To our reader who has found this page somehow:

This document is a text dump for search bots. This document captures the text of our various PDFs. This document was prepared as part of a search engine optimization experiment. These documents in human optimized PDF format can be found on our library page at https://www.vacuumenergyinc.com/library/

If you have any questions please call us at 216-991-7000 or at info@vacuumenergyinc.com with your questions and we will be happy to assist you.

Thanks,

Vacuum Energy inc.

White Paper

Technology White Paper: PolyGettersÔ

Ó 2017 Vacuum Energy Inc.

WP-PG-01 v1.4

About Vacuum Energy

Vacuum Energy has been supplying advanced getter materials to demanding users since 1991. Our original product lines included conventional metal getter systems for vacuum thermal super insulations in the oilfield and appliance markets.

Metal based getters are a mature technology that has been in widespread use for nearly a century. As getter users advanced their own technologies, it became clear that performance limitations in inherent in metal getter systems precluded their use in any number of leading edge applications. In order to meet these needs, Vacuum Energy partnered with Sandia National Laboratories to develop new, technically advanced hydrogen and water getter systems. Vacuum Energy is proud to be Sandia National Laboratories’ exclusive licensee for their polymeric getter systems.

Sandia possesses deep expertise in polymeric getter materials. Sandia and Vacuum Energy worked together to create new types of polymeric getter materials for demanding client applications where operation in the atmosphere, and resistance to atmospheric gases, water, volatile organics and oil mists are critical.

Development work on advanced gettering systems continues to this day. Our getters are used worldwide in millions of products every year. Vacuum Energy’s PolyGettersÔ improve the performance of numerous industrial products, and help prevent the risk of industrial and consumer product explosions by irreversibly scavenging unwanted hydrogen gas.

Whether your conditions call for meeting the extreme requirements of sensitive military components, or insuring the safety of millions of consumer devices, let us help you with your gettering challenges.

Vacuum Energy PolyGettersÔ 2

Characteristics of Polymeric Getters

The PolyGetterÔ polymeric getterswere developed to getter hydrogen. The basic characteristics of the PolyGetterÔ polymeric getter materials are driven by the original need for reliable and long-lived materials to meet the extreme requirements of the nuclear industry.

An important feature of our family of PolyGetterÔ polymeric getter materials is their endless flexibility to be custom formulated to meet customer requirements. A key aspect of our ability to tune getter formulations for the customer’s application is that our getters generate no unwanted byproducts via reactions like PdO + H2 ® H2O + Pd.

Vacuum Energy’s custom PolyGetterÔ formulations build on a set of common base characteristics:

Available in a number of forms:
- - powders - Free flowing, co-deployed with mole sieve desiccants
- - pellets - Small cylinders, good mechanical strength
- - spray coating - Thin, lightweight films, much like spray paints
- - monoliths - Sizes to a few grams co-deployed with desiccants
- - self-adhesive labels - Sprayed or silk screened
Safe and Compatible:
- - superb water & oil resistance, unlike other getters
- - non-pyrophoric
- - non-gas generating
- - widely compatible with other materials, including those used in high

reliability microelectronic and fiber optic packages

- demonstrated resistance to many poisons and radiation

Vacuum Energy PolyGettersÔ 3

Fast absorption rates:
- rapidly reduces target species to sub-ppm concentration levels

Compatible with common process temperatures:
- - Because these are polymeric/organic materials, the useable temperature range should exceed 150-200oC only for short periods, such as during device fabrication.
- - Please ask us about special high temperature materials for special applications.
PolyGetterÔ polymeric getters are long-lived
- - no temperature activation required
- - operate in a variety of atmospheres (air, nitrogen, vacuum, steam, oil

mists)

- no passivating layer formation
- unaffected by oxygen and water
- function well in cold temperatures on ocean floor.
- very low vapor pressure, low outgassing, compatible for vacuum

conditions

- chemically compatible with a wide variety of desiccants, metals, plastics

and glasses

Unlike PdO and similar metal based getter systems, no water is formed as a byproduct of the gettering reaction. When water sorbents are used in our getters they are 100% available to sorb water contamination present in your system.
Unlike metal hydride getter systems based on Pd, Zr, and Ti, there are no dissimilar metals issues or particle shedding when concentrations exceed the a phase hydride and enter the b phase, thereby expanding the metal’s lattice structure.
Sorption mechanisms are fully passive and irreversible at the rated operating temperatures.
Years of shelf life as packaged.

Vacuum Energy PolyGettersÔ 4

Hydrogen PolyGettersÔ

Polymeric hydrogen gettering is based on an invention by Sandia National Laboratories*, whereby additions to carbon-carbon multiple bonds are performed. Numerous patents and other IP have resulted from Vacuum Energy’s work with Sandia expanding this breakthrough into numerous applications.

Key characteristics of Vacuum Energy’s polymeric hydrogen getters include:

No water is formed, in contrast to PdO getters.
No bakeout is required to activate the getter.
Fast absorption rates rapidly reduce hydrogen concentrations to parts-per-million levels.
Commercially available with a range of characteristics and configurations that are custom formulated to specific customer needs.
Capacities range from 10 - 100+ std.cc’s H2/gm depending on the getter format.
Long-lived, not passivated by oxygen or water vapor.
Will operate in a variety of atmospheres (vacuum, steam, air, nitrogen, inert gasses.)
Very low vapor pressure, ultra-low outgassing, specifically designed for vacuum service.
Chemically compatible with a wide variety of desiccants, metals, plastics and glasses.
Years of shelf life as packaged.
For GaAs packages, getter/silicone co-deployment is preferred option.
Demonstrated resistance to many poisons and radiation.
Non-pyrophoric, non-gas generating, widely compatible with other materials.
Compatible with the anticipated materials in fiber optic systems. Typical hydrogen pumping curve:

* http://www.ca.sandia.gov/8700/projects/content.php?cid=41

Vacuum Energy PolyGettersÔ 5

101

100

10-1

10-2

10-3
0 10 20 30 40 50 60

time (min.)

This curve reflects a typical material. Materials are formulated for exact customer needs. Please contact us for examples that are pertinent to your application.
In this example the getter testing apparatus has not been fully degassed of air or dried of water.
Note the low residual pressure of remaining water and air, about <10-2 torr. Getter still absorbing
Hydrogen removal is unaffected by air, humidity, water, seawater, even oil mist contamination.

For more information on our advanced PolyGetterÔ materials, please give us a call!

Vacuum Energy Inc. 2714 W. Park Blvd. Shaker Heights OH 44120 216-991-7000 www.vacuumenergyinc.com info@vacuumenergyinc.com

Vacuum Energy PolyGettersÔ

pressure (torr)

FO Getter

Tech Note

Tech Note Topic: Fiber Optic Hydrogen Getter Gel Comparison Getter Type: Vacuum Energy HiTopTM Hydrogen Getter Gel

page 1 of 2 rev. 2

www.vacuumenergyinc.com

Material Type Example Configurations Engineerable Multi-Gas Getter System

High Purity Polymer Based

Multiple Configurations:
Gels, Powders, Pellets, Dispensable

Hydrogen Uptake Test Results, by Sandia National Laboratories

A series of comparison tests between Vacuum Energy’s HiTopTM gel material, and a getter gel marketed for hydrogen darkening remediation in fiber optic cables were performed. The testing was performed at Sandia National Laboratories in Livermore, CA.

250

200

150

100

Blank

Competing Material Run 1

Competing Material Run 2

VEI Hi-Top

3000 4000 5000

Pressure (torr)

1000 2000
time (min.)

Blank

Sepigel run 1

sepigel run 2

Hi-T op

Hydrogenations

Hydrogen absorption at 125°C for Vacuum Energy’s HiTopTM gel & a competing fiber optic getter gel.

Tech Note

page 2 of 2 rev. 2

www.vacuumenergyinc.com

Hydrogen Uptake Test Results, by Sandia National Laboratories Continued

Testing began with sample preparation. Samples were degassed at 125°C, then equilibrated to hydrogen at ~195 torr. A blank test was also prepared for comparison. The “Blank” test using no getter was stopped at 3500 minutes. After demonstrating a leak-free apparatus it was cooled to room temperature.

The HiTopTM gel test was stopped at 350 minutes, and cooled to room temperature. Test run #1 of the compet- ing fiber optic gel was stopped at 4200 minutes, and cooled to room temperature. Test run #2 of the competing gel was heated to 200°C to see if the reaction rate could be increased. The elevated temperature was main- tained for about 2600 minutes , then returned to 125 °C, and stopped at 3800 minutes and cooled to room tem- perature. The impact of these temperature changes can be clearly seen on the curve.

The results of the comparison test are clear. The Vacuum Energy HiTopTM gel has a higher hydrogen capacity, and produces a lower hydrogen overpressure which might otherwise lead to fiber optic cable attenuation losses:

• The H2 capacity of the HiTopTM gel getter was measured to be 30.3 std cc/gm.

• The H2 capacities of the competing gel, runs #1, #2 were measured to be 1.0 and 1.7 std cc’s/gm respec- tively.

Conclusions

Vacuum Energy's HiTopTM hydrogen getter gels significantly advance the state of the art when compared to conventional hydrogen getter gels. While the test results cited above for the HiTopTM hydrogen getter gel were restricted to a temperature of 125 ̊C, other formulations of HiTopTM hydrogen getter gel are available for operat- ing temperatures exceeding 250 ̊C.

Contact Information

For further information please contact us!

VACUUM ENERGY INC. 2714 W. Park Blvd. Shaker Heights, OH 44120 216-991-7000 (Ohio) 719-966-4296 (Colorado) www.vacuumenergyinc.com

ROHS

Tech Note

Tech Note Topic: RoHS Compliance Getter Type: Vacuum Energy PolyGetterTM

Engineerable Multi-Gas Getter System High Purity Polymer Based

Multiple Configurations:
Sheets, Powders, Pellets, Dispensables

ROHS Compliance

page 1 of 1 rev. 1

www.vacuumenergyinc.com

Material Type

Example Configurations

Vacuum Energy, Inc. is fully committed to the continuous protection and improvement of the environment. Our products are carefully engineered to eliminate the use of hazardous substances wherever possible. As a result of our focus on environmental health and safety, we are proud that our PolyGetterTM hydrogen getters and combination hydrogen/water absorbers do not contain the following substances:

Cadmium (Cd)
Hexavalent Chromium (CrVI+)
Mercury (Hg)
Lead (Pb)
Polybrominated biphenyls (PBB)
Polybrominated diphenyl ethers (PBDE)

Contact Information

For further information please contact us!

VACUUM ENERGY INC. 2714 W. Park Blvd. Shaker Heights, OH 44120 216-991-7000 (Ohio) 719-966-4296 (Colorado) www.vacuumenergyinc.com

Ionics

Tech Note

page 1 of 2 rev. 7

www.vacuumenergyinc.com

Tech Note Topic: Report on testing by an accredited outside laboratory for ionic impurities contained in representative configurations of the Vacuum Energy Hydrogen Getter.

Getter Type: Vacuum Energy PolyGetterTM

Engineerable Multi-Gas Getter System High Purity Polymer Based

Multiple Configurations:
Sheets, Powders, Pellets, Dispensables

Test Protocol

A common concern in microelectronic packaging is corrosion of internal components of the package induced by moisture borne ionic contaminants. This concern is specifically addressed in MIL-STD-883H, Test Method 5011.5, EVALUATION AND ACCEPTANCE PROCEDURES FOR POLYMERIC MATERIALS, section 3.5.4:

“3.5.4 Ionic impurities. The ionic impurity content shall be determined in accordance with 3.8.7 and shall meet the requirements specified in table II.”

and

“3.8.7 Ionic impurities. A water-extract analysis shall be performed to determine the level of ionic contamination in the cured polymeric material. The total ion content (specific electrical conduc- tance) and the specific ionic content for the hydrogen (pH), chloride, sodium, fluoride and potas- sium ions shall be measured. Other ions present in quantities > 5 ppm shall also be reported in ppm.”

Test Results

In order to determine the ionics contribution of PolyGettersTM to a silicone getter matrix, various samples were prepared. Otherwise the getter formulation was typical of our normal product line. Both 1B (with and without any desiccating agents) and 6D getters were tested.

The getter samples were tested per the method called out in 3.8.7 of 5011.4 in MIL-STD-883H by an accredited test lab. A copy of MIL-STD-883H and the test reports are available upon request. Comparative Cookson data was obtained from their Technical bulletin “STAYDRY, Patented Technology, H2-3000 - Hydrogen and Mois- ture Getter” revision 02/02 DML.

As shown in the following table, the ionics contribution of PolyGetterTM is significantly less than that of tradi- tional silicone matrix type getter formulations.

Material Type

Example Configurations

Tech Note

page 2 of 2 rev. 7

www.vacuumenergyinc.com

Ionics Testing

MIL-STD-883H Requirements

Vacuum Energy PolyGetterTM 1B

Vacuum Energy PolyGetterTM 1B w/ mole sieve

Vacuum Energy PolyGetterTM 6D

Cookson STAYDRYTM H2-3000

Total ionic con- tent: ≤4.5 mS/m

2.23 mS/m

< 0.05 mS/m

2.22 mS/m

Not Published

Hydrogen: 4.0 < pH < 9.0

pH = 7.9

pH = 8.4

pH = 7.5

Not Published

Chloride ≤ 200 ppm

3.9 ppm

1.4 ppm

<200 ppm

Sodium ≤ 50 ppm

0.8 ppm

24.8 ppm

4.2 ppm

<204 ppm

Potassium ≤ 50 ppm

< 0.5 ppm

<502 ppm

Fluoride ≤ 50 ppm

< 0.5 ppm

<50 ppm

Contact Information

For further information please contact us!

VACUUM ENERGY INC. 2714 W. Park Blvd. Shaker Heights, OH 44120 216-991-7000 (Ohio) 719-966-4296 (Colorado)

info@vacuumenergyinc.com www.vacuumenergyinc.com

Flares

Tech Note

Tech Note Topic: Hydrogen Control in Pyrotechnic Devices

page 1 of 3 rev. 1

www.vacuumenergyinc.com

Material Type

Vacuum Energy PolyGettersTM

High Purity Polymer Based

USAF Approved Hydrogen Ad- sorption Materials

Technology Developed By San- dia National Labs

Munitions and Pyrotechnics Can Generate Dangerous Amounts of Hydrogen Gas

Magnesium based pyrotechnic devices have well known design and safety issues due to the internal generation of hydrogen gas. Hydrogen gas generation within IR countermeasure flares can cause swelling of the outer casing, preventing proper flare ejection. During boxed storage, this hydrogen can present an explosion risk. The source of hydrogen is the result of a reaction between water (typically humidity in the air) and the magne- sium in the device itself:

Mg + 2H2O → Mg(OH)2 + H2

Hydrogen is a highly flammable gas with a lower explosive limit of 4% in air. Due to this flammability, non- gettered munitions present a significant safety risk. Conventional mitigation techniques such as regular venting of munition containers are expensive, time consuming and hazardous operations.

Vacuum Energy, Inc., in conjunction with Sandia National Laboratories, has developed polymer hydrogen ab- sorbing materials (getters) to eliminate hydrogen build up in pyrotechnic devices. These materials selectively remove hydrogen from air, inert gases, or in vacuum.

The capacity of the various materials has been designed to cope with the very large quantities of hydrogen generated by pyrotechnics over their storage life. Vacuum Energy’s getters are engineered to keep hydrogen concentrations at or below a safety limit of 0.2% in order to prevent explosion risks.

