures that caused both a financial impact to the manufacturers and safety issues with their customers. Recent work showed that corrosion could occur even with measured hydrogen sulfide levels as low as 0.2[micro]g/[m.sup.3] (0.14 ppb). Another reference describes the formation of a 200 angstrom thick layer of silver sulfide in 100 hours at a concentration of just 100 [micro]g/[m.sup.3] [72 ppb] (Anon E 2008).
CAUSES OF CORROSION
Corrosion can be defined as "deterioration of a substance (usually a metal) because of a reaction with its environment" (ISA 1985). Corrosion of metals is actually a chemical reaction caused primarily by attack of gaseous contaminants and is accelerated by heat and moisture. Rapid shifts in either temperature or humidity cause small portions of circuits to fall below the dew point temperature, thereby facilitating condensation of contaminants. Note that even at low relative humidity (RH) corrosion may occur, depending on the temperature and contaminants present (Henriksen et al. 1991). Relative humidity above 50% accelerates corrosion by forming conductive solutions on a small scale on electronic components. Microscopic pools of condensation then absorb contaminant gases to become electrolytes where crystal growth and electroplating occur. These can cause breaks and shorts in the circuits by either coating a conductive surface with a non-conductive layer or by connecting conductive surfaces (Figure 2). Above 80% RH causes electronic corrosive damage regardless of the levels of contamination.
[FIGURE 2 OMITTED]
More generally, corrosive gases and water vapor coming into contact with a base metal result in the buildup of various chemical reaction products. As the chemical reactions continue, these corrosion products can form insulating layers on circuits which can lead to thermal failure, conductive failure, or short-circuits. Pitting and metal loss can also occur as a result of corrosive attack on metal. Pitting can be described as the formation of pits that penetrate into the metal causing structural damage and thus hindering edge connectors, etc. from completing the circuit. Metal loss is self-explanatory and would cause a conductive failure in the circuit.
Corrosive Gases (ISA 1985)
There are three types of gases that can be considered as prime candidates in the corrosion of data center electronics: acidic gases such as hydrogen sulfide, sulfur and nitrogen oxides, chlorine, and hydrogen fluoride; caustic gases, such as ammonia; and oxidizing gases, such as ozone. Of these, the acidic gases are of particular concern. For instance, it takes only 10 ppb (28.98 [micro]g/[m.sup.3]) of chlorine to inflict the same amount of damage as 25,000 ppb (17.40 mg/[m.sup.3]) of ammonia.
Each site may have different combinations and concentration levels of corrosive gaseous contaminants. Performance degradation can occur rapidly or over many years, depending on the specific conditions at a site. Common sources of corrosive gases are shown in Table 1. Descriptions of common pollutants and a discussion of their contributions to equipment performance degradation follow.
Table 1. Sources of Reactive Environmental Contaminants (ISA 1985) Constituent Symbol Category Common Sources Acetic acid [CH.sub.3]COOH Gas Semiconductor manufacturing, wood and wood products, photo developing Active organic [N.sub.2] Gas Automobile emissions, nitrogen animal waste, vegetable combustion, sewage, wood pulping Ammonia [NH.sub.3] Gas Microbes, sewage, fertilizer manufacture, geothermal steam, refrigeration equipment, cleaning products, reproduction (blueprint) machines Arsine [AsH.sub.3] Gas Semiconductor manufacturing Carbon C Solid Incomplete combustion (aerosol constituent), foundry Carbon monoxide CO Gas Combustion, automobile emissions, microbes, trees, wood pulping Chloride ions C1 Liquid Aerosol content, oceanic processes, ore processing Chlorine, [C1.sub.2], Gas Chlorine manufacture, Chlorine [C1O.sub.2] aluminum manufacture, dioxide paper mills, refuse decomposition, cleaning products Ethylene [C.sub.2][H.sub.2] Gas Fruit, vegetable, cut flower storage & transportation Formaldehyde HCHO Gas Wood products, floor & wall coverings, adhesives, sealants, photo developing, tobacco smoke Halogen HBr, HI Gas Automotive emissions compounds Hydrocarbons HC, THC Gas Automotive emissions, (alcohols, fossil fuel aldehydes, processing, tobacco ketones, organic smoke, water acids) treatment, microbes. Many other sources, both natural and industrial, paper mills Hydrogen HC1 Gas Automobile emissions, chloride combustion, oceanic processes, polymer combustion Hydrogen HF Gas Fertilizer fluoride manufacture, aluminum manufacture, ceramics manufacture, steel manufacture, electronic device manufacture, fossil fuel Hydrogen [H.sub.2]S Gas Geothermal emissions, sulfide microbiological activities, fossil fuel processing, wood pulping, sewage treatment, combustion of fossil fuel, auto emissions, ore smelting, sulfuric acid manufacture Inorganic dust Solid Crystal rock, rock and ore processing, combustion, blowing sand and many industrial sources Mercaptans [S.sub.8], R-SH Gas Foundries, sulfur manufacture Oxides of [NO.sub.x] Gas Automobile emissions, nitrogen fossil fuel combustion, microbes, chemical industry Ozone [O.sub.3] Gas Atmospheric photochemical processes mainly involving nitrogen oxides and oxygenated hydrocarbons, automotive emissions, electrostatic filters Sulfur dioxide [SO.sub.2], Gas Combustion of fossil [SO.sub.3] fuel, auto emissions, ore smelting, sulfuric acid manufacture, tobacco smoke
Active Sulfur Compounds. Active sulfur compounds refers to hydrogen sulfide ([H.sub.2]S), elemental sulfur (S), and organic sulfur compounds such as the mercaptans (R-SH). When present at low ppb levels, they rapidly attack copper, silver, aluminum, and iron alloys. The presence of moisture and small amounts of inorganic chlorine compounds and/or nitrogen oxides greatly accelerate sulfide corrosion. Note, however, that attack still occurs in low relative humidity environments. Active sulfurs rank with inorganic chlorides as the predominant cause of atmospheric corrosion.
Sulfur Oxides. Oxidized forms of sulfur ([SO.sub.2], [SO.sub.3]) are generated as combustion products of fossil fuels and from auto emissions. Low parts per billion levels of sulfur oxides can cause reactive metals to be less reactive and thus retard corrosion. At higher levels, however, they will attack certain types of metals. The reaction with metals normally occurs when these gases dissolve in water to form sulfurous and sulfuric acid ([H.sub.2][SO.sub.3] and [H.sub.2][SO.sub.4]).
Nitrogen Oxides ([NO.sub.X]). Some common sources of reactive gas compounds (NO, [NO.sub.2], [N.sub.2][O.sub.4]) are formed as combustion products of fossil fuels and have a critical role in the formation of ozone in the atmosphere. They are also believed to have a catalytic effect on corrosion of base metals by chlorides and sulfides. In the presence of moisture, some of these gases form nitric acid ([HNO.sub.3]) that, in turn, attacks most common metals.
