Why be concerned about moisture?

Moisture-related failures of microelectronic components have occurred across the microelectronics industry for many years. As early as the mid-1950s, Crawford and Weigand¹ showed that water vapor was the greatest contamination problem in standard relays, causing corrosion of contact materials. In 1965, Eisenberg, Brandewie and Meyer² reported that device reliability was influenced by moisture in a hermetic package’s internal atmosphere. Subsequent papers in the 1970s and 1980s described ongoing moisture-caused component and system failures.^3,4 These kinds of failures persist today. Because moisture is difficult to control, preventing the failure of hermetic seals is an ongoing concern for integrated circuits, hybrid assemblies, optoelectronic parts, and microelectromechanical systems (MEMS) installed in military and aerospace, automotive, telecom, and medical device systems.

Moisture-related failure mechanisms are well-documented and have included:

• Corrosion of aluminum wire, interconnect, and bond pads
• Metal migration such as dendritic growth and whisker formation
• Electrical leakage and device instability
• Leaching of ionic species
• Etching/attack on ultrathin film structures (e.g., nichrome)
• Capacitor delamination and popcorning

Though these moisture-related failure modes have been long appreciated, they are still often ignored, misunderstood, or inadequately protected against. Three of the more common failure modes encountered are metal corrosion, metal migration, and parametric device instability.

Low temperature field service conditions where a sealed headspace containing moisture falls below its dewpoint, and the moisture condenses as liquid, can trigger corrosion of aluminum metallized structures, for example:

Al + 3H₂O → Al(OH)₃ + 3/2H₂

2Al + 3H₂O → Al₂O₃ + 3H₂

Ionic impurities, especially anions like chloride and fluoride, or cations like sodium and potassium, can initiate, aggravate, and/or accelerate aqueous chemical reactions, or degrade device function by rendering condensate more electrically conductive. One of the first (Crawford and Weigland¹) and most commonly observed forms of corrosion was the action of chlorine and water to attack metallization, in which chlorine catalyzes or accelerates and the reaction proceeds until either all metallization or all water is consumed.

Al + 3NaCl + 3H₂O → AlCl₃ + 3NaOH + 3/2H₂

6HCl + 2Al → 2AlCl₃ + 3H₂
AlCl₃ + 3H₂O → Al(OH)₃ + 3H₂

2Al(OH)₃ + aging → Al₂O₃ + 3H₂O

While chemistries are different, copper metallization can be similarly vulnerable.

For the mechanism of metal migration, the only factors required are a difference in potential between two adjacent conductors and the presence of ionic contaminants and moisture. The ions, with moisture as a carrier, enable conductor material to migrate from the more negative conductor to the more positive conductor under the influence of bias. With time, thin highly conductive metal dendrites or “whiskers” can grow and eventually bridge conductors, causing a short circuit. Only minute quantities of ionic contaminants are necessary to cause this failure mode. Gold, for example, requires less than 10-7 grams/cm2 of potassium chloride (KCl), with a 12-volt applied bias, in order to migrate. Tin and silver are especially prone to migrate, though many other metals can as well.⁵

For the mechanism of parametric device instability, water in the presence of sodium ion can supply a hydrogen atom which will rapidly diffuse through passivating layer(s) of a semiconductor structure to accumulate destabilizing electrical charge at Si-SiO2 interfaces. Magnitude and impact of this effect depends on design and structure of the device and the amount and distribution of contaminant. It’s not uncommon to see increased leakage currents, failure of Field Emission Transistors (FETs) to enhance or deplete properly, or total lack of function.

Additionally, it is known that as little as three monolayers of water molecules adsorbed on surfaces is the threshold condition for electrical conduction across a surface.⁶ If three or more adsorbed monolayers bridge adjacent conductors, surface electrical leakage and conduction can lead to parametric instability, or galvanic activity that eventually corrodes and/or causes metal migration that physically damages narrow metallization runs, contacts, or wire-bonded connections.

With regard to all these mechanisms, if sealed package headspace is initially processed to exclude water vapor sourced from processes and materials such that it is very dry, and the package maintains a high-integrity hermetic seal preventing ingress of moisture over part lifetime, the mechanisms mentioned cannot occur. But if too much water vapor is or becomes present (military specifications limit it to 5,000 parts per million by volume⁷), devices are susceptible to the failure mechanisms described.

Where does this entrapped moisture, so potentially fatal to device reliability, come from? The belief that a low moisture reading on the sealing box hygrometer will ensure a dry ambient inside the device is a wide-spread myth. It is essential to monitor and maintain a low-moisture sealing environment, but historically this has not been the primary source of high internal moisture levels.

The two major sources of moisture are: (1) water adsorbed on the surface of the package walls and other materials and/or water absorbed within headspace materials that volatilizes post-seal into the headspace, and (2) loss of package hermeticity followed by ingress of ambient humidity. With this in mind, the following summarizes some potential sources of moisture in microelectronic devices:

1. 1. Adsorbed/Absorbed Moisture in Package Parts and Materials⁸
      The following may be sources of moisture when inappropriate times and/or temperatures are used during pre-seal bakeout or when exposure of parts and materials to atmosphere occurs between bakeout and sealing.
      1. Package part surfaces
      2. Bubbles or microcracks in glass/lead/metal interfaces
      3. Base-sidewall junction
      4. Internal plating
      5. Polymeric materials
         • Thermal decomposition
         • Trapped gases under large area devices and substrates
         • Excess of curing agent
         • Moisture-sorbing materials (which retain moisture within the package so that when the device is heated, the material releases large amounts of water, increasing potential for failure)
         • Inadequately cured polymer that outgasses moisture during subsequent heating
      6. Residual hydrogen in the sealed headspace, sourced from Kovar alloy and metal plating, chemically reducing exposed surface metal oxides producing water (H2 + MxOy → H2O + M). Physical chemistry free energy diagrams favor these reactions especially for oxides of iron and silver (silver being commonly present in conductive adhesives)⁹
      7. Particulate matter (which may outgas moisture or provide surface active sites and secondary valence forces allowing moisture to condense at a much higher temperature than the normal ambient dewpoint)
      8. Human contamination during assembly

Mil-Std Test Method 5011 “Polymeric Materials”⁹ gives guidelines and test methods for qualifying, selecting, and using adhesives, coatings, getters, and other materials intended for inclusion in hermetically sealed components. This standard should be consulted when new assemblies are being designed and engineered or when prototypes are built.

1. Loss of Package Hermeticity¹⁰
   1. Loss of hermeticity due to handling, installation, or operation of a device
   2. Temporary loss of hermeticity when bombing pressures are applied during leak testing
   3. Rework of poor package seals
2. Sealer Dry Box Conditions
   1. Ambient air leakage into sealer dry box or gas delivery lines
   2. Contaminated source of sealing blanket gas

Keeping the device as dry as possible during fabrication, assembly, and testing cannot be overly emphasized. To protect the part during its lifetime, a sufficient pre-seal bakeout followed by hermetic sealing of the part in an inert atmosphere with no interim exposure to humidity helps ensure that the device will function as designed. It is always essential to qualify and select the right materials and fully engineer and control treatment and sealing processes to assure control of internal moisture.

Oneida Research Services is especially experienced and equipped to help clients more fully explore and understand these failure mechanisms. We can help you accurately and reliably measure both leakage rates and internal gas compositions of sealed devices and to conduct studies for optimizing materials and processes to achieve control of moisture and other volatiles within hermetically sealed electronic components and systems.

References