Removing hydrogen is just one design factor to consider when designing a getter system. It must also be:

safe to handle
not require any special activation or pretreatment before use
easy to retrofit into existing containers
operate effectively in all expected storage conditions for pyrotechnic devices

PolyGettersTM meet these requirements.

Tech Note

page 2 of 3 rev. 1

www.vacuumenergyinc.com

Typical IR CM Flare Getter Flare Shipping Container Typical IR CM Flare Getter Configuration

Polymer Hydrogenation As A Gettering Mechanism

Vacuum Energy, Inc. has been supplying advanced hydrogen getter materials to demanding government and industrial users since 1991. These materials are used worldwide in environments as varied as vacuum, air, inert gases, and steam. The basic technology was invented and developed by Sandia National Laboratories, in support of their primary mission to safeguard the Nation’s nuclear stockpile. We are proud to be Sandia’s ex- clusive worldwide licensee for their polymer hydrogen getter technology.

The principal of polymer hydrogenation as a gettering mechanism involves a chemical reaction that saturates carbon-carbon double and/or triple bonds to their saturated analogs. In contrast to conventional metal hydride getters such as Palladium/Titanium, the hydrogen removal reaction is completely irreversible. The polymers are completely safe and nontoxic, and meet ROHS standards. The hydrogenation reaction utilizes a metal catalyst and proceeds as follows:

( ) H2, metal catalyst ( nn

The PolyGetterTM family of getters are available in a wide variety of forms. Small molded parts as pictured above have been effectively used within IR CM flares, and large flexible sheets are easily placed in pyrotech- nics munitions boxes minimizing risk of puncture. Other forms of PolyGetterTM can be dispensed within com- plex cavities, with engineerable cure times. High surface area PolyGetterTM powders are available for ex- tremely fast hydrogen removal rates, and are readily co-deployed with conventional water desiccants contained in sachets. Depending on the formulation, typical hydrogen capacities vary from 20 std cc/g to over 100 std cc/ g in special cases.

)

Tech Note

page 3 of 3 rev. 1

www.vacuumenergyinc.com

Getter Performance

A typical getter hydrogen absorption test result is depicted below. Sorption testing is performed internally at VEI, and can easily and independently be performed by clients in their own facilities. Other independent testing serv- ices can be provided by Sandia National Laboratories, and by third party labs.

Conclusions/Contact Information

Vacuum Energy’s PolyGetterTM hydrogen getters are a proven solution for hydrogen mitigation in pyrotechnic devices. This tech note barely scratches the surface of this complicated topic. For further information or sample material for evaluation please contact us!

VACUUM ENERGY INC. 2714 W. Park Blvd. Shaker Heights, OH 44120 216-991-7000 (Ohio) 719-966-4296 (Colorado) www.vacuumenergyinc.com

Storage and Handling

Tech Note

page 1 rev. 0

www.vacuumenergyinc.com

Tech Note Topic: PolyGetterTM Handling & Storage
Getter Type: Vacuum Energy PolyGetterTM Family of Getters & Recombiners

Introduction

Vacuum Energy’s PolyGetterTM family of getters and recombiners builds on our exclusive partnership with San- dia National Laboratories to develop new, technically advanced materials. Our materials are engineered for demanding client applications where operation in the atmosphere, and resistance to atmospheric gases, water, volatile organics and oil mists are critical. A crucial part of engineering our materials is to create safe and easy to handle materials that do not present risks to the end user.

PolyGetterTM Handling and Storage

The purpose of these guidelines is to preserve the gettering or recombining capacity of the parts as delivered. PolyGettersTM are chemically active agents and will react with any material present that they are designed to control. Consequently, any exposure to active agents before device seal must be minimized. That said, Poly- GettersTM are not hazardous and can be stored and handled safely in air.

PolyGettersTM should be stored before use in the unopened original packaging. After the package is opened unused parts should be stored in an equivalent heat sealed bag or other airtight, impermeable container. Stor- age in a dry, inert gas atmosphere such as nitrogen or argon is ideal. Similar storage of WIP with exposed get- ters is strongly recommended.

While there is a very small amount of hydrogen in normal atmosphere, this is not an issue for normal expo- sures. However, if hydrogen is used in a facility for other processes, there have been instances where getters and recombiners have been harmed by a higher than normal partial pressure of hydrogen in the air.

Operators are strongly encouraged to handle getters with gloved hands and or cleaned stainless steel tools in order to prevent contamination.

Operators are strongly encouraged to wear shop or lab coats and eye protection when handling chemically ac- tive materials like getters. Please note that VEI materials are not toxic, but we always encourage safe work- place practices with chemically active materials of all kinds.

Please refer to the MSDS for additional health and safety information.

Contact Information

For further information please contact us!

VACUUM ENERGY INC. 2714 W. Park Blvd. Shaker Heights, OH 44120 216-991-7000 (Ohio) 719-966-4296 (Colorado) www.vacuumenergyinc.com

SPIE 2009

Examining internal gas compositions of a variety of microcircuit package types & ages with a focus on sources of internal moisture

R. K. Lowry* & R. C. Kullberg**
Oneida Research Services, Inc., 8282 Halsey Road, Whitesboro, NY13492

ABSTRACT

The primary cause of corrosion, stiction or other failure mechanisms within hermetically sealed enclosures has historically been viewed as due to increases in internal moisture concentrations. It has historically been postulated that the primary source of moisture in these enclosures is the failure to achieve hermeticity at seal, or the loss of hermeticity post-seal. This postulation is the basis for failure analysis and mitigation both in the appropriate standards like MIL- STD-883 and in industrial QA procedures. Empirical observation of many data sets over the past 20+ years shows that this postulation does not always hold up in practice. The purpose of the current work is to test this postulation through the analysis of archival microelectronic packages and data sets of various ages.

Internal gas composition data for three different sets of packages totaling 165 units is reviewed. Of these, 63 were noncompliant (>0.50v%) on internal moisture, but only 8 (12.7%) showed an internal gas composition “signature” consistent with air leaking into the enclosure. These data suggest that leaks play a minor role in gas composition change within enclosures and that outgassing from materials is the principal contributor to internal moisture concentrations and the failure modes they induce.

Keywords

microelectronic packages, moisture, hermeticity, outgassing, leaks, permeation, corrosion, failure modes

1. INTRODUCTION

Moisture threatens reliability of devices in sealed enclosures by causing corrosion or electrical instability of microcircuits, fogging of optics, or stiction of moving parts in micro and nano machines. Materials outgassing and/or failure to achieve or maintain hermeticity elevates moisture in enclosures. Much attention has been focused on leaks due to lack of hermeticity as a principal cause of elevated moisture. A recent description of leak mechanisms1 confirms that leak avoidance is critical for cavities sealed under vacuum, and depicts Fick’s Law diffusion as the means of moisture ingress for enclosures with no pressure differential with the outside.

New capabilities for helium leak detection approaching 1x10-13 cc atm/sec have extended fine leak detection by more than four orders of magnitude.2 These new capabilities have focused moisture control efforts on leak avoidance. However, little real data have been published differentiating between hermeticity loss and materials outgassing as root causes of elevated moisture. This paper reviews internal gas analysis data for 200 units of various types, providing insight into the prevalence of mechanisms leading to excessive moisture in device enclosures.

* rlowry98@aol.com; phone 1-321-777-9949; www.electronic-materials.com; technical affiliate of ORS & consulting engineer **rckullberg@gmail.com; phone 1-719-966-4296; asanatechne.com; technical affiliate of ORS & consulting engineer

Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and Nanodevices VIII edited by Richard C. Kullberg, Rajeshuni Ramesham, Proc. of SPIE Vol. 7206, 720606
© 2009 SPIE · CCC code: 0277-786X/09/$18 · doi: 10.1117/12.807970

Proc. of SPIE Vol. 7206 720606-1

An important benchmark for understanding what is really occurring within a hermetically sealed cavity is the composition of natural air as shown in Table 1.

Table 1. Chemical composition of natural air.

Species

Content, v %

Comment

Nitrogen

78.08

Constant in natural air

Oxygen

20.95

Constant in natural air; N2/O2 ratio = 3.7

Argon

0.93

Constant in natural air; O2/Ar ratio = 22.5

Water

0.10 – 3.0+

Variable, depends on temperature , humidity, pressure

Carbon Dioxide

0.038

Any amount in enclosures > 0.04% is not from air

Neon

0.0018

Not from air if detected in enclosures

Helium

0.0005

Not from air if detected in enclosures

Methane

0.0002

Not from air if detected in enclosures

Hydrogen

0.00005

Not from air if detected in enclosures

If air has leaked into an enclosure sealed in pure inert gas(es) not containing Ar, the internal gas content should show the O2 and Ar components of air in roughly their natural ratio. Absolute concentrations will be less than those in ambient air (i.e. “diluted”) if the leak has not yet caused the enclosure cavity to reach equilibrium with outside air. Moisture content will vary depending on temperature and humidity of ingressed air and whether materials outgassing has contributed to moisture levels within the cavity. Carbon dioxide in an enclosure exceeding roughly 0.04v%3 is not from air and clearly indicates outgassing, as does the presence of any volatile organic compounds. Helium (if originally present) retained in any unit would argue against a significant leak mechanism.

1.1 Impacts of Ambient Temperature and Relative Humidity on Internal Moisture Concentration

Figure 1. Maximum water vapor content within a cavity due to air leaks of 25%. 50% and 75% relative humidity (RH).

350-l&dnimWSwVzçorContentinPlc DueSolelytoArLSsofVarious%RH

30 250 20 150 10

50 0

Figure 1 shows the maximum internal water vapor concentration at equilibrium within a cavity open to the external atmosphere via a leak. At cool temperatures and "nominal" levels of RH, a package will never become non-compliant to specification, or reach dangerous levels of water vapor, simply by leaking. It must be cautioned that:

0 10203040

Temperature, C

Proc. of SPIE Vol. 7206 720606-2

%Relative Humidity

0.75 0.5 0.25

a) under those temperature conditions, any significant outgassed water vapor may push the cavity into dew point or fogging conditions. While this is not good, it is a materials control problem, not a leak problem.

b) this point may be incorrect for stiction issues. Micromechanical devices can be sensitive to stiction at much lower levels of relative humidity.

At higher levels of temperature and RH, easily reachable in service conditions, packages can leak to non-compliant and/or dangerous levels of moisture. However, prior work on leak modalities into cavities where the cavity pressure is equal to that of the external ambient indicates that systems under these conditions should be modeled using Fickian diffusion.

Pre-existing internal concentrations of water vapor at time zero, or increasing concentrations of water vapor within the cavity due to outgassing during the package’s lifetime, (either or both caused by poor materials selection and process control) take a package to dangerous conditions sooner. As a result, there is less margin of safety if non-hermeticity is present or develops.

The key point is that it is quite possible to engineer and control materials and processes (M&P) to achieve dry packages at time zero, so as to create the most protection for the package. It is much harder to engineer and control M&P to be as certain of leak prevention, as it is to be certain of dry initial conditions. Consequently, leak rates and associated testing ought to be the secondary line of defense against moisture problems, while the primary line of defense ought to be M&P selection and process control. This contention is supported by the results of the current work.

2. SAMPLES & ANALYSIS

Three different groups of internal gas analysis data and samples were reviewed. The data were generated via test methods in MIL-STD-883, Test Method 1018.

2.1 Sample Groups/Data

Sample Group 1

This group contained 15 microcircuit packages built 10-20 years ago by Harris Semiconductor (now Intersil Corporation). The units were never placed in service and were stored for their entire lifetime in a plastic bag in a desk drawer. Eight were TO metal cans, having a nickel lid seam welded to a gold-plated header with eight pins entering through glass-to-metal seals. Seven were gold-plated Kovar flatpacks or DIPs of various sizes and lead counts with braze sealed lids. Most units had cavity volumes 0.1cc. The units were analyzed by Oneida Research Services in June 2008 per Test Method 1018. The analytical focus for this group was to determine if the concentration of internal moisture was elevated after long storage lifetimes by poor hermeticity (leaks) or by materials outgassing within the cavity.

Sample Group 2

This group contained 71 units from a Test Method 50114 qualification study of polymeric adhesives.5 Study of this group evaluated adhesive suitability, so moisture from material outgassing would likely be present. The focus was to determine if poor hermeticity contributed to elevated moisture content.

Sample Group 3

This group contained 114 units from a private database (author R. K. Lowry’s) compiled from many different clients. The group comprised microcircuit packages of a wide variety of styles and cavity volumes. Clients had obtained gas analysis data on the units for a variety of reasons including pre-shipment inspections, materials and process qualifications, DPA, failure analysis, and engineering studies. Details necessarily remain proprietary, but reviewing the gas analysis results with a focus on causes of elevated moisture content is instructive.

Proc. of SPIE Vol. 7206 720606-3

2.2 Data

Internal gas composition data for all sample groups came variously from three different commercial service laboratories, each with suitability for Test Method 1018 procedures. Except for Group 1, the choice of laboratory was by clients and not authors of this paper. Data were not compared for inter-laboratory differences. The study focus was to generally discern causes for noncompliance to the expectation of 0.50v% maximum internal moisture, as defined by MIL-STD- 883.

2.3 Data Analysis

Numerical values in gas analysis reports were rounded to two decimal places and tabulated as volume percent (v%), where 1v% = 10,000 ppmv. No entry in a cell in the data tables indicates that the species reported either as <0.01v% or as not detected.

3. RESULTS AND DISCUSSION

Table 2 summarizes the aggregate results for moisture compliance:
Table 2. Non-compliance to the 0.50v% moisture limit of the three sample groups

The first reaction to 29.5% of units noncompliant on moisture is alarm, justifiably if results were solely from pre- shipment inspections or process control measurements. But this study includes units that might have moisture control problems anyway, so the high incidence of noncompliance is not surprising.

Results for each of the three sample groups were tabulated and their overall gas composition “signatures” considered in detail for the likely causes of internal moisture.

3.1 Group 1, “Old” Units

Tables 3 and 4 summarize results for the TO cans and braze units. Table 3. Internal gas composition for eight TO cans, all about 20 years old.

Group

Total Units

Units >0.50v% H2O

1, “Old” units

2, 5011 Qual units

3, Consultant’s database

114

Total

200

59 (29.5%)

Year sealed

All units pre-1990

Approx. age

All units 18+ years

S/N

3-1

3-2

3-3

3-4

3-5

3-6

3-7

3-8

1014.2 Gross Leak

Pass

1014.2 Fine Leak Rate x 1E-9 atm cc/sec

8.8

6.4

5.8

6.0

5.4

5.0

4.6

4.8

Nitrogen, v%

99.90

99.80

Oxygen, v%

<0.01

Argon, v%

<0.01

Water, v%

0.04

0.03

0.02

0.05

0.03

0.02

0.08

Carbon Dioxide, v%

0.08

0.05

0.04

0.02

0.06

0.04

0.12

0.08

Hydrogen

Proc. of SPIE Vol. 7206 720606-4

Helium

Fluorocarbon

Ammonia

Organics

Table 4. Internal gas composition for seven braze seal units, age from 10-19 years old.