Inorganic Chlorine Compounds. This group includes chlorine ([Cl.sub.2]), chlorine dioxide ([ClO.sub.2]), hydrogen chloride (HCl), etc., and reactivity will depend upon the specific gas composition. In the presence of moisture, these gases generate chloride ions that, in turn, attack most copper, tin, silver, and iron alloys. These reactions are significant even when the gases are present at low ppb levels. At higher concentrations, many materials are oxidized by exposure to chlorinated gases. Particular care must be given to equipment that is exposed to atmospheres which contain chlorinated contaminants. Sources of chloride ions, such as bleaching operations, seawater, cooling tower vapors, and cleaning compounds, etc., should be considered when classifying industrial environments. They are seldom absent in major installations.
Hydrogen Fluoride (HF). This compound is a member of the halogen family and reacts like inorganic chloride compounds.
Ammonia and Derivatives. Reduced forms of nitrogen (ammonia, [NH.sub.3]), amines, ammonium ions ([NH.sup.4+]) occur mainly in fertilizer plants, agricultural applications, and chemical plants. Copper and copper alloys are particularly susceptible to corrosion in ammonia environments.
Photochemical Species. The atmosphere contains a wide variety of unstable, reactive species that are formed by the reaction of sunlight with moisture and other atmospheric constituents. Some have lifetimes measured in fractions of a second as they participate in rapid chain reactions. In addition to ozone ([O.sub.3]), a list of examples would include the hydroxyl radical as well as radicals of hydrocarbons, oxygenated hydrocarbons, nitrogen oxides, sulfur oxides, and water. Because of the transient nature of most of these species, their primary effect is on outdoor installations and enclosures. In general, metals are only slightly susceptible to photochemical effects. However, ozone can function as a catalyst in sulfide and chloride corrosion of metals.
Strong Oxidants. This includes ozone plus certain chlorinated gases (chlorine, chlorine dioxide). Ozone is an unstable form of oxygen that is formed from diatomic oxygen by electrical discharge or by solar radiation in the atmosphere. These gases are powerful oxidizing agents. Photochemical oxidation - the combined effect of oxidants and ultraviolet light (sunlight) - is particularly potent.
Hydrogen sulfide ([H.sub.2]S), sulfur dioxide ([SO.sub.2]), and active chlorine compounds ([Cl.sub.2], HCl, [ClO.sub.2]), have all been shown to cause significant corrosion in electrical and electronic equipment at concentrations of just a few parts per billion in air. Even at levels that are not noticed by or harmful to humans, these gases can be deadly to electronic equipment. Most of the odor threshold levels are much higher than the levels at which corrosive damage will occur.
THE NATURE OF CORROSION
When discussing electronic equipment, the corrosion of copper, silver, or composite materials gives the same result: a disruption of electrical current flow. The severity of the environment (i.e. the types and levels of gases, humidity, and temperature) will determine the speed at which corrosion forms and how soon corrosion-related effects may appear.
Sign of Corrosion
For the purposes of this paper, corrosion can be thought of as two distinct types: the first being the more conventional corrosive attack where the acid gases react with the metals themselves to form non-conductive salts and second type being "whisker growth".
Creep Corrosion. Any place where a non-precious metal meets the atmosphere, corrosion may occur. Some metals undergo self-limiting, or passive corrosion. Once an oxide or sulfide layer forms, it will not grow any further. The layer of corrosion products effectively insulates the underlying base metal from the environment, and further corrosion is prevented.
In active corrosion, the corrosion layer is not self-limiting. The base metal will continually corrode, and the corrosion product will tend to spread out from its point of origin. Any pre-plated or clad contact (coated before stamping) will have bare edges. These edges are free to corrode, and the corrosion product can grow from the edges and slowly spread across the surface as shown in Figure 3. This is known as "creep corrosion" (Anon F 2008).
[FIGURE 3 OMITTED]
Creep corrosion is a mass-transport process during which solid corrosion products migrate over a surface. Eventually, the corrosion product will interfere with the electrical connection by creating unacceptably high levels of contact resistance. Pollutants such as chlorine, hydrogen sulfide, and sulfur dioxide are known to promote creep corrosion. For components with noble metal pre-plated leadframes, creep corrosion is a potential reliability risk for long-term field applications (Zhao and Pecht 2005).
Pore Corrosion. In an ideal world, surface platings would be uniformly thick, continuous, and stress-free. In the real world, there will always be some imperfections in the plating. In many cases, platings will have pores, through which the base metal will be exposed. The degree of porosity depends on the plating thickness, application method, base metal roughness, and base metal cleanliness. As plating thickness increases, there is less likelihood of a pore extending all the way to the base metal. In a very thin plating layer (on the order of those used for gold), the likelihood is high that many pores will extend all the way through this layer. Platings over base metals with rough surfaces show more porosity than those over smooth surfaces. Dirt or oxide on the surface of the base metal can also lead to the appearance of pores.
If the base metal is exposed to a corrosive environment through these pores, pore corrosion can occur. Base metal corrosion can occur entirely in the pores, where it may be hidden from view, or the corrosion product may creep across the surface (Anon F 2008). An example of this is shown in Figure 4. Pore corrosion is promoted by the many of the same gases that promote creep corrosion.
[FIGURE 4 OMITTED]
Via Corrosion. When a multi-layer circuit board is created, vertical interconnects must be added. These interconnects are called vias and can be defined as a plated-through hole used to connect two or more conductor layers of a multilayer board, in which there is no intention to insert a component lead or other reinforcing material.
There are areas on circuit boards that can contain exposed copper, such as the inside of via barrels. The purpose of surface finish (or coating) is to protect the copper beneath. If the coating is complete, corrosion would not take place as rapidly. An experimental investigation of the via holes on a typical silver immersion board shows that the silver coating does not typically extend all the way through the via barrel (Mazurkiewicz 2006).
In situations where the via hole is not completely coated, copper metal is vulnerable to atmospheric attack (Figure 5). This situation which involves a more noble metal coating a metal that is prone to oxidation in the presence of an electrolyte (atmospheric water) is highly prone to galvanic corrosion. Indeed, this appears to be the case when one reviews the dozens of incidents and hundreds of failures in which PCB technology is used in an environment high in reduced sulfur gasses. The majority of the failures show the heaviest corrosion mainly inside the via holes.
[FIGURE 5 OMITTED]
Whisker Growth. This type of corrosion typically refers to electrically conductive, crystalline structures of tin that sometimes grow from surfaces where tin (especially electroplated tin) is used as a final finish. Tin whiskers have been observed to grow to lengths of several millimeters (mm) and in rare instances to lengths up to 10 mm (0.4 in). Numerous electronic system failures have been attributed to short circuits caused by tin whiskers that bridge closely-spaced circuit elements maintained at different electrical potentials (Anon G 2008).
"Whisker growth" can be described as corrosion in which microscopic metal crystals grow out of the surface of the conductive metal. It is caused by the presence of sulfide molecules, e.g., silver sulfide on a silver surface, which can migrate freely over the metallic surface and collect at dendrite boundaries where nucleation takes place and sulfide crystals grow out of the surface of the metal. These whiskers can be of size long enough to connect portions on a board, chip, etc. and cause short-circuits (Figure 6). By some accounts, whisker growth accounts for approximately 10% of documented failures (Brusse et al. 2007).
