Issue



Getters for Microlectronic Packages


12/01/2004







SOLVING HYDROGEN DEGRADATION PROBLEMS

BY RICHARD KULLBERG, HEATHER FLORENCE, MARCO MORAJA, AND RON PETERSEN

Hydrogen contamination of microelectronics packages has been of concern for many years, particularly to the GaAs industry.1 Hydrogen contamination issues impact many areas of the microelectronics industry.2 Getters have been proposed to alleviate these problems.3

Quantifying The Amount Of Hydrogen Present

To determine the amount of getter necessary to remove the hydrogen from a microelectronic package, it is necessary to quantify the amount of hydrogen originally present at the time of sealing the package and the amount expected from outgassing within the package over its lifetime. A normal RGA test can determine the concentration of hydrogen in the package. To determine the amount of outgassing, two techniques typically are considered. The first is the known conductance method, which is carried out in dynamic conditions, i.e. pumping away the gas flow coming from the sample. The second is the pressure rate of rise method, which is carried out in a static, non-pumped system.

The choice between these two methods depends on the outgassing characteristics of the type of material being studied. The known conductance method is used to measure the outgassing rate from metals, glasses, ceramics, and other materials that release a relatively low amount of gas. The pressure rate of rise is useful in making measurements on plastics, foams, and other materials that show a large amount of gas evolved (typically water vapor). Both methods can be used to make measurements on single materials or complete devices. To measure the outgassing rate of gases in microelectronics packages, the known conductance method typically is used.4

Hydrogen Outgassing and Getter Dimensioning

Once the outgassing rates for hydrogen in a system have been determined these rates can be inserted into a model to estimate with a fair degree of accuracy the total amount of hydrogen expected to accumulate over the lifetime of the package.5 To calculate the total quantity of gas outgassed over a time period (t) measured in seconds, assume a time dependence of the outgassing rate (q) of the type:

q = qot-v,

where the time factor (v) has an estimated equivalent of 1 for gases such as carbon monoxide (CO) or nitrogen that are desorbed from the surface of a material. The time factor estimated equivalent is 0.5 for gases such as hydrogen, which desorb by diffusion from the bulk of a material.

By integrating the measured outgassing rates from typical packages, it can be seen that the service life will be quite short if the outgassed species are not trapped.

The total gas load (Q) = qo (t1-v-1)/1-v.

A combination of this total outgassed hydrogen over the system service life, plus the amount of hydrogen originally present, equates to the required getter capacity. Good engineering practice dictates extra getter capacity, which is driven by the cost the market will accept.

Example Calculation

For the purposes of this example, let's assume a package with a volume of 1 cm3. The hydrogen concentration, based on piercing a sealed example and measuring with a RGA, is 5,000 ppm or 0.5%. Furthermore, the pressure in the package is 760 torr (1 atm) at room temperature (STP conditions.) In gettering technology, it is convenient to work in units derived from the Ideal Gas Law:

PV = nRT.

The mass of gases are measured in pressure and volume units:

n = PV/RT.

Working in mass units derived from the Ideal Gas Law, the PV units of choice are cm3-torr. This means that the mass of hydrogen present in the package as measured by RGA is:

(0.005×1)cm3 × 760 torr = 3.8 cm3-torr.

The outgassing rate of hydrogen in a package typically is considered proprietary information. It is acknowledged, however, that the variability of outgassing rates of hydrogen in a package is dependent upon the material of construction, types of coatings, heat treatments, means of hermetic sealing, and storage conditions.6 For the sake of discussion, a typical value can be taken to be on the order of 1.6×10-8cm3-torr/sec-cm2. The 1-cm3 cube used in this example has an interior surface area of 6 cm2. A typical service life for a package is 10 years. This information can be plugged into the total gas load equation to calculate the total amount of additional hydrogen to be gettered in the package. In this example, the additional amount of hydrogen outgassed over 10 years is 0.2-cm3-torr, giving a total hydrogen quantity to be pumped of 4 cm3 torr.

One company's* new getter can vary from 200 to 700 cm3-torr/cm2 in hydrogen capacity depending upon the application it is optimized for. For this example, we will take the lesser capacity because it is typical of getters optimized for microelectronic packages. Assuming that the getter coats 1 cm2 or one interior face of the cube, for example, the lid (typically the sealing area of the lid is not coated, which insignificantly reduces the gettering surface area), and given that the need is to pump 4 cm3-torr of hydrogen, the safety margin offered by the new getter is 50× (200 cm3-torr/4 cm3-torr = 50). This is a best-case scenario. However, even considering other potential outgassing sources of hydrogen, it is estimated that the getter offers safety margins of 30× or more.

A Getter Solution For Hydrogen

The new type of getter was designed to remove hydrogen contaminants from hermetically sealed microelectronics packages. It selectively pumps hydrogen, and can be used in package ambients where earlier-generation, non-evaporable getters (NEGs) cannot work. A key feature is that the hydrogen getter requires no activation.