18 leadd braze seal DIPs

40 lead braze DIP

84 lead PGA

16 lead Flatpack

Post seal treatment

no bake

1 hr p.s. bake

no bake

1/2 hr bake

none

None

Year sealed

1989

1991

1998

1990

Approx. age

19 y

17 y

10 y

18 y

S/N

4-1

4-2

4-3

4-4

4-5

4-6

4-7

1014.2 Gross Leak

Pass

1014.2 Fine Leak Rate x 1E- 9 atm cc/sec

5.6

4.8

4.4

4.0

4.6

9.6

Nitrogen, v%

98.60

98.50

99.00

97.60

96.80

94.20

99.90

Oxygen, v%

<0.01

Argon, v%

<0.01

Water, v%

0.01

0.02

0.04

0.02

Carbon Dioxide

0.05

0.07

0.03

0.02

1.11

0.06

Hydrogen

1.33

1.37

0.91

2.33

3.13

4.65

<0.01

Helium*

Fluorocarbon*

Ammonia*

Organics*

All 15 TO99 cans and braze seal units are clearly free of significant moisture, from either materials outgassing or ingress through leaks, throughout their entire lifetimes.

These units had eutectic substrate attach and no polymeric materials inside, which helps achieve and maintain moisture outgassing control over long periods of time.

The findings in Tables 3 and 4 were gratifying. Robust M&P rendered these package styles both dry and truly hermetic for long periods of time of practical significance to device reliability. The materials and technology remain available today.

The need for robust M&P cannot be over emphasized. As package internal volumes decrease, the necessity for a truly hermetic seal rapidly increases. For example, a device with an internal volume of 0.1 cc cannot exceed leak rates of 10-10 sccm if service lives exceeding 10 years are to be reached. 1

3.2 Group 2, 5011 Adhesive Qualification Units

All units in this group contain polymeric substrate attach materials which require careful choice and robust pre-seal processing to insure moisture control. There were 71 total units, of which 18 were noncompliant and 53 compliant on moisture. Table 5 summarizes results for the noncompliant units.

Proc. of SPIE Vol. 7206 720606-5

Table 5. Noncompliant units in 5011 adhesive qualification study.

Seq. S/N

5-1

5-2

5-3

5-4

5-5

5-6

5-7

5-8

5-9

Study ID

C3-45

A22-14

A38-2

C6-B5

C5-19

C5-20

C5-B8

C2-63

A38-1

Nitrogen

74.00

90.90

59.00

93.10

90.90

90.20

95.20

95.40

95.30

Oxygen

19.40

0.02

0.52

2.39

Argon

0.83

0.11

0.68

0.16

0.18

0.04

.11

Water

1.68

2.50

11.70

1.68

0.68

0.91

1.10

0.54

2.33

Carbon Dioxide

0.41

0.88

3.93

0.48

3.71

6.37

0.16

0.27

2.02

Hydrogen

0.33

0.29

0.09

.03

Helium

4.83

2.19

4.01

1.68

3.41

Fluorocarbon

3.65

0.76

24.00

Ammonia

0.07

0.09

Organics

0.01

0.14

0.33

0.24

3.82

0.20

Table 5, continued

Seq. S/N

5-10

5-11

5-12

5-13

5-14

5-15

5-16

5-17

5-18

Study ID

C2-64

C5-18

A22-4

A22-5

A22-6

A22-12

A22-13

A37-1

A38-3

Nitrogen

95.20

91.80

91.50

91.30

91.20

92.60

91.20

99.00

96.00

Oxygen

Argon

0.06

0.03

0.02

0.07

Water

0.52

0.71

0.96

1.13

0.98

0.96

0.86

0.63

1.71

Carbon Dioxide

0.27

3.12

0.73

0.78

0.76

0.69

1.24

0.21

2.08

Hydrogen

0.32

0.02

Helium

3.80

6.69

6.75

6.99

5.65

6.57

Fluorocarbon

Ammonia

0.04

Organics

3.99

0.20

0.04

0.03

0.09

0.16

S/Ns 5-1, 5-2, and 5-3 contain components of air and fluorocarbon. This defines these units as “variable” or “one-time” leakers, as discussed later. These units are a special case of induced non-hermeticity. They are most likely hermetic except during stresses of burn in or leak test.

S/N 5-4 contains O2 and Ar in a ratio that, while not identical to air, probably identifies it as a leaking device. Its He content is less than half that of a brother unit (data not shown), and it contains no FC. However, its CO2 content at 0.48v% is too high to be explained by air ingress alone, so outgassing is also occurring in this unit. The 14 remaining noncompliant units S/Ns 5-5 through 5-18 contain no detectable O2 and negligible Ar (though three have somewhat elevated Ar). All contain some level of volatile organics, and most contain CO2 at concentrations far above that of natural air. These units are noncompliant due solely to materials outgassing.

None of the 53 compliant units contain O2 and most contain no Ar. Absence of O2/Ar indicates that up to the time of analysis none of the units had begun to acquire moisture by air ingress.

Units with polymeric materials tend to have levels of CO2 and organics that track with moisture content. This is reflected in Table 6, showing that on average compliant units contain only about one-third as much as much CO2 and hydrocarbons as noncompliant units, further indicating outgassing as the source of moisture in units in this group.

Proc. of SPIE Vol. 7206 720606-6

Table 6. Comparison of gas compositions between compliant and noncompliant units.

Twelve of the 53 compliant units contained moisture between 0.40-0.50v%. Any loss of hermeticity, or any additional outgassing, would quickly push those units above the recommended maximum moisture content. This underscores the importance of qualifying polymeric materials and their processing with respect to their moisture behavior.

Thus in Group 2, 17 of 18 noncompliant units contain moisture due solely to outgassing from materials. One unit showed evidence of air ingress, however outgassing also contributed to its total moisture content.

3.3 Group 3, Database of Miscellaneous Part Types

This was a group of 114 units of a wide variety of package styles and sizes. It had 41 units (from 31 different lots) noncompliant on moisture content. The gas composition “signatures” of the 41 noncompliant parts fell into four distinct categories, shown in Table 7.

Table 7. Categories of nonconforming units among 100 unit (31 sample group) database.

Category 1 is that of probable leakers, in which O2 and Ar are present. Seven of the 41 noncompliant units exhibit this “signature”, as summarized in Table 8.

Table 8. Apparent non-hermetic units, those with air and no FC.

Number of units

Average H2O v%

Average CO2, v%

Average HC, v%

Noncompliant, >0.50v% H2O

1.75

1.60

0.69

Compliant, <0.5v%

0.23

0.66

0.22

Category

Status

No. Units

1. Nonconforming units with components of air

Non-hermetic enclosure

2. Nonconforming units with no components of air

Outgassing materials

3a. Nonconforming units with components of air and FC

Probable one-time leaker

3b. Nonconforming units with no air, but FC

Definite one-time leaker

Total

Gp 23

Gp 27

Gp 28

Gp 42

Gp 44

Gp 31

TO-OC-4

TO257-132

Weld unit

Cer FP-1

Cer. FP-2

S/N

8-1

8-2

8-3

8-4

8-5

8-6

8-7

Nitrogen

93.90

96.80

90.40

78.30

81.66

77.50

76.90

Oxygen

3.55

1.63

4.11

18.60

16.44

20.50

21.20

Argon

1.15

0.14

0.40

0.82

0.62

0.93

0.96

Water

1.04

0.80

1.96

1.68

1.02

0.81

0.86

Carbon Dioxide

0.32

0.40

2.07

0.66

0.03

0.28

0.13

Hydrogen

0.03

0.01

Helium

0.24

1.08

Fluorocarbon

Ammonia

Organics

S/N’s 8-4 through 8-7 contain more-or-less stochiometric air. The others contain O2 and Ar in “diluted” amounts, with O2/Ar ratios from 3.1 to 11.6, unlike that of natural air. But the O2/Ar presence probably indicates at least some air

Proc. of SPIE Vol. 7206 720606-7

ingress (though the He in units 8-2 and 8-3 begs that question). Outgassing contributed to the levels of moisture in all units except 8-5, as CO2 is elevated far above that of natural air in the others. These seven units are classified as containing at least some moisture due to air ingress, per the considerations of this study.

Category 2 was a group of 20 units that are noncompliant, but contain negligible or no components of air, and no fluorocarbon. Not all the results are shown, but all the units had gas composition signatures like the typical examples in Table 9.

Table 9. Typical results for the 27 nonconforming units containing no air or FC.

C2- 63

C2- 64

C5- 18

C5- 19

A22- 4

A22- 5

Metal Can C

06- TO- 143

06- TO- 214

T-1

T-2

S/N

9-1a

9-1b

9-2a

9-2b

9-3a

9-3b

9-4a

9-4b

9-5

9-6a

9-6b

Nitrogen

95.40

95.20

91.80

90.90

91.50

91.30

94.20

94.30

92.90

96.00

96.40

Oxygen

Argon

0.06

0.03

0.02

0.01

0.13

Water

0.54

0.52

0.71

0.68

0.96

1.13

0.73

1.03

2.69

1.50

1.21

Carbon Dioxide

0.27

3.12

3.71

0.73

0.78

0.12

0.13

1.16

0.01

Hydrogen

0.32

0.33

0.27

0.01

0.19

0.15

0.12

Helium

3.80

4.01

6.69

6.75

4.71

4.56

2.90

2.31

2.29

Fluorocarbon

Ammonia

0.04

0.07

Organics

3.82

3.99

0.20

0.33

0.04

S/N’s 9-1a and 9-1b are brother units whose gas compositions are very similar. While they are barely noncompliant, it is clear that excess moisture is from organic compounds in the enclosures. S/N’s 9-2a and 9-2b are also brother units with a similar situation, but much more CO2. These units have also retained their He. S/N’s 9-3a and 9-3b are another set of brother units, with higher moisture yet lower organics and lower CO2. Clearly, the differences between S/N groups 9-1, 9-2, and 9-3 are the post-seal outgassing behavior of the organic materials inside the enclosures. Failure to properly cure these materials allows post-seal outgassing to elevate enclosure moisture content.

S/N’s 9-4 through 9-6 are a variation on the theme. Organics do not seem to play a role in their gas compositions, yet all are noncompliant, exceeding the moisture limit by factors ranging from 1.5x to 5.4x. This is attributed to poor pre-seal baking of package piece parts, failing to permanently remove adhered moisture, which subsequently outgassed from part surfaces.

Category 3 is a group of 14 units that are noncompliant and contain fluorocarbon. Within this group, subgroup 3a has 9 that also contain components of air, and subgroup 3b has 5 that do not contain components of air. Table 10 shows typical results for these kinds of units.

Table 10. Typical results for noncompliant units that contain fluorocarbon and no air (S/Ns 10-1 through 10-4), and those that contain fluorocarbon and air (S/Ns 10-5 through 10-6c).

FP3

T10-06-3

T10-06-4

T3-4

A22-14

A38-2

SB-AP-23

SB-AP-87

SB-AP-95

S/N

10-1

10-2a

10-2b

10-3

10-4

10-5

10-6a

10-6b

10-6c

Nitrogen

93.60

93.80

94.10

93.60

90.90

59.00

81.30

80.00

72.00

Oxygen

0.01

0.02

0.52

15.30

16.20

8.91

Argon

0.02

0.09

0.07

0.11

0.68

0.70

0.72

0.40

Water

0.86

3.22

2.75

0.86

2.50

11.70

1.18

1.13

1.07

Proc. of SPIE Vol. 7206 720606-8

Carbon Dioxide

0.15

0.30

0.29

0.15

0.88

3.93

0.22

0.61

0.96

Hydrogen

0.04

0.02

0.55

Helium

4.57

2.49

2.71

4.57

4.83

0.03

0.09

0.36

Fluorocarbon

0.29

0.07

0.10

0.29

0.76

24.00

1.28

1.25

16.30

Ammonia

Organics

0.01

0.14

3.4 Pressure-Sensitive Leakers

Three units in Group 2 and 14 units in Group 3 resemble those with glass-to-metal seals having variable leak rates,6 e.g. units with leaks apparently induced by external thermal or physical stresses such as clamping during burn-in which temporarily breaks the oxide-sealed surfaces in glass-to-metal seals. Table 11 contains data typical of units with pressure sensitive leaks described in that study.

Table 11. Noncompliant units, which contain FC in the pressure-sensitive leaker study.

Gp 49

Gp 50

Clarke 425

Clarke 432

Clarke 20

Nitrogen

94.60

92.10

82.20

Oxygen

0.34

2.58

Argon

0.53

0.05

0.17

Water

0.77

0.78

0.81

Carbon Dioxide

0.04

0.03

1.06

Hydrogen

Helium

3.90

6.70

12.20

Fluorocarbon

0.08

0.04

0.88

Ammonia

0.13

Organics

0.01

0.02

Because of this characteristic, noncompliant units in Sample Group 2 or 3 that contain fluorocarbon and/or a gas signature like that in Table 10 are not counted as “on-the-shelf” or “in-service” units with moisture elevated by normal air ingress.

3.5 Moisture From Chemical Reactions

Within-enclosure chemical reactions, such as H2 outgassing7 and H2O production by reaction with surface-exposed oxides8, or reaction of H2 and O2 in the presence of nickel as a catalyst producing H2O9, are special cases of the outgassing mechanism rather than a consequence of air ingress. Those mechanisms may have contributed to excess moisture in some of the samples, but they were not studied in detail.

3.6 Peculiar Data

Not all gas compositions are readily explainable. Table 12 shows some examples to challenge the reader. The data on the explanted hermetic biomedical device was obtained from a lawsuit filing in public record. The device contained negligible air, or at least not enough to explain 33v% H2O. (Not all data were reported in the filing). How can so much water be inside this unit? The large TO’s, A and B, are brother units. A has the right amount of Ar for air, but no O2. The level of moisture is too high to be explained purely by air ingress. Was the O2 consumed by some kind of chemical reaction that produced water? The organic substance reported was specifically methanol, which implies a chemical reaction of O2 with CH4 to make water and methanol (no balanced reaction is evident unless H2 is available). Unit B also has much water but a completely different gas signature. The LCC is noncompliant by a small amount, but has 23 times

Proc. of SPIE Vol. 7206 720606-9

as much CO2 as air. Where did all the CO2 come from, and is that the explanation for noncompliance? Proposed explanations for these data are solicited from the reader.

Table 12. Peculiar hermetic enclosure gas compositions.

Explanted Biomed device

Large TO A

Large TO B

48 ld LCC

Nitrogen

70.60

87.20

98.40

Oxygen

0.17

Argon

0.91

Water

32.91

24.70

10.10

0.67

Carbon Dioxide

0.40

0.06

0.91

Hydrogen

0.04

0.07

Helium

13.55

2.56

Fluorocarbon

Ammonia

Organics

3.77

4.1 Results

nr=not reported

4. RESULTS & CONCLUSIONS

1) Fifteen TO can and braze seal units more than 20 years old maintained negligible internal moisture due to robust materials processing that prevents both outgassing and air ingress.
2) Seventy-one units from a Test Method 5011 adhesive qualification study had 18 units noncompliant on moisture content. One of these contained an internal gas signature consistent with air ingress. Air ingress did not account for all the moisture in that unit.
3) One hundred fourteen units of a wide variety of package styles and sizes had 41 units noncompliant on moisture content. Seven contained an internal gas signature consistent with air ingress, but only one of these contained air exclusively.

Table 13 summarizes the results with respect to sources of internal moisture.