[FIGURE 6 OMITTED]
Other specific areas on PCB that are particularly sensitive to corrosion attack are briefly described below.
Edge Connectors. These contacts are typically gold-plated over a nickel-plated copper substrate that is intended to ensure that the electrical contact between the board and the connector is maintained with the highest integrity over a long period of time. Since the gold layer is typically only between 4 and 8 [micro]m thick, and at these plating thicknesses fairly porous, the passage of corrosive gases can occur through the gold layer and attack the underlying layers of nickel and/or copper.
The salts formed by corrosion reactions are of a higher volume than the pure metal and can either "flake off" the gold plating or be forced back up through the pores onto the surface of the gold plating. In both cases the contact arm no longer rests on the conductive metal but rests on a high resistance coating.
Earlier generations of electronic equipment with operating voltages as high as 24-48 volts DC, equipment failures only occurred after many years of operation since these voltages were high enough to break down the thin high-resistance layers caused by salt formation. Modern electronics are, however, much more sensitive to this type of corrosion since operating voltages are lower and cannot break down these corrosion films.
Pin Connectors. The problems experienced on pin connectors and IC plug-in sockets show the same problems as edge connectors. However, edge connectors appear to be more susceptible to corrosive attack and failures occur earlier than on pin connectors and IC sockets.
Wire-wrap Connections. Wire-wrap connection pins are particularly sensitive to corrosive attack because the wire forms a "smear" contact between the pin and the wire with up to four different metallic alloys being exposed to the atmosphere. Corrosive attack can thus occur concurrently with electrolytic action which greatly increases the corrosive influence.
Electrical Systems. Previous generations of electronic equipment contained only heavy current systems in which the build-up of corrosion by-products could cause overheating that result in a reduction in equipment life. Today, however, central offices, internet data centers, and remote locations using the most advanced microelectronics operate with much lower voltages and are much more sensitive to corrosive attack as described above. As such, the air quality in these rooms will have a great effect on the life of the equipment.
There are many methods available to measure ambient air quality and two methods have been used extensively for environmental characterization with regards to corrosion. One is a direct measure of selected gaseous air pollutants. The other, which can be termed "reactivity monitoring," provides a quantitative measure of the overall corrosion potential of an environment.
Pollution analysis may provide short-term estimates for specific sites. High values will confirm that a severe environment exists. The reverse, however, is not necessarily true. Many environments may contain a complex mixture of contaminants that interact to greatly accelerate (or retard) the corrosive action of individual gas species.
As a direct measure of overall corrosion potential, reactivity monitoring involves the placement of specially prepared metal coupons in the operating environments. (3) Analyses may consist of measurements of film thickness, film chemistry, or weight loss. However, ISA Standard 71.04-1985 references measurement of film thickness by cathodic (electrolytic, columeteric) reduction as a preferred method to perform reactivity monitoring. By using this method, the air with contaminant gases can be rated as to its effect on the electronic equipment.
Quantifying the Corrosive Potential of an Environment
Environmental Reactivity Coupons. Since the ISA Standard was published hundreds of thousands these "environmental reactivity coupons," or ERCs, have been used to provide a measure of the corrosive potential of an environment (Figure 7). ERCs may be used to indicate the presence of [SO.sub.2], [NO.sub.2], [Cl.sub.2], [O.sub.3], and many other corrosive materials which can cause damage to PCBs and many types of electronic equipment.
[FIGURE 7 OMITTED]
ERCs originally used only copper reactivity to establish environmental classifications. However, copper is not sufficiently sensitive to many of those pollutants ubiquitous to most urban environments - the same environments in which most data centers are located. Further, copper coupons cannot detect the presence of chlorine, a particularly dangerous contaminant to metals. Finally, with the passage of RoHS legislation, and the switch to Imm-Ag and other lead-free finishes, many control systems manufacturers and end-users are now questioning whether this type of environmental monitoring is adequate or not.
Due to all of this, the use of silver reactivity monitoring, in addition to copper reactivity monitoring, has seen an increase in use in these environments. Although silver is not an equipment reliability determinant in the ISA standard, it has been long been included because of independent studies citing silver being able to detect changes in the levels of gaseous contaminants in the ambient environment as small as 1 ppb and differentiate between different classes of contaminants as well as being a better indicator of environmental chlorine than copper (Rice et al 1981, Muller 1999).
The use of these two types of coupons presented, on occasion, results that were quite surprising. Some environments that were non-corrosive to copper, and thus considered harmless to equipment (by the ISA standard), were extremely corrosive to silver. While this alone does not necessarily indicate an environment requiring direct control of contaminants, it does point out the potential for corrosion-related effects. Many in the process industries (e.g., pulp and paper, petrochemical) feel that using copper-only reactivity monitoring results can seriously understate the potential for equipment failure in these environments (Muller 1990).
The primary reasons for using both copper and silver reactivity monitoring to gauge data center air quality are:
1. Corrosion on silver is not humidity-dependent. In a temperature and humidity-controlled data center, any corrosion that forms on silver can be attributed solely to the presence of chemical contaminants. Copper corrosion is humidity dependent and can therefore be attenuated in these environments.
2. Silver can be used to positively indicate the presence of inorganic chlorine which is an extremely reactive contaminant. Environmental chlorine cannot be definitively determined from copper ERC analysis.
3. Silver is much more sensitive to low levels of contaminants. This is due to a passivating oxide layer that can form on copper. Silver is not reactive with oxygen and, again, any corrosion that forms is due to the presence of reactive contaminant gases.
The corrosion reported from reactivity monitoring with ERCs is actually the sum of individual corrosion films. For copper coupons, sulfide and oxide films are most commonly produced and are reported as copper sulfide ([Cu.sub.2]S) and copper oxide ([Cu.sub.2]O), respectively. For silver coupons, sulfide, chloride, and oxide films may be produced and are reported as silver sulfide ([Ag.sub.2]S), silver chloride (AgCl), and silver oxide ([Ag.sub.2]O), respectively. Each coupon is analyzed as to the type and amount of film present and its relative contribution to the total corrosion produced.
Environmental Reactivity Monitors. One consideration faced in designing an air quality monitoring program is the choice of passive vs. active sampling. The immediate feedback of an active monitor is a most desirable aspect and is what often precludes the use of passive monitors. The main limitation in the use of ERCs is their inability to provide a continuous environmental classification. To address this, reactivity monitoring has been taken a step farther through the development of a realtime monitoring device employing metal-plated quartz crystal microbalances (QCMs, Figure 8). These microprocessor-controlled devices are able to measure the total environmental corrosion attributable to gaseous pollutants. ERMs employing QCMs can detect and record changes <1 ppb for many common contaminants. This ability is regarded as one of the main requirements for any real-time monitoring protocol to be used in data center environments (England et al. 1991, Weiller 1994).