Figure 1. Hydrogen getter deposited on Si substrate.
Click here to enlarge image

The getter is formed via a PVD process to create a two-layer thin film structure of titanium and palladium on common substrate materials, including Kovar, silicon, and other metallic or ceramic substrates up to 9 × 12 in. (Figures 1 and 2). The PVD process allows the getter films to be patterned.

Gettering Mechanisms


Figure 2. Getter deposited on Kovar lids.
Click here to enlarge image

The theory of bulk gettering of diatomic gasses is highly evolved.7 In a traditional getter system, the getter is activated by a thermal treatment to depassivate the surface of the getter material — rendering it chemically active. Once the surface is chemically active, two gettering mechanisms come into play, depending on the gas to be pumped.

In the instance of gases like oxygen or nitrogen, the gas molecules are first physisorbed onto the getter surface. If the physisorbtion occurs on a chemically active site, further reaction occurs to permanently capture the gas. Hydrogen is pumped by a different mechanism. On a chemically active getter surface, a hydrogen molecule that is physisorbed cracks into monatomic hydrogen. As protons, the hydrogen easily diffuses into the lattice or bulk of the getter material.8

Palladium is transparent to the passage of hydrogen gas.9 A palladium film is deposited on top of the titanium layer to prevent its exposure to atmospheric gases that will repassivate it by the formation of oxides, carbides, and nitrides. Once the hydrogen has passed through the palladium film, it is pumped in normal fashion by the titanium.

H2 Getter Performance and Dynamic Sorption Test Results

Representative getter films were tested for performance in a dynamic sorption test at 0.1 torr of H2. The test was performed per ASTM F 798-97, Standard Practice for Determining Gettering Rate, Sorption Capacity, and Gas Content of Non-evaporable Getters in the Molecular Flow Region. The getter shows consistent pumping speed at 1-2 cm3/s.cm2 until the film has reached its capacity for hydrogen. The capacity of the film is 700 cm3torr/cm2 (Figure 3).


Figure 3. SEM of a cross section through the getter film.
Click here to enlarge image

null

Film Adhesion and Particles

A qualitative test of film adhesion was performed per MIL-STD-977, Method 4500, Metallization Adherence, on deposited films. In addition, a qualitative test for particulation using ASTM D 3359-97 was performed on both freshly deposited films and ones that had pumped 700 cm3torr/cm2 of hydrogen. In both series of tests, there was no evidence of loss of adhesion or particulation (Figure 4).


Figure 4. 10× photomicrograph of a tape sample from adhesion and particulation testing.
Click here to enlarge image

null

Conclusion

A hydrogen-only getter, requiring no thermal activation, has been industrialized. The film structure passes qualitative tests for adhesion and particulation. The getter has a capacity of 700 cm3torr/cm2, offering at least a 30× safety margin for most devices. It also can be deposited on standard substrate materials, including silicon and Kovar.

References

  1. "Hydrogen Effects On GaAs Microwave Semiconductors," Shason Microwave Corp. for JPL, report number SMC97, 0701, July 1997.
  2. vPrivate communications, Minnowbrook Microelectronics Conference, 2002 and 2003.
  3. A. Immorlica Jr., S.B. Adams, and A.R. Reisinger, "Practical Approaches to Remediation of Hydrogen Poisoning in GaAs Devices," GaAs MANTECH, 1999.
  4. Stefano Tominetti and Anna Della Porta, "Moisture and impurities detection and removal in packaged MEMS," Proc. SPIE Vol. 4558, pp. 215-225, Reliability, Testing, and Characterization of MEMS/MOEMS; Oct. 2001.
  5. C.Carretti, P.Manini, A.Renzo, G.Andreoletti, G.Carcano and G.Lambrughi, Vuoto, Vol. XXIII N.2 (1994) p. 73.
  6. "Hydrogen Effects On GaAs Microwave Semiconductors," Shason Microwave Corp. for JPL, report number SMC97, 0701, July 1997.
  7. R. J. Knize and J. L. Cecchi, "Theory of Bulk Gettering," J. Appl. Phys. 54 (6), June 1983, pp. 3183-3189.
  8. R.C. Kullberg, "Processes And Materials For Creating And Maintaining Reliable Vacuum And Other Controlled Atmospheres In Hermetically Sealed MEMS Packages," Proc. SPIE Vol. 3880.
  9. John F. O'Hanlon, A User's Guide To Vacuum Technology, New York: John Wiley & Sons, 1989, p. 281.

*SAES Getters.

RICHARD KULLBERG, application development manager, HEATHER FLORENCE, MARCO MORAJA, application engineer, and RON PETERSEN, business development manager, may be contacted at SAES Getters/USA Inc., 1122 E. Cheyenne Mountain Blvd., Colorado Springs, CO 80906; (719) 527-4123; e-mail: rck@saes-group.com; haf@saes-group.com; marco_moraja@saes-group.com; rop@saes-group.com.