Sample Group

Number of Units

Noncompliant Units

Noncompliant: Air ingress

Noncompliant: outgassing exclusively

Noncompliant: variable leak

1. Units 10-20 years old

2. 5011 Adhesive Qual

3. Consultant’s database

114

Totals

200

Percentages

29.5% of units are noncompliant

13.6% of the noncompliant units

57.6% of the noncompliant units

28.8% of the noncompliant units

This unit showed evidence of outgassing also.
Only one of these units showed air ingress exclusively.

Proc. of SPIE Vol. 7206 720606-10

4.2 Conclusions

Overall, 57.6% of noncompliant units in this study contained elevated moisture due solely to materials outgassing with no evidence of air ingress. An additional 28.8% of the units behaved as variable leakers, a condition not attributed to simple air ingress during storage or service. Only 13.6% of noncompliant units showed evidence of air ingress, and only one of those appeared to contain air exclusively with no evidence of outgassing.

Piece part and materials selection and robust processing are the essential first line of defense for internal moisture control of sealed enclosures. Concern for hermeticity becomes important only after materials and processes are in place to assure that product is dry as-sealed and is as free as possible of outgassing.

It is recognized that certain package types, sealing equipment, suppliers of materials and equipment, and seal processes may be unique special causes of hermeticity issues. Engineers must address these special causes, while maintaining control of materials outgassing, as an integral part of investigating and eliminating special causes of hermeticity failure.

It is recognized that this is a small dataset. Much larger databases are available. The authors solicit inputs from anyone who can share data with the community as a whole to enlarge this study.

ACKNOWLEDGEMENTS

The authors wish to extend our thanks to our colleagues in industry who supplied the archival packages for analysis and the support of Oneida Research Services who supplied the necessary laboratory testing of the “old” units in group 1.

REFERENCES

1. Kullberg, R. C. and Lowry, R. K., “Hermetic package leak testing re-visited”, IMAPS International Conference and Exhibition on Device Packaging, (March 17-20, 2008)

2. Pernicka, J. C., “Cumulative helium leak detection (CHLD); first new test method for testing hermetic packages in over 30 Years, case studies of previously manufactured parts that failed after years on shelf”, Military, Aerospace, Space, and Homeland Security Topical Workshop and Exhibition, Baltimore, MD. (May 7-10, 2007)

3. http://scrippsco2.ucsd.edu/program_history/keeling_curve_lessons.html

4. Military Standard MIL-STD-883 Method 5011, “Evaluation and acceptance procedures for polymeric materials,” Columbus, OH, (31 October 1995)

5. Schuessler, P. W. private communication.

6. Clarke, R. A. and DerMarderosian, A., “Variable leak rate phenomena in glass to metal seals”, International Symposium on Microelectronics, 828 (1998)

7. Schuessler, P. W. and Gonya, S. G., “Hydrogen desorption from base and processed packaging alloys”, Proceedings, RL/NIST Workshop on Moisture Measurement and Control for Microelectronics, 67 (April 5-7, 1993)

8. Ellingham, H.J.T., “Reducibility of oxides and sulfides in metallurgical processes”, J. Soc. Chemical Industry, Transactions and Communications (May, 1944)

9. Gorodejskii, Sobyanin, and Bulgakov, “Low temperature reaction of hydrogen with preadsorbed oxygen on iridium surfaces”, Surface Sciences Vol. 82, 120-138 (1979)

Proc. of SPIE Vol. 7206 720606-11

IMAPS 2009

Advanced Getter Materials for GaAs, RF/MW, MEMs and Other Microelectronic Packages

Richard C. Kullberg & Bradley L. Phillip, Vacuum Energy Inc.1 &
Timothy J. Shepodd, Sandia National Laboratories2,3

INTRODUCTION

Various getter materials have been developed over the years to deal with hydrogen and moisture caused problems in microelectronic packages. These materials have ranged from adaptations of classical metal getter systems, to the currently prevalent PdO/desiccant mixtures in silicone RTVs, to esoteric metal thin film structures. All of these approaches have been successful when used appropriately, but all of them have also presented implementation difficulties to the packaging engineer. In addition, other issues can arise during a package’s service life. A brief overview of these traditional materials and the design concerns they raise will be presented.

To meet community driven needs and concerns in regards to the primary family of getters used in microelectronic packages, i.e. the silicone RTV mixtures with PdO and desiccants, a new class of getter materials have been developed as a drop in functional replacement for them. Key design goals for these materials has been to increase the capacities for hydrogen and water, while minimizing outgassed VOCs, which is of particular importance to optical applications, while eliminating corrosion inducing ionics and lowering the attainable vapor pressures for moisture inside the package. In addition, the material system allows the incorporation of RF absorbers.

IMPLEMENTATION ISSUES

To date the major approaches to gettering hydrogen and water from microelectronics have been based on traditional metal getters used in vacuum systems, mixtures of PdO and desiccants in silicone RTVs, and various thin film structures deposited by PVD techniques. All of these approaches have created implementation issues for the packaging engineer:

Traditional metal getters are exemplified by the classic barium ring getter used in CRT TV sets. Metal getter systems need to be used in a vacuum or noble gas environment and require a high temperature activation step. Mounting the getter is often an issue.

PdO/desiccant mixtures, while conceptually easier to implement, since they will operate in normal fill gases, still need an activation step at an elevated temperature in order to remove existing moisture from the desiccants used. In addition, the use of desiccants to capture water formed as part of the hydrogen gettering process is mandatory in these materials as hydrogen is gettered by the following reaction:

PdO+H2 ->Pd+H20
It is somewhat incongruous to put a material into a water sensitive device that creates water itself.

1 Richard C. Kullberg & Bradley L. Phillip, Vacuum Energy Inc., 2714 W. Park Blvd., Shaker Heights OH 44120,

216-991-7000, rckullberg@vacuumenergyinc.com & bradlphillip@vacuumenergyinc.com

2 Timothy J. Shepodd Ph.D., Materials Chemistry Department 8223, Sandia National Laboratories, 7011 East Ave., MS9403, Livermore, CA 94551-0969, 925-294-2791, tjshepo@sandia.gov

3 Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Co., for the United States Department of Energy’s National Nuclear Security Administration under Contract No. DE-AC04-94AL85000.

In order to deal with these issues recent work has been done to create thin film metal structures that pump hydrogen without creating water vapor. While these films are excellent hydrogen getters, they do nothing to pump water from a package and can be both difficult to integrate into a package design, as well as expensive.

DESIGN GOALS

Based on the existing issues with current hydrogen and water getters for microelectronic packages a simple set of design goals was established:

no activation
irreversible pumping of hydrogen & water in any sort of normal operating conditions
meet or exceed the requirements MIL-STD-883, Test Method 5011.4 for minimizing corrosion inducing ionics in polymer based materials used in microelectronic packages
minimize outgassing of any organics into the headspace of the package

MATERIAL CONCEPT

Extensive work has been done over the years at Sandia National Laboratories to develop hydrogen gettering materials for use in many different defense & industrial applications. The resulting materials, which work via hydrogenation, are completely passive, completely irreversible with no possibility of hydrogen re‐emission, and non pyrophoric. They efficiently scavenge hydrogen down to the sub‐ppm range in the cold conditions of the Arctic, as well as at temperatures as high as 300 ̊C.

In practical terms this means that compounds can be developed to getter hydrogen from microelectronic packages at room temperature that 1) do not require a thermal activation step, 2) will operate in air or other atmospheres, and 3) irreversibly sorb the hydrogen without creating water as a by product.

With these compounds in hand, gettering water from a package becomes a much simpler task as any moisture getter present need only deal with water already present in the system. Such a water getter becomes simple to define in that it should have a reasonably high capacity for moisture, not need any heat treatment to activate it, and pump water irreversibly under the desired operating conditions.

In some cases the presence of a desiccant is a by product of the hydrogen gettering chemistry used and additional desiccant is added to sorb any additional water present in the package:

PdO+H2 ->Pd+H20

Typical desiccants used in these scenarios are zeolites with an equilibrium water vapor pressure in the range of 10-3 torr, which can be an appreciable amount of water vapor. A downside to the use of zeolites is that they need a bakeout step in order to prepare them for use. This step is typically accomplished at some phase of the material’s introduction into the package.

As package volumes decrease the acceptable water concentrations inside the package decrease well below the 5,000 ppmv called out in MIL-STD-883. 5,000 ppmv is a historical compromise designed to prevent the formation of 3 monolayers of water on surfaces inside a typical package of the day. 3 monolayers is a critical metric as it is the boundary condition above which ionics begin to induce corrosion processes on a surface. It therefore becomes critical to not only remove the hydrogen from a package in a manner that doesn’t generate water as a byproduct, but to use a desiccating agent that operates at much lower vapor pressures.

A typical desiccant considered in these scenarios is anhydrous CaO, which reacts with water in an irreversible manner under normal operating conditions:

CaO + H2O -> Ca(OH)2

CaO provides much lower water vapor pressures that range from 10-11 torr at 0C to 10-9 torr at room temperature to 10-5 torr at 100 C. Consequently, under normal room temperature operating conditions, the water vapor pressure is 6 orders of magnitude lower than that of a typical zeolite.

Lastly, any resulting gettering system should be able to be integrated in to a package using normal industrial methods. I.E. the getters should be able to be mixed into various polymer substrates in order to facilitate their use as molded forms, pastes, or dispensable liquids depending on the preferred method of integration into a package.

MATERIAL PERFORMANCE VERIFICATION

Hydrogen Sorption Measurement

While hydrogen sorption testing can get quite complicated, particularly when one is studying metal based getter systems, polymeric hydrogen gettering systems lend themselves to simple, yet still quantitative approaches that can be implemented even in incoming inspection scenarios.

There are two basic approaches to simple sorption measurement. These are based on the getter’s pumping action changing either the pressure or the volume of a system. These are often referred to as the ∆P or ∆V methods. For this development effort we used the ∆V method, as it is the simplest of all and takes advantage of the insensitivity of polymeric getters to water exposure.

Simple schematic of ∆V set up.

To implement the ∆V method cut off a piece of getter material with a mass of approximately 0.5 g. Note the actual mass. Wrap a thin copper wire around it and suspend it in a graduated cylinder. Place the cylinder in a vessel containing water and allow it to fill. Invert the cylinder without re-exposing it to air. Put the tubing from the hydrogen source into the vessel and flush with hydrogen for a few seconds. Then bubble a few tens of cc’s of hydrogen at a pressure of 1 atm into the inverted cylinder and make note of how much gas is captured. Note the starting volume of gas and gather data over time until the captured volume of gas no longer decreases. The slope of the curve will tell you the pumping speed of the getter sample and the total volume of hydrogen pumped will tell you its capacity per unit mass.

Water Sorption Measurement

Verifying the performance of proposed materials for water sorption is a simple matter of measuring the change in mass of a sample over time. The purpose of such testing is more of a snap gauge test to ensure that the amount of water gettered approaches the stoichiometric endpoint of the reaction, rather than a detailed study of the reaction itself and any environmental impacts caused by the operation ambient. In other words, desiccants are well understood and all that is necessary for the current work is performance verification. To that end samples were exposed to normal laboratory ambients and the increase in mass over time measured with a normal electronic lab balance.

RESULTS

For the purposes of this paper sample material was prepared containing both a proprietary hydrogen getter material and CaO as a desiccating agent. Testing on candidate materials began with exposure to atmospheric moisture to verify the water gettering ability of the samples. Once water sorption performance was measured the samples were tested for hydrogen capacity as well as ionics content.

Water

CaO desiccants have a high capacity for water. Stoichiometrically, one mole of CaO will getter one mole of water, or for every 3g of CaO 1g of water will be gettered. For testing purposes a small batch of getter material was prepared with 5.5 wt % CaO for sorption testing. The amount of desiccant added was purely arbitrary and can be engineered over a broad spectrum to meet customer needs. At the end of the test the sample gained 2 wt%, or within the theoretical water capacity of the CaO used.

Performance of a polymeric getter structure with CaO desiccant.

As can be seen from the data, appreciable water pumping speed is available but there is also adequate ‘throttling’ of water diffusion into the getter structure to allow reasonable working times.

Hydrogen

A 0.5g sample of the original polymeric getter structure, after exposure to water vapor, was tested for hydrogen sorption performance via the ∆V method.

Performance of a 0.5g sample after moisture saturation.

As can be seen, even after gettering water to capacity and subsequent exposure to liquid water as part of the ∆V method set up, the getter was on track to its nominal 50 cc/g hydrogen capacity when the test was ended.

Ionics

Testing materials for ionics per MIL-STD-883, Test Method 5011.4 is an established procedure and is supplied by DSCC certified laboratories. For this talk, rather than reinvent the wheel, we had samples tested by Pacific Testing Laboratories, who is DSCC certified:

MIL-STD-883G Requirements

Vacuum Energy Inc. PolyGetterTM 1B

Vacuum Energy Inc. PolyGetterTM 6D

Total ionic content: ≤4.5 mS/m

Total ionic content: 2.23 mS/m

Total ionic content: 2.22 mS/m

Hydrogen: 4.0 < pH < 9.0

pH = 7.9

pH = 7.5

Chloride ≤ 200 ppm

3.9 ppm

1.4 ppm

Sodium ≤ 50 ppm

0.8 ppm

4.2 ppm

Potassium ≤ 50 ppm

< 0.5 ppm

Fluoride ≤ 50 ppm

< 0.5 ppm

Outgassing

At this time outgassing testing is still underway and is planned to incorporate concepts incorporated in standards and guidelines currently under development by SEMI. Consequently we will report on this work in a future paper.

SMTA 2010

HARSH ENVIRONMENTS AND VOLATILES IN SEALED ENCLOSURES

Robert K. Lowry
Consultant, Electronic Materials
Phone/Fax 321-777-9949 rlowry98@aol.com
Electronic materials chemical, physical, and failure analysis

Richard C. Kullberg Vacuum Energy, Inc. Phone/Fax: 719-966-4296 rckullberg@gmail.com Getters for sealed enclosures

Daniel J. Rossiter
Oneida Research Services Phone: 315-736-5480
Fax: 315-736-9321 djrossiter@ors-labs.com Sealed enclosure analysis services

HARSH ENVIRONMENTS AND VOLATILES IN SEALED ENCLOSURES

OUTLINE • HermeticEnclosures

• HowCanASealedEnclosureBeAHarshEnvironment?

• HarshEnvironmentsandFailureModes • Water

• Oxygen
• Hydrogen
• Ammonia
• Hydrocarbons
• Carbondioxide;Argon • Vacuum
• Air

• AvoidingHarshEnvironmentsinHermetics

HERMETICALLY SEALED ENCLOSURE

●Headspace volume can range from thousands of liters to nanoliters ●There is gas in the headspace

VOLATILE IMPURITIES IN HERMETIC HEADSPACE

A B1C B2

A: Fill gas impurities: moisture, O2, volatile hydrocarbons
B1: Offgassing by package piece parts: moisture, H2
B2: Offgassing by assembly materials: moisture, volatile hydrocarbons, NH3, reactive gases C: Permeation: moisture, O2
D: Ingress via poor seal integrity: moisture, O2, SO2, etc.