[FIGURE 8 OMITTED]
To date, there are two commercially available ERMs that employ copper and silver-plated QCMs that provide real-time information on the amount of corrosion forming due to the presence of gaseous contaminants. These devices measure film thickness similar to the ERCs and report cumulative and incremental corrosion rates in angstroms, correlating to the reporting requirements of ISA 71.04-1985.
One device measures corrosion on a continuous basis and allows for preventive action to be taken before serious damage has occurred. Appropriate reactivity and alarm levels for a particular application can be adjusted. It can be operated independently as a battery-operated data logger unit or operated as a data transmitter wired directly into a facility monitoring system (FMS). Environmental classification databases can be established and maintained to provide historical data.
The second device is similar in that it employs QCMs to monitor the environment in real-time. However, it does not store data but transmits it directly to an FMS or computer directly via a 4-20 mA output signal.
Classification of Reactive Environments (ISA 1985)
The classification of reactive environments per ISA Standard 71.04 is based on the thickness of the reaction film(s) formed on the copper coupon when determined by cathodic reduction. The total film thickness is measured in angstroms ([Angstrom]) normalized to a one month exposure and correlated to severity levels that predict effects of corrosion on electronic equipment reliability. This classification scheme also provided approximations of the concentrations of gases in the environment being analyzed. The classification of reactive environments is shown in Table 2 and described and illustrated below (Figure 9).
[FIGURE 9 OMITTED]
Table 2. ISA Classification of Reactive Environments * Severity Level G1 G2 G3 GX Severe Mild Moderate Harsh Copper Reactivity Level <300 <1000 <2000 [greater than or [(in angstroms).sup.[section]] equal to]2000 * The gas concentration levels shown below are provided for reference purposes. They are believed to approximate the Copper Reactivity Levels stated above, providing the relative humidity is less than 50%. For a given gas concentration, the Severity Level (and Copper Reactivity Level) can be expected to be increased by one level for each 10% increase in relative humidity above 50% or for a relative humidity rate of change greater than 6% per hour. Gas Concentrations (in ppb) Contaminant Gas Reactive Species **, Group A [H.sub.2]S <3 [dagger][dagger] [SO.sub.2], [SO.sub.3] <10 [C1.sub.2] <1 [NO.sub.X] <50 Group B HF <1 [NH.sub.3] <500 [O.sub.3] <2 Gas Concentrations (in ppb) Contaminant Concentration Reactive Species **, Group A <10 <50 50 [dagger][dagger] <100 <300 300 <2 <10 10 <125 <1250 1250 Group B <2 <10 10 <10,000 <25,000 25,000 <25 <100 100 [section] Measured in angstroms after one month's exposure. ** The Group A contaminants often occur together and the reactivity levels include the synergistic effects of these contaminants. [dagger][dagger] The synergistic effects of Group B contaminants are not known at this time.
Severity Level G1--Mild. An environment sufficiently well-controlled such that corrosion is not a factor in determining equipment reliability.
Severity Level G2--Moderate. An environment in which the effects of corrosion are measurable and may be a factor in determining equipment reliability.
Severity Level G3--Harsh. An environment in which there is a high probability that corrosive attack will occur. These harsh levels should prompt further evaluation resulting in environmental controls or specially designed and packaged equipment.
Severity Level GX--Severe. An environment in which only specially designed and packaged equipment would be expected to survive. Specifications for equipment in this class are a matter of negotiation between user and supplier.
This classification scheme allows for the evaluation of a subject environment with respect to its corrosion potential and ultimately the reliability of electronic equipment being used in the environment. From this a decision can be made as to whether or not action is needed to improve the environment for the protection of the electronic equipment.
SO WHAT DOES ALL THIS MEAN?
Corrosion-induced failures are frequent in electronics products used in industrial environments. A typical failure mechanism of electronic systems in these environments is the reaction of atmospheric sulfur with exposed metals - particularly copper and silver. These metals are found in PCB traces, integrated circuit (IC) leads and device terminations. Copper sulfide or silver corrosion products can grow and creep across surfaces such as IC packages and PCB substrates.
Historically, the use of silver in electronic assemblies has been a reliability risk unless the silver is protected from the environment. Creep corrosion (electromigration) can occur quite readily in humid environments especially in the presence of small amounts of atmospheric sulfur and chlorides which are common in many industrial and other environments.
Manufacturers of industrial computer equipment specify the control of corrosive gases in their site planning guides and/or their terms and conditions to maintain warranties and service contracts. Now even environments previously considered benign with regards to electronics corrosion are experiencing serious problems as a direct result of RoHS compliance. Data centers in many urban locations have reported failures of servers and hard disk drives due to sulfur corrosion. There are hints that this may even trickle down into personal computers and electronic devices.
Many millions of lead-free desktop and notebook systems, and related peripherals, were put into service by Dell Computers starting in 2005, a year prior to the effective date of the RoHS legislation and a large amount of resources were invested in optimizing and qualifying lead-free assemblies. Field data showed the overall quality of Dell's lead-free products to be as good, and in some cases, better than the previous generation of tin-lead products (Schueller 2007).
However, one failure mechanism caused by the lead-free transition was not foreseen by Dell or the industry. It was soon discovered that lead-free products with ImmAg surface finish will creep corrode in high sulfur environments (ISA Class G2 or higher). The majority of creep corrosion failures occurred on hard disk drives (HDD), graphic cards, and motherboards in desktop or workstation systems (only those with ImmAg PCB finish were affected).
Echoing the study by Rockwell, Hewlett Packard (Mazurkiewicz 2006) documented several case studies where creep corrosion of ImmAg finish caused failures of computer systems in high sulfur environments. It was stated that sulfur-based corrosion failures increased dramatically upon introduction of ImmAg surface finish on computer products (due to ROHS requirements). Alcatel-Lucent also has experience with this issue in a paper detailing their work with mixed flowing gas testing of various lead-free surface finishes and their resistance to creep corrosion (Reid et al. 2007).
Analysis revealed the corrosion products to be copper sulfide ([Cu.sub.2]S) and silver sulfide ([Ag.sub.2]S). Other studies have shown that high amounts of [Cu.sub.2]S typically indicate the presence of active sulfur compounds such as elemental sulfur, [H.sub.2]S, or organic sulfur compounds. [Ag.sub.2]S can also be formed by these contaminants but can also be formed by exposure to [SO.sub.2] contamination - a primary urban contaminant.
The most common failures are with the most common components, with the highest incidence of failures being found in capacitors, plastic encapsulated microcircuits (PEMs) and printed circuit boards (PCBs). A University of Maryland study of field failures from 70 companies revealed 30% of ALL failures were capacitors--mainly MLCCs [multilayer chip capacitors] (Anon H 2008).
CORROSION OF DATACOM EQUIPMENT
The reason for the dramatic increase in corrosion failures due to the use of ImmAg involves the propensity of silver and copper to react with sulfur. This is further complicated by the fact that there are areas on ImmAg circuit boards that contain both exposed silver and copper in contact, such as the inside of via barrels. The purpose of the silver coating is to protect the copper beneath. If the coating was complete, corrosion would not take place as rapidly (Mazurkiewicz 2006).