To mass spec, E-8 Specimen temp 100±5oC Pre-bake 24hrs at 100oC

ANALYZING SEALED ENCLOSURE HEADSPACE GAS

MASS SPECTROMETRY

INTERPRETING HEADSPACE GAS COMPOSITIONS COMPOSITION OF AIR IS THE BENCHMARK

Nitrogen

Oxygen

Argon

Water

Species

Carbon Dioxide

Neon

Helium

Methane

Hydrogen

Volume Percent

78.084 ± 0.004

20.946 ± 0.002

0.934 ± 0.001

0.1-3.0+

0.038

0.0018

0.0005

0.0002

0.00005

Constant

Comment

Variable, f(RH, temp, P)

Constant

Not from air if detected

MOISTURE IN PACKAGES Moisture-related failure modes

– Condensate (liquid phase) • Chemical corrosion

• Optics fogging

– Adsorbate (molecular monolayers) • Electrical leakage, instability
• Dendritic growth

– Vapor (gas phase) • Stiction

• Vacuum deterioration

INTERNAL WATER VAPOR LIMIT

3 monolayers H2O adsorbate supports surface electrical conduction

Condensation dewpoint at 1 atm is -2oC

0.5v%

Threshold for adsorbing 3 monolayers H2O molecules

Measurable with some degree of precision by mass spectrometers;

no circulatable primary standard

POWER MANAGEMENT HYBRID

IN-FLIGHT ON ISS, 5 YEARS, NO FAILURES TYPICAL INTERNAL GAS CONTENT

Species

Hybrid

Nitrogen

99.80

Oxygen

<0.01

Argon

<0.01

Water

0.02

Carbon Dioxide

0.01

Hydrogen

0.19

Helium

<0.01

Hydrocarbon

<0.01

The GOLDEN unit!

SILVER DENDRITES

FLATPACK SILVER DENDRITE SHORT

Species

Flatpack

Nitrogen

93.90

Oxygen

3.55

Argon

1.15

Water

1.04

Carbon Dioxide

0.32

Hydrogen

0.03

Helium

<0.01

Hydrocarbon

<0.01

CORRODED METAL STRIPES ON FR4 IN A HIGH-MOISTURE SEALED ENCLOSURE

METAL PACKAGE EPOXY LID SEAL

Species

Epoxy Seal

Nitrogen

53.50

Oxygen

<0.01

Argon

0.13

Water

39.00

Carbon Dioxide

2.59

Hydrogen

<0.01

Helium

<0.01

Hydrocarbon

<0.01

HARSH ENVIRONMENT SPECIES

MOISTURE All values in volume percent

Species

Flatpack; Aluminum Corrosion

Telecom Device Electrical Leakage at >0.7v%

Med Device

Operating Housing

Nitrogen

96.00

98.60

76.10

Oxygen

<0.01

21.00

Argon

<0.01

0.01

0.93

Water

1.50

0.72

1.46

Carbon Dioxide

0.01

0.04

0.08

Hydrogen

0.15

0.66

<0.01

Helium

<0.01

0.40

Hydrocarbon

<0.01

HARSH ENVIRONMENT SPECIES OXYGEN

THRESHOLD CONDITION ≥0.20v%

• Enterscracksorpinholesinsolderattachingdietosubstrate
• Oxidizesmechanicaldefectsurfaceswithinthesolder
• Introducescoefficientofthermalexpansiondifferencestothe bulk of the solder attach
• Causessoldereddietoliftfromsubstrateduringthermal excursions due to CTE mismatches

HARSH ENVIRONMENT SPECIES

OXYGEN All values in volume percent

Species

Unit A, No fail

Unit B, Lifted Die

Nitrogen

95.78

92.96

Oxygen

<0.01

1.33

Argon

<0.01

0.14

Water

<0.01

0.35

Carbon Dioxide

0.01

<0.01

Hydrogen

4.33

0.15

Helium

<0.01

5.23

Hydrocarbon

<0.01

HARSH ENVIRONMENT SPECIES HYDROGEN

THRESHOLD CONDITIONS Variable for device effects 4.0v% for LEL in air

• “Poisons” MOS and compound semiconductor devices. Tolerable level varies with device design and other factors but effect can occur with as little as 0.1v%
• Reduces surface metal oxides within enclosures, producing moisture as by-product
• Lower explosive limit in air is 4v%

HARSH ENVIRONMENT SPECIES

HYDROGEN All values in volume percent

Species

MOS Unit A1, unstable threshold voltage

MOS Unit A2, Stable

Air-sealed Marine Nav Aid

Nitrogen

95.41

99.46

64.79

Oxygen

<0.01

17.08

Argon

<0.01

0.84

Water

0.06

0.31

3.75

Carbon Dioxide

0.21

0.22

1.12

Hydrogen

4.33

<0.01

12.50

Helium

<0.01

Hydrocarbon

<0.01

HARSH ENVIRONMENT SPECIES AMMONIA

THRESHOLD CONDITIONS Any is too much

• Raises pH of any liquid moisture enhancing corrosion potential
• Promotes silver dendrite growth
• Avoid by not using organic adhesives containing dicyandiamide curing agent

HARSH ENVIRONMENT SPECIES

AMMONIA All values in volume percent

Species

Hybrid 1
with Silver Dendrites

Hybrid 2 Dendrite-Free

Nitrogen

95.37

99.37

Oxygen

0.04

0.03

Argon

0.03

0.04

Water

0.19

0.35

Carbon Dioxide

0.34

0.15

Hydrogen

0.06

0.08

Helium

<0.01

Hydrocarbon

0.09

<0.01

Ammonia

2.57

<0.01

Methanol

1.31

<0.01

HARSH ENVIRONMENT SPECIES HYDROCARBONS

THRESHOLD CONDITION Varies, but preferably none

• Fog optical surfaces
• Collect on moving MEMs parts and interfere with mechanical function

HARSH ENVIRONMENT SPECIES

HYDROCARBONS All values in volume percent

Species

Sealed Optoelectronic Device

MEMs Device with Jammed Microgear

Nitrogen

85.10

95.90

Oxygen

<0.01

Argon

<0.01

0.05

Water

0.47

0.07

Carbon Dioxide

0.58

3.48

Hydrogen

<0.01

0.26

Helium

≈13.30

<0.01

Hydrocarbon

0.41*

0.18

Notes:

*Included methyl ethyl ketone, methanol, and tetrahydrofuran

Gears jammed with microresidues of behenonitrile and stearonitrile

HARSH ENVIRONMENT SPECIES CARBON DIOXIDE, ARGON

THRESHOLD CONDITION Unknown for carbon dioxide 60v%+ for argon

• Carbon dioxide is soluble in water condensate, lowers pH, rendering any condensate more chemically aggressive
• This mechanism for CO2-accelerated moisture corrosion mechanisms is postulated and has never been confirmed.
• High concentrations of argon support arc discharge. Do not seal any device that can have an ESD or any other arc or spark discharge event in pure argon.

HARSH ENVIRONMENT SPECIES VACUUM

THRESHOLD CONDITION Any increase in pressure

• Volatilazation or ingress of any gas molecules into a vacuum sealed headspace deteriorates vacuum quality intended to protect and enable operation of MEMs, quartz crystal oscillators, and other vacuum-sealed parts

HARSH ENVIRONMENT SPECIES

VACUUM All values in volume percent

Species

MEMs Device Function Good

MEMs Device Function Poor

Ion source pressure

3.0E-8 torr

1.4E-7 torr

Nitrogen

81.20

41.90

Oxygen

<0.01

Argon

1.86

1.44

Water

<0.01

49.30

Carbon Dioxide

7.17

6.23

Hydrogen

9.75

1.13

Hydrocarbon

<0.01

AVOIDING HARSH EXTERNAL ENVIRONMENTS THAT EXACERBATE HARSH INTERNAL ENVIRONMENTS

• Avoid post-seal process temperature excursions that exceed pre-seal cure temperatures
• Avoid post-seal process temperature excursions that exceed pre-seal bakes
• Avoid post-mount test temperature excursions that exceed the maximum pre-seal temperature
• Avoid post-mount field-service temperature excursions that exceed the maximum pre-seal temperature
• Avoid post-seal process mount handling, mechanical stress/shock, or vibration conditions that threaten lid seal or glass-to-metal feedthrough integrity → leaks
• Avoid post-mount field-service mechanical stress/shock or vibration conditions that threaten lid seal or glass-to-metal feedthrough integrity → leaks
• Utilize robust statistically-controlled process monitoring to insure that none of the above happens unexpectedly or accidentally

AVOIDING HARSH EXTERNAL ENVIRONMENTS THAT

EXACERBATE HARSH INTERNAL ENVIRONMENTS

Component Manufacturers

• Select package parts with attention to composition, surface chemistry, cleanliness, and outgassing behavior.
• Select attachment and protective materials with volatility properties suited to expected field service conditions.
• Deploy getters to capture and bind volatiles discussed in this paper.
• Assemble in clean environments with humidity control.
• Cure all organic materials completely per manufacturer recommendations
• Engineer post-cure pre-seal bakes to maximize loss of volatile materials before seal without altering material physical or chemical properties.
• Pre-seal bake assembled components, in vacuum <10mTorr if possible.
• Do not overload ovens or short-cut cycle times.
• Do not re-expose pre-seal baked components to ambient air.
• Seal in moisture-controlled equipment under pure N2 or He.

In-House/Contract Assemblers

• Fully inform customers of assembly and handling process conditions: soldering and baking temperature profiles, extremes of thermal cycling, mechanical stress, thermal or mechanical shock, vibration.
• Fully inform customers of product test conditions with respect to above- mentioned factors.
• Know users’ expected field service conditions.

SPIE 2010

Characterization of polymeric getter materials for MEMs/MOEMS and other microelectronic package service

Richard C. Kullberg1 & Bradley L. Phillip2
Vacuum Energy Inc., 2714 W. Park Blvd., Shaker Heights OH 44120

ABSTRACT

Various getter materials have been developed over the years to deal with hydrogen and moisture caused problems in MEMS/MOEMS packages. These materials consist of two major families, metal alloy systems and polymeric systems. Both systems have pluses and minuses for the MEMS/MOEMS packaging engineer. In order to determine applicability, careful characterization of these systems is critical.

Keywords

polymer getter desiccant ionics sorption hydrogen water gravimetric MIL-STD-883

1. INTRODUCTION

The advent of a new class of polymeric getter materials originally developed for other industries has brought a need to readdress the issue of how to characterize such materials for MEMS/MOEMS service. While MIL-STD-883 gives a degree of guidance, it is not the complete answer. Characterization of these materials has consequently been clean sheeted to address the key areas of hydrogen and water capacity determination, and determination of ionics concentrations. A critical driver for determining the methods to use was to replace expensive vacuum methodologies as expressed in ASTMs F 111-96[1] and F 798-97[2]. Our goal was to qualify simple methods using normal laboratory equipment available to the getter end user, either as a part of their own development work or as part of an ongoing quality assurance effort.

2. GETTER MATERIAL PERFORMANCE VERIFICATION

2.1 Hydrogen Sorption Measurement

1 719-966-4296, rckullberg@vacuumenergyinc.com 2 216-991-7000, bradlphillip@vacuumenergyinc.com

discuss the ∆V method, as it is the simplest of all and takes advantage of the insensitivity of polymeric getters to water exposure.

Simple schematic of ∆V set up.

To implement the ∆V method, prepare a sample of getter material with a mass of approximately 0.5 g. Note the actual mass. Wrap a thin copper wire around it and suspend it in a graduated cylinder. Place the cylinder in a vessel containing water and allow it to fill. Invert the cylinder without re-exposing it to air. Put the tubing from the hydrogen source into the vessel and flush with hydrogen for a few seconds. Then bubble a few tens of cc’s of hydrogen at a pressure of 1 atm into the inverted cylinder and make note of how much gas is captured. Note the starting volume of gas and gather data over time until the captured volume of gas no longer decreases. The slope of the curve will allow calculation of the pumping speed of the getter sample and the total volume of hydrogen sorbed indicates the getter’s capacity per unit mass.

2.2 Water Sorption Measurement

Verifying the performance of proposed materials for water sorption is a simple matter of gravimetric analysis, i.e. measuring the change in mass of a sample over time. The purpose of such testing is more of a snap gauge test to ensure that the amount of water gettered approaches the stoichiometric endpoint of the reaction, rather than a detailed study of the reaction itself and any environmental impacts caused by either the test or operational ambient. In other words, desiccants are well understood and all that is necessary for typical applications is performance verification. To that end samples are exposed to normal laboratory ambients and the increase in mass over time monitored with an adequately sensitive electronic lab balance. If some idea of the sorption rate of a desiccant is required, monitoring the relative humidity in the test space and/or doing the test in a volume controlled humidity can be considered.

2.3 Ionics

A common concern in microelectronic packaging is corrosion of internal components of the package induced by moisture borne ionic contaminants. This concern is specifically addressed in MIL-STD-883, Test Method 5011.4, EVALUATION AND ACCEPTANCE PROCEDURES FOR POLYMERIC MATERIALS, section 3.5.4 and 3.8.7[3]:

“3.5.4 Ionic impurities. The ionic impurity content shall be determined in accordance with 3.8.7 and shall meet the requirements specified in table II. Ionic content analysis shall be in triplicate for certification and single analysis for acceptance testing. Failure at acceptance shall require the passing of two additional samples.”

And

“3.8.7 Ionic impurities. A water-extract analysis shall be performed to determine the level of ionic contamination in the cured polymeric material. The total ion content (specific electrical conductance) and the specific ionic content for the hydrogen (pH), chloride, sodium, fluoride and potassium ions shall be measured. Other ions present in quantities > 5 ppm shall also be reported in ppm.”

The actual test method as called out in MIL-STD-883 consists of two primary steps; first a sample of the material is ground and placed in deionized water under controlled conditions to allow the ionic species to leach out of the sample. The electrical conductance of the solution is then measured with a conductivity meter to determine the total ionic content. Specific concentrations of the species present are measured with ion chromatography or an equivalent technique.

3. EXAMPLES

For the purposes of this paper sample material was prepared containing both a proprietary hydrogen getter material and CaO as a desiccating agent. This sample material was tested using the methods described in this paper. Testing on candidate materials began with exposure to atmospheric moisture to verify the water gettering ability of the samples. Once water sorption performance was measured, the samples were tested for hydrogen capacity as well as ionics content.

3.1 Hydrogen

A 0.5g sample of the original polymeric getter structure, after exposure to water vapor, was tested for hydrogen sorption performance via the ∆V method.

35 30 25 20 15 10

0
0 5 10 15 20 25 30 35 40 45 50

hours

Performance of a 0.5g sample after moisture saturation.

3.2 Water

CaO desiccants have a high capacity for water. Stoichiometrically, one mole of anhydrous CaO will getter one mole of water, or for approximately 3g of CaO 1g of water will be gettered. For testing purposes a small batch of getter material was prepared with 5.5 wt % CaO for sorption testing. The amount of desiccant added was purely arbitrary and can be engineered over a broad performance spectrum. At the end of the test the sample gained 2 wt%, or within the theoretical water capacity of the CaO used.

cc hydrogen gettered

Mass increase over time of a polymeric getter structure with CaO desiccant.

As can be seen from the data, appreciable water pumping speed is available but there is also adequate ‘throttling’ of water diffusion into the getter structure to allow reasonable working times.

3.3 Ionics

Testing materials for ionics per MIL-STD-883, Test Method 5011.4 is an established procedure performed by DSCC certified laboratories. For this talk, rather than reinvent the wheel, we had samples tested by Pacific Testing Laboratories, who is Defense Supply Center Columbus (DSCC) certified:

MIL-STD-883G Requirements

Vacuum Energy Inc. PolyGetterTM 1B

Vacuum Energy Inc. PolyGetterTM 6D

Total ionic content: ≤4.5 mS/m

Total ionic content: 2.23 mS/m

Total ionic content: 2.22 mS/m

Hydrogen: 4.0 < pH < 9.0

pH = 7.9

pH = 7.5

Chloride ≤ 200 ppm

3.9 ppm

1.4 ppm

Sodium ≤ 50 ppm

0.8 ppm

4.2 ppm

Potassium ≤ 50 ppm

< 0.5 ppm

Fluoride ≤ 50 ppm

< 0.5 ppm

4. CONCLUSION

Verifying the performance of critical active agents like getters in a MEMS/MOEMS package need not be a complicated task requiring expensive equipment and complicated procedures. So long as the chemistry of the getter material is understood it is often possible to develop simple tests to verify their performance and suitability for use.

REFERENCES

[1] ASTM F 111-96, Standard Practice for Determining Barium Yield, Getter Gas Content, and Getter Sorption Capacity for Bariaum Flash Getters, 1996.

[2] ASTM F 798-97, Standard Practice for Determining Gettering Rate, Sorption Capacity, and Gas Content of Nonevaporable Getters in the Molecular Flow Region, 1997.

[3] MIL-STD-883G, TEST METHOD STANDARD, MICROCIRCUITS, 28 February, 2006.

SPIE 2012

The Unsettled World of Leak Rate Physics:
1 Atm Large - Volume Considerations Do Not Apply to MEMS Packages, A Practitioner's Perspective

Richard C. Kullberg*a, Arthur Jonath*b, Robert K. Lowry*c
aVacuum Energy Inc., 2412 W. Park Blvd, Shaker Heights, OH 44120
bJonath Associates, 335 Golden Oak Dr., Portola Valley, CA 94028 cElectronic Materials Characterization, 3000 Purple Martin Lane, Indiatlantic, FL 32903

ABSTRACT

The world of leak testing, and the applicable physics, is unsettled. While globally lower MIL-STD leak rate criteria are under consideration even for 1 atm-large volume packages, industry is conversely moving rapidly into very small volume MEMS and vacuum packaging for advanced devices. These changes point out serious conceptual disconnects between the reality of properly characterizing a leak and the conceptual tools used to ensure the desired lifetime. The physical understandings and associated tool sets used to test and model the leaks are described. We modeled two actual packages, a large, ≈200 cc volume multichip module for aerospace applications and a small ≈0.01cc volume MEMS package for sensor applications. Impacts of various physical models of leak flow into a package are compared to include Fickian Diffusion, The Davy Model, Howl-Mann, and an empirically derived model based on Kr-85 leak testing as called out in the most recent edition of MIL-STD-883. As shown in the comparisons, simple He leak testing and physical models based thereon fall apart in the small volume MEMS packaging space.

Key Words: MEMS Packaging Leaks Fick Davy Helium Howl-Mann Krypton 1. INTRODUCTION

The world of leak testing, and the applicable physics is unsettled. While globally lower MIL-STD leak rate criteria are under consideration, even for 1 atm-large volume packages, industry is moving rapidly into very small volume MEMS and vacuum packaging for advanced devices. These changes point out serious conceptual disconnects between the reality of properly characterizing a leak and the conceptual tools used to ensure the desired lifetime. In fact, practical experience has repeatedly shown that other contributors to moisture mass flows within and without a package to include outgassing of materials and the actual packaging process are the major contributors to moisture issues.

For example, review of a private hermetic package database of one of the authors revealed the extent and nature of hermetic package moisture noncompliance.1 This database comprised both single chip microcircuit packages as well as some hybrid units. In the review, any unit exceeding 0.5v% internal water vapor was termed “noncompliant”. There were 69 noncompliant units with no components of air, another 69 units with approximate air composition, 30 noncompliant units with more moisture than simple air ingress can explain, and 35 noncompliant units that were gross leakers indicated by presence of fluorocarbon.2

* rckullberg@vacuumenergyinc.com; 719-966-4296; www.vacuumenergyinc.com arthur@jonathassociates.com; 650-851-8852; www.jonathassociates.com rlowry98@aol.com; 321-777-994; www.electronic-materials.com

Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and Nanodevices XI, edited by Sonia M. García-Blanco, Rajeshuni Ramesham, Proc. of SPIE Vol. 8250, 82500H © 2012 SPIE · CCC code: 0277-786X/12/$18 · doi: 10.1117/12.921305

Proc. of SPIE Vol. 8250 82500H-1

Table 1. Reasons For Moisture Noncompliance of 1,889 “Hermetic” Units

Category

Number of Units

Percent of total units

Compliant

1686

89.3%

Noncompliant, no air present

3.7%

Noncompliant, evidence of air leak

3.7%

Noncompliant, fluorocarbon present

1.9%

Noncompliant, small amount of air present

1.6%

Total Units

1889

Review of a larger database of over 10,000 units reported similar results, with a total of 14% of units noncompliant and a similar breakdown of non-compliants.3 The reviewed data shows that the general extent of moisture noncompliance in hermetic packages is surprisingly high. Humidity ingress through leaks, and post-seal internal materials volatility, are roughly equally responsible for moisture noncompliance.

2. CLASSICAL LEAK PHYSICS: MOISTURE INGRESS INTO HERMETIC PACKAGES

Even without a leakage path, moisture can be present inside sealed enclosures due to either moisture impurity in the blanket seal gases or volatility from package materials.4 Assuming that moisture from these sources both at time of seal and over unit storage and service life remains negligible (a perilous assumption), a failure of seal integrity then becomes a potentially significant source of moisture. Based on the assumption that a failure of seal integrity is the root cause for elevated concentrations of moisture and other species within a package, workers in the field have modeled leaks and leak physics in great detail.

3. LEAK MODELING

3.1 Primary Approaches to Leak Modeling

Context is critical to quantitatively assessing moisture flows through a capillary as the physics behind this topic has been massively confused by historical events and engineering rules of thumb developed in response to those events. This confusion is the root cause for leak testing, modeling, and regulatory efforts to be a current hot topic in the industry.

Commonly used approaches to understanding leaks into a package assume a positive flow of gas. Such flow is typically assumed to be pressure driven and is, in fact, actually so when traditional leak check methods are used. But is this truly applicable when it is considered that there are two pressure regimes to consider:

P1=P2 (1) P1≠P2 (2)

In a typical microelectronic package, not purposely sealed under reduced pressure, and in storage or installed in a system on the surface of the earth, P1=P2 applies. This puts the physics into a pure Fickian Diffusion mode, which is more appropriate than flow based physics to understand moisture transport into a typical microelectronic package.

The significant differences between diffusion-based mass flow and physical-pressure-based mass flow physics have not stopped practitioners in the field from trying to apply data obtained by flow based methods, e.g. leak testing, to model what is in actuality often a diffusion-based regime.

The three main approaches to modeling leaks into packages are Fick, Davy and Howl-Mann. We will give a brief overview of each. As the applicable physics for most devices is P1=P2, the authors view Fickian diffusion as the most

Proc. of SPIE Vol. 8250 82500H-2

important method, with flow based methods like Davy and Howl-Mann are special case tools. We shall also briefly touch on current work by Rossiter & Neff to generate an empirically derived relationship between krypton leak rates and permissible leak rates.

3.2 Fick’s Diffusion Physics

The authors’ preference for understanding ‘leaks’ into packages using Fick’s Law is based upon decades of analyzing real world hermetic systems, ranging in size from very small MEMS packages to huge cryogenic vessels. Analyzing failure modes across this broad range of systems does not agree with a leak or flow-based perspective of hermeticity.

Our analysis separates the data sets into two main families based on pressure regimes. Fundamentally, there is either a ΔP across the boundary between the inner cavity of the hermetic system and the outer ambient, or there is not. Where a true ΔP exists, typically when the internal cavity is under a vacuum, classic flow physics apply and the traditional approaches to calculating a mass flow through a leak are applicable. Focusing solely on leaks considers only air and water flows into the system. This ignores other potentially adverse mass flows due to mechanisms such as outgassing or permeation. While these flows are often the true root cause of a system’s failure to reach its designed lifetime, further discussion of them is outside of the scope of the current paper.

For the purposes of the current discussion a leak can be viewed as a very tiny capillary connecting the internal cavity of a package with the outer world. As the package has been sealed at the same atmospheric pressure as the outer world, P1=P2 applies. Diffusion physics determines how long it will take for an equilibrium to form between the partial pressure or concentration of moisture within a sealed cavity and that in the external ambient.

where:

(3)

m = molecular weight of the diffusing species (18 for water vapor) t = time
D = diffusion coefficient (for water, 2.4E-5 m2/sec at 20°C)
A = the smallest cross-sectional area of the leak path

L = the length of the leak path
C2 – C1 = the concentration difference between the cavity and the outside ambient. Where materials & processes are effectively executed, C1 = 0.

Inside the cavity C1 is dynamic, as it is diminished by physisorption on internal surfaces and augmented by outgassing from those surfaces or by influx from the external ambient.

Leaks described by Fick’s Law thus are driven by temperature, concentration (partial pressure) differences, and leak geometry (primarily leak path diameter). These are the conditions that apply to enclosures nominally at 1 atm after seal, when stored or operating in an ambient air environment.

Care should be taken when comparing the outcomes of diffusion-based models with flow-based models. As Davy writes:5

“... the rate at which helium leaks out of a package when it is being tested (under vacuum by a helium mass spectrometer leak detector) sets an upper limit to the rate at which air may leak into the same package while it is in use in air at one atm, but that upper limit may be orders of magnitude greater than the true value.”

3.3 Davy’s Combined Flow Equation

Davy made an effort to combine all of the flow regimes, viscous, molecular and diffusional in his combined flow equation (CFE):6

Proc. of SPIE Vol. 8250 82500H-3

P
1.249 10 ∆ 4.961 10 43 1.509 ∆

(4)

where:

R = the measured leak rate of one gas species through the capillary in atm-cc sec-1
Y & Z = viscous flow dimensionless parameters correcting for end effect and molecular slip l & d = length and diameter of the capillary in cm

P = the average pressure in atm
ΔP = the difference in total pressure in atm
Δp = the difference in partial pressure of measured gas species in atm λ = mean free path in cm

Pressure-driven mass flow regimes account for the first half of the CFE equation. However, since the pressures are equal on both sides of the leak, ΔP = 0, the results are determined by the diffusion-driven second half of the equation, where the dimensions of the leak are critical.

3.4 Howl-Mann Leak Rate Equation

In early work Howl and Mann derived an equation7,8 that is the basis for many leak rate considerations and supports the physics of helium and krypton leak rate testing specified in Mil-Std test methods:9

(5)

where:

R1 = the measured leak rate of tracer gas (He) through the leak in atm-cc sec-1of He L = the equivalent standard leak rate in atm-cc sec-1of air
PE = the pressure of exposure in atmospheres absolute of He
PO = the atmospheric pressure in atmospheres absolute

MA = the molecular weight of air in g (28.7 per Method 1014)
M = the molecular weight of the tracer gas in g
t1 = the time of exposure to PE in sec
t2 = the dwell time between release of pressure and leak detection, in sec V = the internal volume of the device package cavity in cc

Howl-Mann is solved by considering the three terms separately:

First: Converts true air leak rate to that of helium.
Second: Calculates the amount of helium entering the package during the bomb cycle T1. Third: Represents the amount of helium remaining in the package at test time T2.

From a practical standpoint the important parameters are bomb time T1, bomb pressure PE, package volume V, and dwell time T2. When Howl-Mann is looked at from this perspective it is clear that what it is actually measuring is the ability of an internal overpressure of He to leak out of a package. Given this reality and the lack of correlation to internal moisture build up within a package over its service life, the fact that Howl-Mann is honored more in the breach than in day-to-day practice becomes understandable.

Proc. of SPIE Vol. 8250 82500H-4

3.5 Rossiter & Neff

In order to square the circle between diffusion and flow based understandings of leaks into packages, particularly small packages of the sort used in MEMS, T. Rossiter of Oneida Research (RGA service laboratory) and G. Neff of IsoVac Engineering (krypton leak testing) made an in depth study of leak rates of very small packages using the Kr85 method as commercialized under the Radiflo trademark. This study used oxygen concentration as the key marker. Measurements of the oxygen concentration were done by internal vapor analysis (a form of RGA.) This study resulted in an empirically derived equation giving the permissible air leak rates into these packages.10 The relationship is simple:

0.5 (6)

where:

Lair = the permissible air leak rate
V = the volume of the internal cavity in cubic centimeters T = the desired system life in seconds

While there is a reasonable correlation for permissible air leak rates in larger systems, this work was done on very small MEMs packages and has not been verified for larger volume systems

4. COMPARISONS

It mustn’t be forgotten that moisture movement through capillaries into a sealed cavity is a subset of the larger volatile dynamics within and without a sealed cavity. These mass flows consist not only of air, water vapor and other species through a capillary, but also occur due to permeation, outgassing, and physisorption.11 Understanding these mass flows is further complicated by the fact that moisture is not the only species involved, nor are there any lack of potential chemical reactions within or without a sealed cavity that can complicate understanding of the total mass flow system.

To give a sense of scale to where leaks fit within these mass flow mechanisms for MEMS packages, a simple boundary condition calculation can be performed using the equation:

3 (7)

Where:

t3ml = the time for adequate moisture to flow into a package to create 3 monolayers Mml = the mass of a monolayer of water (30 ng/cm2)12
Ai = the internal surface area of the package
Q = the mass flow rate of moisture into the package

Assuming that:

There are no mass flows (Q) in or out of the package other than an arbitrary leak of mass n/unit time.
Moisture sorption on the surface is instantaneous for practical purposes.
The number of monolayers at t=0 equal 0. I.E. the package is bone dry.
The pressures within and without the package are equal.
Only the mass needed to form 3 monolayers of water is determined by this approach, not the total mass necessary to generate an internal concentration of 5,000 ppmv.

Proc. of SPIE Vol. 8250 82500H-5

Performing this calculation, which is a variant of the DerMardosian equation for moisture ingress time,13 for a relatively small package with dimensions of 0.2x0.2x0.05 cm and a leak rate of 1e-8 sccm of water vapor, results in time to form three monolayers of 15,508 years! There is obviously a serious disconnect here that supports the conclusions in our prior work that materials and processes issues are greater contributors to adverse accumulations of water within packages than leaks are, so long as the package seal has passed a reasonable level of leak check (e.g. 1-0X sccm.) 14

4.1 Comparisons with Fick, Davy, and Rossiter& Neff

Given the conceptual differences between these approaches to modeling leaks a true correlation is beyond the scope of the current work. However, norming inputs in two cases, a large package with 200 cc internal volume and a small package of 0.01 cc does give an indication of the differing results given by the various methods. In each instance a fine leak rate of 1e-8 sccm will be assumed. Both Fick and Davy are dependent on the geometry of the leak. In this comparison a leak length of 0.01 cm and a diameter of 3e-05 cm are used.

Time to 5000 ppmv Moisture

5. CONCLUSIONS

Very large packages, with volumes in the range of 10’s of cm3 and larger, can accommodate fine leaks below 1e- 07 atm-cc sec-1 (nominal) without compromising reliability via moisture failure mechanisms.
Boundary condition calculations indicate that very small packages with high surface area-to-volume ratios can withstand very large leak rates before 3 monolayers of water form, creating the conditions for corrosion.
Calculated times to unacceptable moisture levels can differ significantly between gas-flow driven and simple gaseous diffusion models.
Regardless of models used, none give exactly the same results.