ERCs and ERMs have been used to evaluate the severity levels of data center environments all over the world. It has been found that many areas have levels of corrosive contaminants high enough to be harmful to electronic equipment performance to the point of failure. Examples of reactivity monitoring data collected from data centers with documented cases of corrosion-related equipment failure (PCBs, servers, hard disk drives, etc.) is shown in Table 3 (Purafil 2008).
Table 3. Reactivity Monitoring Results for Data Centers Copper Corrosion Type Location [Cu.sub.2]S [Cu.sub.2]O Bank Inside DC 0 75 FSC Server Room 122 * 70 TCC Inside DC 162 * 105 Bank Inside DC 193 * 94 Bank DC supply air 0 288 Bank Under raised floor 258 * 89 Bank Data storage 265 * 91 HQ Inside DC 299 * 80 Bank Inside DC 334 * 75 FSC Inside DC 209 * 240 FSC Outside air 355 * 137 Bank DC supply air 426 * 85 Bank Inside DC 415 * 179 TCC Server room 474 * 143 Bank Under raised floor 587 * 180 FSC # Hub room # 1,702 *# 320 # HQ # Outside air # 620 *# 1,549 # FSC # Hub room # 1,595 *# 686 # IC # Outside air # 1,749 *# 268 # Bank # Outside air # 0 *# 258 # FSC # Hub room # 3,297 *# 778 # HQ # Outside air # 6,855 *# 6,794 # HQ # Outside air # 10,357 *# 7,434 # TCC # Inside DC # 20,000 *# 0 # ISP # Server room # 20,000 *# 0 # Bank # Outside air # 20,000 *# 0 # Bank # Outside air # 20,000 *# 0 # FSC # Hub room # 20,000 *# 0 # FSC # Hub room # 20,000 *# 0 # FSC # Hub room # 20,000 *# 0 # ISP # Hub room # 20,000 *# 0 # Copper Corrosion Type Location Cu-Unk Total ISA Level Bank Inside DC 0 75 G1 FSC Server Room 0 192 G1 TCC Inside DC 0 267 G1 Bank Inside DC 0 287 G1 Bank DC supply air 0 288 G1 Bank Under raised floor 0 347 G2 Bank Data storage 0 356 G2 HQ Inside DC 0 379 G2 Bank Inside DC 0 409 G2 FSC Inside DC 0 449 G2 FSC Outside air 0 492 G2 Bank DC supply air 0 511 G2 Bank Inside DC 0 594 G2 TCC Server room 0 617 G2 Bank Under raised floor 0 767 G2 FSC # Hub room # 0 # 2,022 # GX # HQ # Outside air # 0 # 2,169 # GX # FSC # Hub room # 0 # 2,281 # GX # IC # Outside air # 478 # 2,495 # GX # Bank # Outside air # 2,315 # 2,573 # GX # FSC # Hub room # 0 # 4,075 # GX # HQ # Outside air # 0 # 13,649 # GX # HQ # Outside air # 0 # 17,792 # GX # TCC # Inside DC # 0 # 20,000 # GX # ISP # Server room # 0 # 20,000 # GX # Bank # Outside air # 0 # 20,000 # GX # Bank # Outside air # 0 # 20,000 # GX # FSC # Hub room # 0 # 20,000 # GX # FSC # Hub room # 0 # 20,000 # GX # FSC # Hub room # 0 # 20,000 # GX # ISP # Hub room # 0 # 20,000 # GX # Silver Corrosion Type Location AgCl [Ag.sub.2]S Ag-Unk Total Bank Inside DC 0 # 636 *# 0 # 636 # FSC Server Room 0 524 0 524 TCC Inside DC 0 # 982 *# 0 # 982 # Bank Inside DC 0 # 785 *# 0 # 785 # Bank DC supply air 23 *# 227 # 0 # 250 # Bank Under raised floor 0 # 1,047 *# 0 # 1,047 # Bank Data storage 0 # 3,200 *# 0 # 3,200 # HQ Inside DC 0 # 1,603 *# 0 # 1,603 # Bank Inside DC 0 # 1,833 *# 0 # 1,833 # FSC Inside DC 0 106 0 106 FSC Outside air 135 *# 353 # 0 # 488 # Bank DC supply air 0 # 861 *# 0 # 861 # Bank Inside DC 0 # 1,122 *# 0 # 1,122 # TCC Server room 0 # 852 *# 0 # 852 # Bank Under raised floor 45 *# 1,742 *# 0 # 1,787 # FSC # Hub room # 0 # 916 *# 0 # 916 # HQ # Outside air # 0 465 0 465 FSC # Hub room # 0 # 1,309 # 0 # 1,309 # IC # Outside air # 835 *# 96 # 96 # 1,027 # Bank # Outside air # 87 552 0 639 FSC # Hub room # 0 # 1,440 *# 0 # 1,440 # HQ # Outside air # 0 # 5,044 *# 2,421 # 7,465 # HQ # Outside air # 0 # 1,651 *# 2,631 # 4,282 # TCC # Inside DC # 0 # 655 *# 0 # 655 # ISP # Server room # 0 # 810 *# 0 # 810 # Bank # Outside air # 387 *# 374 # 715 # 1,476 # Bank # Outside air # 0 # 2,855 *# 0 # 2,855 # FSC # Hub room # 0 # 3,404 *# 0 # 3,404 # FSC # Hub room # 0 # 3,535 *# 0 # 3,535 # FSC # Hub room # 0 # 3,927 *# 0 # 3,927 # ISP # Hub room # 0 # 4,114 *# 0 # 4,114 # Key: FSC = financial services company, HQ = company headquarters, IC = investment company, ISP = internet service provider, TCC = telephone call center Note: Gray-shaded areas indicate # locations where the total copper and/or silver corrosion rates exceed those recommended for data centers. Bold text indicates * the presence of active sulfur contamination (as [Ag.sub.2]S) and/or inorganic chlorine compounds (as AgC1), either of which indicate a higher corrosion potential.
When interpreting the analysis results for the individual corrosion films, the detection of a silver sulfide ([Ag.sub.2]S) film without a corresponding copper sulfide ([Cu.sub.2]S) film usually indicates the presence of oxidized forms of sulfur such as [SO.sub.2] and [SO.sub.3]. When both films are detected - as is the case here -it more often than not indicates the presence of active sulfur compounds such as elemental sulfur, [H.sub.2]S, and organic sulfur compounds (e.g., mercaptans) as well. When both films are present and the amount of [Cu.sub.2]S is greater than 50% of the total corrosion, this is further evidence of the presence of active sulfur compounds in the subject environment.
Silver chloride (AgCl) indicates the presence of (an) inorganic chlorine compound(s), e.g., [Cl.sub.2], [ClO.sub.2], HCl.
The highlighted values in Table 3 under the [Cu.sub.2]S and AgCl columns indicate an increased risk for corrosive attack regardless of the ISA Severity Level indicated. The shaded rows are locations where computer equipment would not be expected to survive without some type of additional environmental control, i.e., air cleaning to remove corrosive gases.