Are leaks really the issue? Our analysis, using various accepted industry models as well as the actual physics involved, Fickian diffusion, points towards materials and processing issues as the major contributors of moisture within a package. A materials and process approach both helps the practitioner to understand and solve more problems than a simple leak rate perspective.

The need to solve day-to-day problems in the field force the empirical approach. That these problems continue to bedevil practitioners after decades of work, indicates a need to refine the underlying science of mass flow mechanisms within the package and its environment. Leaks continue as a hot topic because everything looks like a leak when all one has is a leak tester. The issue is further clouded by commercial agendas of leak equipment suppliers.

We see a wonderful opportunity for academia and the national laboratories to reset the science of mass flows in microelectronic packaging using a hard, analytical approach. The end result should be a ‘unified field theory’ of leaks. The questions to be answered are many, but the results are critical as packages become ever smaller in the era of MEMS.

References
[1] Lowry, R. K. and Kullberg, R. C., “Examining internal gas compositions of a variety of microcircuit package types

& ages with a focus on sources of internal moisture,” Proceedings of the SPIE, Vol. 7206, (2009) [2] Lowry,R.K.,“HermeticityandRGA,”2011CMSE,LosAngeles,CA,February2011

200 cc internal volume

0.01 cc internal volume

Fick

160 years

0.008 years

Davy

89 years

0.004 years

Rossiter & Neff

850 years

4 years

Proc. of SPIE Vol. 8250 82500H-6

[3] Rossiter, T. J., “Searching for leakers”, Minnowbrook Microelectronics Conference, October, 2008
[4] Kullberg, R. C. and Rossiter, D. J., “Measuring mass flows in hermetically sealed MEMs & MOEMs to ensure

device reliability,” Proceedings of the SPIE 6884 (2008)

[5] Davy, J. G., “Calculations for leak rates of hermetic packages,” IEEE Transactions on Parts, Hybrids, and

Packaging. Vol. 11, No.3, 177-189 (1975)

[6] Ibid
[7] Howl,D.A.andMann,C.A.,Vacuum,15,347(1965)
[8] Greenhouse,H.,Lowry,R.,andRomenesko,B.,[HermeticityofElectronicPackages],WilliamAndrew

Publications, Elsevier Publishers, 2nd edition, 108, (2012)

[9] MIL-STD 883H, Test Method 1014, 26 Feb., 2010
[10] Rossiter, T. J. and Neff, G., “Measurement of the rates of ingress of air and moisture into small hermetically sealed

devices as a function of Kr-85 leak rate and cavity size,” JEDEC, Columbus, OH, (2010)

[11]Kullberg, R. C. and Rossiter, D. J., “Measuring mass flows in hermetically sealed MEMs & MOEMs to ensure

device reliability,” Proceedings of the SPIE 6884 (2008)
[12]Brown, M. E. and Gallagher, P. E., editors, [Handbook of Thermal Analysis and Calorimetry: Recent Advances,

Techniques, & Applications], Elsevier, Amsterdam, The Netherlands, 162 (2008)

[13]DerMarderosian, A., "Permissible leak rates and moisture ingress", ARPA/NBS/NIST Workshop on Moisture Measurement Technology and Control for Microelectronics, 15 (1987)

[14] Lowry, R. K. and Kullberg, R. C., “Examining internal gas compositions of a variety of microcircuit package types & ages with a focus on sources of internal moisture,” Proceedings of the SPIE, Vol. 7206, (2009)

Proc. of SPIE Vol. 8250 82500H-7

IMAPS 2012

Getters and Design to Reliability: A Tool For Lifetime Assurance

Richard C. Kullberg & Bradley L. Phillip
Vacuum Energy Inc.
2714 W. Park Blvd.
Shaker Heights, OH 44120
1-216-991-7000 (T) rckullberg@vacuumenergyinc.com, bradlphillip@vacuumenergyinc.com

Abstract

The key to reaching multi-decade package lifetimes and device reliability is to not just take a snap shot with a RGA and declaring a part passed per MIL-STD-883. It requires a deep understanding of the sources of unwanted gases in a package, characterizing their true flow rates within, without and through the package system, and carefully choosing processes and materials, including getters, to manage the unwanted gases. This is true whether the package is hermetic or non-hermetic.

A multi-step process is discussed to include identifying the gas sources, the species present and their quantities, modeling the true quantities of gas generated over the lifetime of the package, and removing it, either through process or materials. When package service lifetimes reach decades, traditional understandings start to fall apart and careful quantitative analysis is rewarded.

Getters play a key role in attaining multi-decade lifetimes. Getter selection and sizing is discussed. Included in the discussion will be a brief synopsis of the current state of the art of gettering technology.

Key words: getters reliability hermetic packaging leaks outgassing

Introduction

To truly understand the service life impacts of contaminants like water vapor on the wide array of modern package designs and sizes requires a deep understanding of the sources of unwanted gases in a package, characterizing their true flow rates within, without and through the package system, and carefully choosing processes and materials, including getters, to manage the unwanted gases. This is true whether the package is hermetic or non-hermetic.

Indeed, the science and technology behind these understandings has reached a juncture where it has been strongly suggested that academia and the national labs clean sheet the issue as a whole. [1]

The Problem

The problem of unwanted gases within a package is conceptually simple: [2]

Figure 1: [2] A simple schematic of mass flows within a microelectronic package where:

P = pressure

C = the concentration of the species of interest

V = volume

T = Temperature

Prm = the permeation rate of a species into the internal volume of a package

Ps = the physisorption rate of the species onto the interior surfaces of the package

Og = the outgassing rate of the species from the interior surfaces of the package

∅ = the diameter of a leak path L = the length of a leak path.

Industry and the standards community have historically focused on leaks and the internal concentration of the species of interest, e.g. water vapor and hydrogen. This made sense given that packages were typically made of impermeable materials like metals. In addition, the typical internal pressure of 1 atm and volumes on the order of a few cm3 masked the impact of mass flow mechanisms such as outgassing and physisorption.

Figure 2: Time to reach 1 atm from a vacuum for various leak rates and device volumes. [3]

At low levels the rates of leak mass flow and outgassing mass flow begin to overlap. This begins to create diagnostic difficulties at mass flow rates on the order 10-11 atm cc s-1

The response of industry and the standards bodies to this changing reality has been to focus on reducing permissible leak rates, as much because they can arguably be measured, as for any other reason. The debate over the issue of leaks and their real impact has reached the point where calls have been made to start fresh on the whole issue. [4] [5]

1.E+08 1.E+07 1.E+06 1.E+05 1.E+04 1.E+03 1.E+02 1.E+01 1.E+00 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 1.E-07 1.E-08 1.E-09 1.E-10 1.E-11 1.E-12

E-12 E-10 E-8 E-6 E-4 E-2

Vacuum-Sealed Hermetic Enclosure Pressure Degradation, Time to Leak to 1 Atm

1.E+03 1.E+02 1.E+01 1.E+00 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06

Sealed Enclosure Internal Volume, cc

As the interior volume of packages continue to shrink, the impact of outgassing in particular becomes an ever more significant contributor of mass flow into the total system. Consider matters from a simple mass flow perspective. Taking guidance from MIL-STD-883H, Test Method 1014, a leak rate of 5x10-8 atm cc s-1 is the maximum acceptable for volumes of 0.01 cm3 or less. Indeed it is the lowest leak rate mentioned for any volume device in the test method. Compare it with other leak rates for the time to reach 1 atm from a perfect vacuum:

This becomes painfully clear when considering the different estimated package life times that can be calculated when using different leak models ranging from pure Fickian diffusion to the latest empirical work by Rossiter and Neff:

Time to 5000 ppmv Moisture [6]

200 cc internal volume

0.01 cc internal volume

Fick

160 years

0.008 years

Davy

89 years

0.004 years

Rossiter & Neff

850 years

4 years

Quantifying the Problem

Quantitative and scientifically applicable data on the mass flows within a package is crucial to understanding the scope of the problem to be solved. That is if there is a problem to be solved! Historical observations based on a study of archival parts have shown that if a part is hermetic and properly processed it tends to have excellent service life. Good

Time, years

material and process engineering continues to be critical for success. [7]

The two main mass flows that impact package life are leaks and outgassing. The primary methods of quantifying this mass flows and their constituents are the various forms of leak testing and residual gas analysis, or as it is known in MIL-STD- 883, internal gas analysis. In recent years the sensitivity of these methods has increased significantly through the use of time of flight spectrometers and Kr85 leak testing. [8][9]

Sensitive determination by mass spectrometer of the species present is critical to determine if a leak or outgassing is the source of the species of concern.[10][11] Mass spectrometry is also used to measure outgassing rates.[12] Outgassing is a significant source of gas in small volume systems.

In order to estimate the total mass flow into a package of a species of interest, leaks and outgassing are of the most interest in terms of magnitude. When calculating the total mass flow of a leak there are a number of conflicting approaches. Given the actual conditions observed in most packages, the authors and their colleagues lean towards the Fickian understanding of what is occurring.

Outgassing is a much simpler mechanism to model. Quantitative models have been developed that are in daily use in industry. [13] [14]

To calculate the total quantity of gas outgassed over a time period (t) measured in hours, it is customary to assume a time dependence of the outgassing rate q of the type:

t1−v
(2b) =q01−v forv≠1

(2c) =qlnt forv=1 0

Evaluating the resultant equations over the range t = 1 to t = t gives us the following:

(3a)

(3b)

t=0

With a quantitative estimate in hand of the mass flows of interest over the service life of the package a getter solution for removing that mass can be developed based on the amount and type of getter needed.

Getters

Getters have a long history of solving contamination issues within hermetic systems. Research on inorganic or organic getter materials that are able to sorb small quantities of reactive gases in vacuum devices began late in the 19th century. The first use of the term “getter” was by Edison’s assistant Malignani in 1882. Malignani developed the technique of coating components of incandescent lamps with red phosphorous. Red phosphorous reacts with, or getters, water vapor, thereby breaking the water-tungsten cycle that limits lamp lifetime. This process is still used today in the lamp industry, over a century later.

During the early 20th century successful electron tubes were first developed, but tube lifetimes proved to be impractical. Tube lifetime was limited by degradation of the internal vacuum due to outgassing. New forms of getters were developed as a successful solution to this problem. Getters based on alloys or compounds of barium were developed to supply the necessary sorption capabilities. These getters are referred to as evaporable getters because they are heated to deposit a barium as thin film on the inner surface of vacuum tubes. Such films maximize the available gettering capacity.

Early forms of evaporable barium getters included pure barium encapsulated in small iron or nickel tubes, barium-thorium alloys (Telefunken) and barium-strontium carbonate mixtures (RCA). All of these approaches had stability problems. These problems were solved by the development of the

(1) q=q t−v 0

where the time factor (v) is normally estimated to be equivalent to 1 for gases, such as carbon monoxide (CO) or nitrogen (N2), which are desorbed from the surface of a material, and is estimated to be equivalent to 0.5 for gases, such as hydrogen (H2), which desorb by diffusion from the bulk of a material.

The quantity of gas released can be obtained by integrating equation 1 over the

desired time period, t:

(2a) q= ºqt−vdt=q 00

t−vdt

t1−v t1−v −1 0 t=1 0

q 1−v t=t =q 1−v t=t

qlnt =qlnt−0 att=1,ln 0 t=1 0

BaAl4 alloy by Paolo della Porta of SAES Getters S.p.A. in the early 1950s. This alloy is stable in atmosphere and made the high volume use of getters much more practical. BaAl4 getter technology extended vacuum tube life to thousands of hours. [15]

Metal getter technology expanded beyond Ba to include Ti, Zr and their alloys. Metal getters work well for removing gases like O2, N2, CO, CO2, and H2O from a hermetic device with an internal pressure in the vacuum regime. An example gettering reaction is:

Ti+O2 →TiO2

It must be noted that H2 does not follow this mechanism. When H2 physisorbs on a chemically active site on a metal getter surface it is split into monoatomic hydrogen which then dissolves into the bulk metal, going into solution. The amount of hydrogen that can go into solution in a metal getter is inversely proportional to temperature and follows a Sievert’s Law relationship.

The hey day of classical metal getters was the barium ring era from the 1950s through the 1990s. Hundreds of millions of barium getters were produced every year in the US, Asia, and Europe. With the passing of vacuum tube technology both in electronics and in displays (CRTs) this era has passed, albeit there are still small specialty applications in industries spanning the gamut from aerospace to high end music playback systems.

Figure 3: Typical barium getter of the type developed by Dr. Paulo della Porta of SAES Getters S.p.A. [16]

The second wave of metal gettering technology is based on Zr and Ti alloys. These alloys are typically called non-evaporable or NEG getters. The classic application for NEGs and their derivatives such as sintered porous structures is to

maintain a vacuum used for thermal isolation in applications like thermos bottles or IR detectors.

Figure 4: Sintered porous getter structure of a Zr alloy. [17]

Beginning in the 1990s growing issues with contaminating species like hydrogen in non-vacuum packages became more common. A classic example is hydrogen induced degradation of GaAs devices. [18] To solve this problem a new generation of getters was developed.

From a gettering perspective a package filled with a gas as opposed to a vacuum presents significant difficulties. A broad spectrum metal getter that can remove everything from O2 to H2O is passivated in short order by the immense (from a metal getter perspective) amount of gas present. In order to provide a useable getter solution two factors came into play. First the focus shifted from taking all potentially harmful species out of a system, as is the norm in a vacuum system, to taking out the critical species, typically H2 and H2O. The second factor was to develop active gettering materials that can remove one or two species in an atmosphere of gas without being destroyed by the other gases present.

There are four mechanisms by which H2 can be gettered from a package. These mechanisms include the formation of metal hydrides, reducing metal oxides, hydrogen re- combination (forming water while in the presence of oxygen) and hydrogenation.

The metal getter community took a thin film deposition path. The original gas of interest was H2. It is possible to deposit a thin film metal structure that will effectively sorb H2 without being passivated by the other chemically active gases present. [19] Another pathway was to take gettering approaches used in other industries, for example PdO as used cryogenics to sorb H2, and incorporate them in structures suitable for use in microelectronic packages. [20]

Both of these approaches present difficulties to the packaging engineer. A metal thin film needs an adequate footprint within a package for deposition. This deposition typically occurs at an outside facility, complicating the workflow. The use of PdO addresses the footprint and workflow issues, but the gettering reaction creates its own issues:

PdO+H2 →Pd+H2O

Introducing additional water into a package that the user is already working to keep dry is disconcerting to say the least. The supplier of these formulations does address the issue with the addition of a desiccating agent, but it is still an ongoing concern.

At the current time the most advanced solutions for removing H2 side step the whole issue of metal films or metal oxides entirely by leveraging hydrogen gettering compounds developed by Sandia National Labs. These compounds work by hydrogenation and can incorporate the end user’s desiccating agent of choice.

Originally pioneered by Sandia National Laboratories, these materials are based on the selective hydrogenation of unsaturated carbon-carbon triple bonds and double bonds to their saturated carbon-carbon single bond analogs. These materials scavenge hydrogen in an irreversible manner in contrast to the aforementioned metal hydrides. Additionally, they do not generate water as a part of the reaction chemistry, in contrast to the metal oxide systems such as PdO. The organic hydrogen getter materials offer package designers another option in dealing with the removal of hydrogen from hermetically sealed devices. [21] [22]

Figure 5: Typical Hydrogenation Getter Reaction

Figure 6: Typical Vacuum Energy polymer hydrogen getters. [23]

Choosing a Getter

Based on this short discussion of the types of getters commonly in use, simple guidelines for choosing a gettering technology to investigate can be given.

Is your package under vacuum or is it gas filled? If it is under vacuum a traditional metal getter system is the most likely candidate. These systems are well understood and the suppliers can assist you in engineering a solution. Caution is urged however, in the MEMS case given the typically very high surface area to volume ratios in these packages as well as limited footprint availability for getter integration. These factors make careful evaluation of the outgassing load critical, as well as absolute hermeticity with no actual leaks.

If a package is gas or air filled matters become more complex. Careful needs analysis is required to identify the actual problem to be solved. Some of the questions to be answered during such analysis are the actual species of concern (H2, H2O or something else entirely?), the desired service life, which leak model and test method are most applicable, and more. As volumes decrease and surface area to volume ratios increase mechanisms

like outgassing or the formation of addition water within the package due to hydrogen reducing any metal oxides present becomes more critical to understand.

Removing moisture vapor only from a package is the simplest scenario, solvable by introducing a desiccating agent. Typical agents used in microelectronics include molecular sieves or anhydrous CaO. Usage should be per the manufacturer’s directions for best results. Do note that both approaches present issues either of activation for the molecular sieves or preventing hydration of the CaO until the package is sealed. The addition of desiccating agents does present a useful tool for the packaging engineer.

Desiccating agents work by two different mechanisms. These are physisorption and chemisorption. Historically the bulk of water sorbing materials used have been of the physisorption variety, e.g. mole sieves and zeolites. These materials work by presenting very large surface areas to a system for water to sorb on. Molecular sieves and zeolites are often used in conjunction with metal oxide hydrogen getters, where they are used as much to getter the water generated by the hydrogen guttering reaction, as to getter any other water present in the package. To say that it seems somewhat incongruous to create water in moisture sensitive packages is a bit of an understatement.

As packages shrink and have ever higher surface area to volume ratios, the MIL-STD-883 5000 ppmv maximum permissible water vapor concentration specified to keep the number of water monolayers at 3 or less, is ever less applicable to the real world. There are programs now specifying maxim water concentrations as low as 1000 ppmv to prevent corrosion and stiction. When operating at these very low maximum water concentrations, vapor pressure issues become important as well.

In such circumstances a better technical choice (albeit not necessarily the better choice from a process flow perspective) is a chemisorbant like anhydrous CaO. Anhydrous CaO irreversibly (in a practical sense) getters water from a system. CaO provides much lower water vapor pressures that range from 10-11 torr at 0 C to 10-9 torr at room temperature to 10-5 torr at 100 C. Consequently, under normal room temperature operating conditions, the water vapor pressure is 6 orders of magnitude lower than that of a typical zeolite.

H2 only removal can be equally simple in principle given the available thin film and polymer gettering materials. Both options will sorb H2 without activation or the generation of H2O. In the case of the polymer materials a desiccating agent, complete with the associated issues previously discussed, can be incorporated into the final part.

Designing a hydrogen getter is complicated by the source of hydrogen within a package. Hydrogen outgasses from metal components, plating, and metallization of the various internal surfaces within a package. Uniformity of this outgassing rate can not be assumed even within the same lot of materials. It has been observed to vary by up to two orders of magnitude in packages assembled from the same lots of source materials and parts.

A design rule has been developed to accommodate this extreme variability. Historical data shows that H2 concentrations typically fall within a range of a few hundred PPMV to 27,000 ppmv. An outlier population of approximately 7-8% of the total exceed 27,000 ppmv. Experience shows that even this population rarely exceeds 50,000 ppmv. Consequently hydrogen getters can be engineered to the 50,000 ppmv case plus the desired factor of safety. [24]

Factors of Safety

It is very common for hermetically sealed systems to be used in mission critical components. Applications for gettered systems literally range from the bottom of the ocean to the moons of the outer planets. [25] [26] Getters play a key role in mission assurance for these types of applications.

When considering a getter solution not only must the cost of the getter material and its integration into the package be considered, but also the price of failure should adverse species be allowed to increase in concentration to the point where failure mechanisms are induced. Factors of safety in getter capacity of 2-3 are very common and it is not unheard of to reach factors of safety as high as 10 in applications like space flight.

Conclusions

Getters are not a black art nor do they consist of magic pixie dust. Rather they are highly engineered materials that have solved real world problems for over a century at the leading edge of technical developments from the light bulb to the latest in MEMS based sensor systems.

In Memoriam

The authors would like to dedicate this paper to a key figure in the history of getters. Dr.. Paolo della Porta, the founder of SAES Getters S.p.A., passed away this year. His technical rigor and entrepreneurial drive continue to inspire.

References

[1] R.C. Kullberg, A. Jonath, & R.K. Lowry, “The Unsettled World of Leak Rate Physics: 1 Atm Large - Volume Considerations Do Not Apply to MEMS Packages, A Practitioner's Perspective,” Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and Nanodevices XI, Proc. of SPIE Vol. 8250, 82500H-1

. [2] R.C. Kullberg and D.J. Rossiter, “Measuring mass flows in hermetically sealed MEMs & MOEMs to ensure device reliability” Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS VII, Proc. of SPIE Vol. 6884, 68840L-1, 2008

[3] R.C. Kullberg & R.K. Lowry, “Hermetic Package Leak Testing Re-Visited,” IMAPS International Conference and Exhibition on Device Packaging, Presentation TP24, Scottsdale, Arizona, March 2008

[4] R.C. Kullberg, A. Jonath, & R.K. Lowry, “The Unsettled World of Leak Rate Physics: 1 Atm Large - Volume Considerations Do Not Apply to MEMS Packages, A Practitioner's Perspective,” Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and Nanodevices XI, Proc. of SPIE Vol. 8250, 82500H-1, 2012

[5] Richard C. Kullberg & Robert K. Lowry, “Hermetic Package Leak Testing Re-Visited,” IMAPS International Conference and Exhibition on Device Packaging, Presentation TP24, Scottsdale, Arizona, March 2008

[6] R.C. Kullberg, A. Jonath, & R.K. Lowry, “The Unsettled World of Leak Rate Physics: 1 Atm Large - Volume Considerations Do Not Apply to MEMS Packages, A Practitioner's Perspective,” Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and Nanodevices XI, Proc. of SPIE Vol. 8250, 82500H-1, 2012

[7] R.K. Lowry and R.C. Kullberg, “Examining internal gas compositions of a variety of microcircuit package types & ages with a focus on sources of internal moisture”, Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and

Nanodevices VIII , Proc. of SPIE Vol. 7206, 720606- 1, 2009

[8] R.C. Kullberg and D.J. Rossiter, “Measuring mass flows in hermetically sealed MEMs & MOEMs to ensure device reliability” Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS VII, Proc. of SPIE Vol. 6884, 68840L-1, 2008

[9] Rossiter, T. J. and Neff, G., “Measurement of the rates of ingress of air and moisture into small hermetically sealed devices as a function of Kr-85 leak rate and cavity size,” JEDEC, Columbus, OH, 2010

[10] R.C. Kullberg and D.J. Rossiter, “Measuring mass flows in hermetically sealed MEMs & MOEMs to ensure device reliability” Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS VII, Proc. of SPIE Vol. 6884, 68840L-1, 2008

[11] [7] R.K. Lowry and R.C. Kullberg, “Examining internal gas compositions of a variety of microcircuit package types & ages with a focus on sources of internal moisture”, Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS and Nanodevices VIII , Proc. of SPIE Vol. 7206, 720606- 1, 2009

[12] C. Carretti, “General Introduction to Outgassing Data and Measurement Methods”, White paper presented by the Gas-Surface Laboratory, SAES Getters S.p.A., October, 1995

[13] A. Corazza and R.C. Kullberg, “Vacuum Maintenance In Hermetically Sealed MEMs Packages,” Micromachined Devices and Components IV, Proceedings of SPIE Vol. 3514 p. 82-89, 1998

[14] [12] C. Carretti, “General Introduction to Outgassing Data and Measurement Methods”, White paper presented by the Gas-Surface Laboratory, SAES Getters S.p.A., October, 1995

[15] R. Ramesham, “Getters for Reliable Hermetic Packages”, Jet Propulsion Laboratory Publication D- 17920, 1999

[16] http://www.saesgetters.com

[17] A. Corazza and R.C. Kullberg, “Vacuum Maintenance In Hermetically Sealed MEMs Packages,” Micromachined Devices and Components IV, Proceedings of SPIE Vol. 3514 p. 82-89, 1998

[18] “Hydrogen Effects on GaAs Microwave Semiconductors”, Prepared for the Jet Propulsion Laboratory, Report Number: SMC97-0701, July 1997

[19] R. Kullberg, M. Moraja, C. Carretti, “Advances in Hydrogen Gettering Technology”, OnBoard Technology, February, 2005, p. 42-45

[20] “Hydrogen Effects on GaAs Microwave Semiconductors”, Prepared for the Jet Propulsion Laboratory, Report Number: SMC97-0701, July 1997, P. 28

[21] R.C. Kullberg, B. L. Phillip, & T. J. Shepodd, “Advanced Getter Materials for GaAs, RF/MW, MEMs and Other Microelectronic Packages”, WA11, IMAPS Advanced Technology Workshop on RF and Microwave Packaging, San Diego, California, September 22-24, 2009

[22] P. Koning, “Sandia’s getter technology comes full circle”, http://www.sandia.gov/LabNews/091106.html

[23] http://www.vacuumenergyinc.com

[24] data set privately communicated to the author

[25] R.C. Kullberg, “An Overview of Gettering Technology for Sensitive Systems”, SEAFOM, Houston, TX, 2010

[26] R.C. Kullberg, “Extreme Reliability Getter Pumps and the Huygens GCMS”, IMAPS Advanced Technology Workshop on Reliability of Advanced Electronic Packages and Devices in Extreme Cold Environments, Pasadena, California, 2005

Sandia

Sandia’s getter technology comes full circle

igniting anything — it moderates the reaction by never letting the temperature get to the ignition point.”

About five years ago, Vacuum Energy expanded the getter application space into flares. Flares are sealed to enable a chemical reaction when they are exposed to oxygen. That sealed compartment, like a waterproof flashlight, can become explosive if too much hydrogen builds up inside.

Tim says he receives several calls each year about new uses for hydrogen getters. “The common theme is often water. I’ve been contacted repeatedly by subma- rine and buoy operators, even water-themed Las Vegas shows,” he explains. “We are always looking for new applications.”

In many of those calls, Tim says, people tell him they have a hydrogen problem but they have no idea how much hydrogen they are dealing with. Sandia’s getters can be used to determine the extent of hydrogen build-up, as you can count the number of hydrogen atoms absorbed by a getter over a set period of time.

Last year Sandia’s getter passed rigorous DOE testing for use in plutonium transport. Tim says that without getter commercialization, Sandia never would have been able to develop the technology for this applica- tion. Vacuum Energy is now supplying getters to DOE’s Savannah River Site.

“The biggest hurdle was that the plutonium ship- ping container contains a mixture of carbon dioxide

By Patti Koning

Hydrogen is also dangerous, especially when it builds up in sealed compartments, because it is chemically reactive and has a large explosive range with air and high thermal conductivity.

“Have you ever opened your TV remote to change the batteries and noticed that the old batteries are bulging slightly? That’s hydrogen pressure,” explains Tim. “A typical AA battery is semihermitically sealed, so when it fails, it fails suddenly.” That released hydro- gen can be ignited with static electricity, such as static generated from sliding across a synthetic bed com- forter or carpet.

The getters contain unsaturated organic compounds that react directly with the hydrogen and remove it from the atmosphere, a process called “gettering.” If oxygen is present, the getter enables the hydrogen to react with the oxygen to form water in a controlled cat- alytic reaction. Tim describes getters as being like ice cream, with many different flavors designed to work in high or low temperature, radioactive, or noxious gas environments. While the flashlight getters are pellets, getters are also used in powder, spray coating, and sheet form.

Easy to get rid of

“It’s easy to get rid of hydrogen,” says Tim. “The challenge is to move an explosive atmosphere below an explosive concentration in a controlled fashion. That’s what our technology does extremely effectively without

Hydrogen getters are not sexy, but they are reliable. Sandia has been in the getter business for nearly 20 years through a flourishing partnership with Vacuum Energy Inc., which has the exclusive license to the technology.

“This is really the textbook case of how technology transfer should work,” says Tim Shepodd, manager of Materials Chemistry Dept. 8223. “We developed our hydrogen getters for applications in nuclear weapons. Then they were commercialized, which kept the pro- gram going at Sandia, and they have come back to us for plutonium transport.”

About 15 years ago, Brad Phillip of Vacuum Energy approached Sandia for help removing gases from vac- uum-insulated refrigerator panels to extend the prod- uct life, improve efficiency, and decrease size. Seeing the potential for hydrogen removal applications, Vac- uum Energy licensed Sandia’s getter technology.

If you’ve used a waterproof flashlight, there’s a good chance you’ve encountered one of Sandia’s getters. Each year, Vacuum Energy sells millions of Sandia’s getters to flashlight manufacturers. The getters, which are about the size of a pencil eraser, scavenge the hydrogen that accumulates in the sealed compartment to prevent an explosion.

Hydrogen is ubiquitous because moisture is always present, despite efforts to remove it. Water reacts with metals, making metal oxides and hydrogen gas.

and air, and getters had never been used in that envi- ronment before,” says April Nissen (8223), who over- saw the testing for Sandia. “We had to ensure that the getters would perform perfectly in those conditions. Failure is not an option.”

An important component of the testing was how the getters performed as they aged in the radioactive environment. Recently, April ran tests on a getter that had been in service for two years and found that they performed better than the laboratory-aged samples, which were subjected to worst-case conditions. “The getters sailed through a very rigorous set of tests,” says April. “We knew the technology would perform, but you never expect this sort of thing to come off without a hitch.”

Another potential getter application is in radar sources, which can degrade and malfunction when exposed to hydrogen. “A lot of people have an issue with hydrogen-sensitive devices that are placed with or in microchips and sealed,” says Tim. “Existing technol- ogy used for radar sources deactivates very quickly when exposed to room air, making this technology very difficult to work with.”

Tim presented Sandia’s getter solution at the

Advanced Technology Workshop on RF and Micro- wave Packaging, sponsored by the International Microelectronics and Packaging Society last month in San Diego, Calif. An advantage of Sandia’s tech- nology is that it can be worked on in air for several hours before it needs to be sealed. “Our technology also has orders of magnitude less equilibrium water vapor over the surface of the material, minimizing the chance of corrosion, and it passes low ionic content test,” says Tim. “Our presentation was very well received. We are definitely expanding this line of business.

Sandia’s getter work has taken Tim and other researchers in some unusual directions. About a year and a half ago, a salmon supplier asked Sandia for help in maintaining a specific environment in shipping containers that were cooled using a technology that leaked a little hydrogen.

“If the supplier could keep the fish at a certain pres- sure and temperature consistently, they could change their shipping method and save a lot of money,” he says. “So for a few months we had refrigerators out in the lab with trays full of salmon and getters as we worked on the environment.”

Another aspect of the project was quantifying the odor emitted by the salmon to determine what factors might be changed to keep the fish fresher. “That’s the beauty of working at Sandia,” says Tim. “You never know what kind of project you’ll be working on next — salmon or radar sources or something else totally unexpected.”