WHERE DO WE GO FROM HERE?
The requirement for corrosion control in industrial environments remains constant. However, more companies are now taking a much closer look at developing or updating specifications due to the changes made by controls manufacturers to comply with the RoHS restrictions on the use of lead. This includes specifying an ISA Class G1 environment for control rooms, etc., where in the past a Class G2 environment was considered acceptable. Specifications are also now showing up requiring the measurement and quantification of silver corrosion rates and related corrosion products.
With a majority of control systems, manufacturers opting to use the ImmAg process for their PCBs and other electrical components, and the fact that silver is much more sensitive to lower levels of corrosive gases, we have already seen increased concerns over equipment reliability. As one Automation Engineer for an Australian pulp and paper manufacturer described it, "The ISA Standard is becoming irrelevant because it does not take into account silver corrosion. All of the new equipment I have purchased over the last year contains silver and the failure rate for some components is now being measured in weeks instead of years."
Another contributing factor to concerns over the use of silver in industrial applications is that even with tightening the environmental control requirements for temperature and humidity and the positive effect this has on copper corrosion, silver can still exhibit high rates of corrosion in well-controlled environments. Examination of one environmental reactivity coupon (ERC) database has shown that in locations reported as ISA Class G1 for copper corrosion, the corresponding silver corrosion rate can be up to 10x higher. Furthermore, every ERC analyzed shows evidence of sulfur contamination (as silver sulfide, [Ag.sub.2]S). On average, the amount of silver corrosion measured is double that of the copper corrosion reported.
Corrosion control in industrial environments is acknowledged as a requirement to assure electrical and electronic equipment reliability. However, up to now corrosion control in non-industrial applications has only been a concern in isolated instances. Desktop and laptop computers, servers, data communications (datacom) equipment and other equipment used in information technology (IT) applications were not particularly sensitive to the types and concentrations of contaminants present in urban environments. However, even these types of computer equipment are now at risk due to RoHS.
Datacom equipment center owners and operators have traditionally focused most of their attention on the physical structure and performance of the data center infrastructure (e.g. power, cooling, raised access floor equipment). Most IT hardware is designed to operate in a controlled environment. There are few moving parts (air handling equipments, drive motors, etc.) and electronics are not expected to simply "wear out."
High reliability is demanded for today's IT equipment and projections to first failure are more than 11 years of power on operation, which is far longer than in the real world, especially when considering that the equipment becomes obsolete in half that time. However, that high reliability is only if the product is maintained in the optimum environment. Any site climate deviations or deviations in electrical power, static electricity or air quality will have a negative impact on the equipment and failure rates.
Even though today's complex and sensitive IT equipment requires a higher level of environmental control for gaseous contamination than would have been considered before RoHS took effect, this has been one aspect of data center environmental control that was mostly overlooked. Datacom equipment center contamination if left unchecked can influence the reliability and the continuous operation of mission critical IT equipment within a facility. Gaseous contamination can result in intermittent equipment glitches, unplanned shutdowns, or failure of critical systems that often result in significant business and financial loss. Harsh environments are also not always obvious or restricted to industrial settings.
Contaminant gases containing sulfur, such as [SO.sub.2] and [H.sub.2]S, are the most common gases in data centers causing hardware corrosion. One example of component failure was due to sulfur gases entering a component package and attacking the silver resulting in the formation of [Ag.sub.2]S. The mechanical pressure created by the [Ag.sub.2]S formation inside the package damaged its mechanical integrity and caused the device to fail. This and other failure mechanisms are becoming common occurrences within data centers using RoHS-compliant equipment and components produced using the ImmAg process, and to a lesser degree, the ENIG process.
To maintain a high level of equipment dependability and availability, it should be understood that a data center is a dynamic environment where many maintenance operations, infrastructure upgrades, and equipment change activities occur on a regular basis. Airborne contaminants harmful to sensitive electronic devices can be introduced into the operating environment in many ways (e.g., chlorine can be emitted from PVC insulation use on wires and cables if temperatures get too high.) However, the outdoor ambient air that is used for both cooling and pressurization is often the primary source of corrosive contaminants and should be cleaned before its introduction into the data center environment. And now with the rediscovery of "free cooling" using air-side economizer technology, more outside air will be brought into the data center and along with it higher levels of contamination than previously encountered. Loosening control limits for humidity will also contribute an increased potential for corrosion-related problems.
With the changes to IT equipment due to the RoHS directives, data center designers, managers, and operators should include an environmental contamination monitoring and control section as part of an overall site planning, risk management, mitigation, and improvement plan.
Included in this plan should be considerations for the assessment of the outdoor air and data center environment with regards to corrosion potential. Even though ISA 71.04 is considered a standard for industrial applications, it still has relevance in datacom applications and can be used to provide site-specific data on the types and levels of gaseous contamination. ERCs can be used as a survey tool to establish baseline data necessary to determine if and what type of environmental controls are needed.
The second part of the plan should be the development and specification of a specific contamination control strategy. Corrosion in an indoor environment is most often caused by a short list of contaminants or combinations of a few contaminants. The compounds present in an individual data center are highly dependant on the location and the controls put in place to mitigate them. Most often this would involve the selection and application of the appropriate chemical filtration systems to clean both the outdoor ventilation air as well as the air inside the data center.
The final part of the plan would be to establish a real-time environmental monitoring program based on the severity levels established in the ISA Standard. Real-time environmental reactivity monitors (ERMs) can provide accurate and timely data on the performance of the chemical filtration system as well as the air quality within the data center itself.
The absence of gaseous contamination controls can be the result of a lack of knowledge and education. Often the relationship between corrosion levels and hardware failures in data centers is overlooked or unknown. However, due to the efforts of companies like Rockwell, Dell, IBM, HP and others, this knowledge gap is shrinking and successful corrosion monitoring and control programs can be developed and implemented - assuring reliable operation of data center equipment.
SUMMARY AND CONCLUSION
Most computer systems and associated components, such as telecommunications and storage systems used in data centers are protected against the potential threats posed by fire, power, shock, humidity, temperature and particulate contamination. Unfortunately, the potential damage to this equipment caused by the corrosive effects of gaseous contaminants due to RoHS compliance has still not been fully recognized or addressed.
Manufacturers have to comply with RoHS if they want to continue to do business in the EU, China, etc., and many have taken the ImmAg route as their path to compliance. This has taken care of one issue but has presented new challenges with regards to equipment reliability.
The first stages of corrosion may show up as untraceable alarms, as corrosion products grow and provide random circuit paths that clear themselves as soon as current flows through them. However, some information may have already been lost. In the worst cases, substantial resistance to current flow is developed, variations in voltages occur, and the entire circuit may become useless due to changes in component value. These circuit failures, whether in a server, hard disk drive, switch, or other electrical component, increase equipment capital cost and warranty and maintenance costs. Something as simple as a corroded chip may destroy the promise of realtime connectivity, resulting in consumer dissatisfaction and lost profit.
Recognizing the severity of the problem, most of the world's leading manufacturers of computer systems have placed in their site planning manuals references to the ISA or similar standards for acceptable levels of airborne contaminants. This is because they provide achievable and effective guidelines to protect electronics and electrical equipment from the damaging effects of corrosive gases.
The use of reactivity monitoring using both copper and silver corrosion data is being used to provide a more complete environmental assessment than the copper-only monitoring described in ISA Standard 71.04-1985. This is based on studies that have shown that while copper coupons may be good indicators of corrosive gases in an environment, they are not sufficiently sensitive in all cases to many of the contaminants encountered in many environments (Muller 1991, Muller et al 1991). The standard is currently undergoing a long past-due update to incorporate the use of silver corrosion data into quantifiable severity levels relating to equipment reliability. It is also planned to add the use of real-time reactivity monitors as an acceptable approach for environmental classification using reactivity monitoring. Most likely this will be a reference to mass-gain monitoring techniques (e.g., quartz crystal microbalances).
When both copper and silver monitoring is used and the copper reactivity is equal to an ISA G1 classification, the corresponding silver corrosion rate can be expected to be approximately double that of the copper coupon for indoor environments (Purafil 2000). Therefore, total silver corrosion less than 600[Angstrom] with no evidence of chlorine contamination (as AgCl), would be considered an acceptable complement to a class G1 copper environment with no evidence of sulfur contamination (as [Cu.sub.2]S). Total silver corrosion above 600[Angstrom] with or without evidence of chlorine often indicates an environment where the effects of corrosion are measurable and there is an increased probability of corrosive attack.
The issue and potential for corrosion-related problems in data centers has been presented. Data from many different sites shows that corrosive atmospheres exist in locations that most would consider otherwise benign if not for the changes in electronics mandated by RoHS legislation. These problems can be addressed by continuous monitoring of the data center environment and removal of corrosive contaminants where indicated. Ultimately, the successful implementation of a corrosion protection program requires:
* Knowledge and understanding that corrosion of electronic equipment is a serious problem.
* Commitment to a monitoring program to describe the potential for electronic equipment failure.
* Commitment to an integrated contamination control system.
* Commitment to take corrective action whenever necessary.
Anonymity. 2008. http://www.emsnow.com/npps/story.cfm?id=24880.
Anonymity A. 2008. http://smt.pennnet.com/Articles/Article_Display.cfm?Section=ARCHI&ARTICLE_ID=266474&VERSION_NUM=2&p=35
Anonymity B. 2008. http://www.uyemura.com/library-2.htm.
Anonymity C. 2008. http://www.smta.org/knowledge/proceedings_abstract.cfm?PROC_ID=1765.
Anonymity D. 2008. http://www.era.co.uk/index.asp.
Anonymity E. 2008. http://www.era.co.uk/news/rfa_feature_03.asp.
Anonymity F. 2008. http://www.brushwellman.com/alloy/tech_lit/May02.pdf.
Anonymity G. 2008. http://nepp.nasa.gov/WHISKER/back-ground/index.htm.
Anonymity H. 2008. http://www.calce.umd.edu/lead-free/longterm.htm.
Brusse J., H. Leidecker and L. Panashchenko. 2007. Metal Whiskers: Failure Modes and Mitigation Strategies, Microelectronics Reliability & Qualification Workshop, National Aeronautics and Space Administration, Goddard Space Flight Center, Greenbelt, Maryland.
England, W.G., M.W. Osborne, X. Zhang. 1991. Applications of a Real-Time Electronic Contact Corrosion Monitor, Proceedings of Advances in Instrumentation and Control Vol 46: pp 929-955.
ISA. 1985. ANSI/ISA-71.04-1985 - Environmental Conditions for Process Measurement and Control Systems: Airborne Contaminants, Research Triangle Park: International Society for Automation.
Henriksen, J., R. Hienonen, T. Imrell, C. Leygraf, and L. Sjogren. 1991. Corrosion of Electronics: A Handbook based on Experiences from a Nordic research Project. Litohuset AB, Stockholm: Swedish Corrosion Institute, page 26.
Mazurkiewicz, P. 2006. Accelerated Corrosion of Printed Circuit Boards due to High Levels of Reduced Sulfur Gasses in Industrial Environments, Proceedings of the 32nd International Symposium for Testing and Failure Analysis, pp. 469-473.
Muller, C.O. 1990. Combination Corrosion Coupon Testing Needed for Today's Control Equipment, Pulp & Paper Magazine, February, 1990.
Muller, C.O. 1991. Multiple Contaminant Gas Effects on Electronic Equipment Corrosion, Corrosion Journal, 47(22):146-151.
Muller, C.O. C.A. Affolder, and W.G. England, "Multiple Contaminant Gas Effects on Electronic Equipment Corrosion: Further Studies," Proceedings of Advancements in Instrumentation and Control, Instrument Society of America, October 1991, Anaheim, CA.
Muller, C.O. 1999. The Use of Reactivity Monitoring as an Alternative to Direct Gas Monitoring for Environmental Assessments in Cleanrooms. Proceedings of Clean-Rooms East '99 Conference pp. 145-156.
Reid, M, J. Punch, C. Ryan, J. Franey, G.E.Derkits, W.D. Reents, and L.F. Garfias. 2007. The Corrosion of Electronic Resistors, IEEE Transactions 30(4): pp. 666-672.
Rice, D.W., P. Peterson, E.B. Rigby, P.B.P. Phipps, R.J. Cappell, and R. Tremoureaux. 1981. Atmospheric Corrosion of Copper and Silver. Journal of the Electrochemical Society: Electrochemical Science and Technology 128(2): pp. 275-284.
Schueller, R. 2007. Creep Corrosion on Lead-free PCBs in High Sulfur Environments, Journal of Surface Mount Technology 21(1): 21-29.
Unpublished data. 2000. Purafil, Inc.
Unpublished data. 2008. Purafil, Inc.
Weiller, A.J. 1994. Electronic Monitoring of Indoor Atmospheric Pollutants, Proceedings of Healthy Buildings '94, pp 241-243.
Zhao, P. and Pecht, M. 2005. Mixed flowing gas studies of creep corrosion on plastic encapsulated microcircuit packages with noble metal pre-plated leadframes, IEEE Transactions on Device and Materials Reliability (TDMR) 5(2): 268-276.
Michael McKenna, Sr. Mechanical Engineer, CCG Facilities, Baltimore, MD: Why wouldn't manufacturers move to more corrosive-resistant processes? Ventilation rates in data centers are very low, so why is contamination being seen in data centers?
Chris Muller: This is a goal of all manufacturers, but the passage of RoHS regulations have resulted in problems related to the choices and selection of lead-free surface finishes and solders for printed wiring boards (PWBs) and printed circuit boards (PCBs). By the time equipment is sold as a finished product, it has passed through many hands from beginning form circuit board production, to lead frames, to chip packaging, to assembly and test, etc., so we are not really sure where corrosion might have started. Also, it's not just circuit boards that are failing, it's hard drives and other types of electronic equipment that are experiencing corrosion-related failures.
The previously widely used PWB, hot air solder leveling (HASL), tin/lead (SnPb) finish has been replaced with lead-free and coplanar PWB finishes such as organic solderability preservative (OSP), immersion silver (ImAg), electroless nickel immersion gold (ENIG), and immersion tin (ImSn). SnPb HASL offers excellent corrosion protection of the underlying copper and has inherent corrosion resistance. Lead-free board finishes provide reduced corrosion protection to the underlying copper due to their thin coating. For ImAg, the coating material can also corrode in more aggressive environments, which has become an issue for products used in regions with high levels of sulfur-containing pollutants, such as China where a large percentage of the world's PWBs and PCBs are being manufactured.
Accelerated corrosion leading to system failure has been observed on printed circuit boards used in industrial environments that contain abnormal levels of reduced sulfur gasses, such as hydrogen sulfide ([H.sub.2]S) and elemental sulfur. The problem is compounded by the fact that elemental sulfur is regulated by OSHA as a nuisance dust and is allowed in a human working environment at the parts per thousand levels. Anecdotal data clearly shows that elemental sulfur gas present at the parts per million level can cause computer systems to fail within two months of use. Technologies such as immersion silver are especially susceptible to this type of corrosion. With the continued use of organically coated copper (OCC) and immersion silver finishes, the number of failures due to reduced sulfur gasses in the environment has risen substantially.
In these corrosive environments, creep corrosion can lead to product failures in short service life (1-5 years). In some of the worst cases, creep corrosion failures within one year of product deployment have been reported. This has prompted an industry-wide effort to understand creep corrosion, although minimal progress has been made, due to an inability to reproduce creep corrosion in the lab using realistic accelerated aging tests. It has been reported that creep corrosion is highly sensitive to the surface chemical properties, which must be considered when simulating creep corrosion in the laboratory.
The bottom line is that there are active efforts to develop lead-free surface finishes, solders, and PWB/PCB manufacturing processes that provide a higher level of corrosion resistance. But, the fact that corrosion-related failures are showing up in nonindustrial urban settings with relatively low levels of corrosive gases present means that it is most likely a combination of factors that result in these failures. Manufacturers of electronic equipment are now looking to their upstream suppliers to see if boards and components are predisposed to corrosion due to exposure to a "dirty" environment during the manufacturing process, or if evidence of corrosion can be observed on components prior to final assembly.
As for your second question, we are not seeing corrosion in all data centers just yet. Since the switch to lead-free manufacturing, the majority of corrosion-related problems in commercial data centers have been reported from China, India, and other developing countries where ambient air quality is poor. (Industrial locations have always presented challenges with respect to electronic equipment, so I am not going to address those here.) We have not seen as many problems in the United States or Europe, but we believe that is due to better ambient air quality (for the most part) and that these lead-free components have not been exposed long enough to these lower levels of contamination for the effects of corrosion to present themselves. We believe that in the next year or two we will begin to see corrosion issues crop up in the United States and Europe.
Vali Sorell, Associate Partner, Syska Hennessy Group, Charlotte, NC: Based on your experience, who is most qualified to select, specify, and design the monitoring system for gaseous contamination?
Chris Muller: That's a loaded question, because I do not wish to inject personal bias into my answer. However, from an objective viewpoint you want to use a monitoring technique that provides information relative to the corrosion risk potential by measuring corrosion levels directly, as opposed to simply measuring levels of gaseous pollutants. Direct gas monitoring is OK, but when presented with a list of contaminants and their concentrations, that's all you really have. This does not account for the potential synergies (positive and negative) between contaminants, the effects of fluctuations in temperature and/or humidity, etc., that could result in higher or lower levels of corrosion in the space. If you are trying to prevent corrosion and the effects of corrosion, you should use a technique that measures corrosion directly.
Corrosion monitoring in a data center would provide information relative to the type(s) of corrosion that is present as well as the rate at which this corrosion is being formed. This can be done using passive devices such as corrosion classification coupons using both copper AND silver metal or realtime atmospheric corrosion monitors, again using sensors that measure the copper and silver corrosion rates. The most qualified supplier for this type of contamination monitoring would be able to provide the monitoring devices and corrosion analysis themselves and need to have years of experience with both passive and real-time monitoring of airborne corrosion. They would have an extensive database of monitoring results from data centers in many different locations that could be used to develop risk potentials SPECIFIC to these environments and provide an objective interpretation of the results.
Hans Besselink, BSC, Royal Haskoning, Rotterdam, Netherlands: In Europe, it became common to design all air cooling systems for data centers by means of heat recovery rotation wheel with additional dx cooling, so called Kyoto cooling, see www.kyotocooling.com. My question is what is your opinion on the implications regarding corrosion issues for all air systems in data centers?
Chris Muller: Asking about "corrosion issues for all air systems in data centers" is a very broad question, and I am not sure what specific information you are looking for. Let me state that most, if not all, of the contaminants that have been identified as the major causes of corrosion are sulfur gases, such as sulfur oxides and active (reduced) sulfur compounds, that have sources that are almost exclusively from outdoor air. If no outdoor air is being directly introduced into the data center, then the risk of corrosion would be expected to be low. However, with the free cooling concept being championed by many, any contaminants in the outdoor air would be introduced into the data center environment, and the potential risk for corrosion goes up.
There are internal sources of corrosion, such as off-gassing from data center personnel. We have monitored facilities where we can tell when someone who has been outdoors or has even been smoking has entered due to an increase in real-time corrosion rates due to off-gassing from clothing. It may be a small transient increase in corrosion, but corrosion is cumulative, and over time, the chance of equipment/device failure increases. Electronic equipment using PVC wire insulation can produce chlorine if temperatures get too high. This is an extremely aggressive contaminant toward metals.
Due to the wide range of data center designs with respect to air handling and cooling and the types of air handling and air-conditioning systems being used, our recommendations have always been to develop the risk potential for the data center environment using passive and/or real-time corrosion monitoring. Since the switch to lead-free manufacturing, many data center owners/operators use or have installed corrosion monitors as a preventive measure. However, it is not until the datacom equipment manufacturers have added corrosion control (monitoring) to their terms and conditions--just as they have specifications for temperature, humidity, shock, vibration, ESD, etc.--that we will be able to say whether one type of air handling system design will be better than another.
(1.) Note that Croatia, Norway, and Switzerland are not part of the EU. They may nevertheless have legislation implementing EU WEEE and RoHS rules, or similar legislation.
(2.) Known as the Instrumentation Society of America when the standard was published and later as the Instrumentation, Systems, and Automation Society.
(3.) Copper has been the primary coupon material because data exist which correlates copper film formation with reactive (corrosive) environments and it has proven to be particularly useful for environmental characterization.
Chris Muller is the Technical Director at Purafil, Inc., in Doraville, GA.
|Printer friendly Cite/link Email Feedback|
|Date:||Jan 1, 2010|
|Previous Article:||SAA integrates with ASHRAE.|
|Next Article:||Contamination in the Data Center--the impact of CRAC belt dust.|