Challenges and solutions for germane (GeH4) packaging
06/01/2003
Overview
Germane gas is increasingly called for as a deposition reactant in wafer processing applications needing silicon germanium layers. Because this gas is a flammable, toxic, colorless gas reactive with oxidizers and halogens, its use in wafer fabs demands packaging that can ensure safe and environmentally friendly handling. Specific tests have analyzed these issues, particularly any potential for deflagration, when germane is packaged in a subatmospheric delivery system.
Germane is plasma deposited with silane or disilane to form amorphous silicon germanium (SiGe) alloys (10–50% germanium). SiGe layers in a multiple junction device efficiently capture red and infrared photons and significantly enhance overall solar energy conversion. The demand for efficient, environmentally friendly solar energy sources has increased at a steady pace over the last several years, which translates into a greater demand for germane.
There is also a growing demand for germane gas for the production of SiGe semiconductor devices as this technology is poised to enter mainstream production. In one family of devices, germanium is alloyed with silicon to produce a strained heterostructure when silicon is deposited on the SiGe layer. This strained silicon technology provides significant benefits for the production of heterojunction bipolar transistor (HBT) devices, including the ability to engineer band gaps to specific device needs for high frequency optical networking, wireless, and other communications applications.
IBM pioneered the use of SiGe to produce a new generation of 350Ghz bipolar transistors and has led the commercial development of this technology [1, 2]. Intel is also adopting SiGe technology, and the company will use SiGe buffer layers to implement strained silicon technology in its new 90nm CMOS process on 300mm wafers [3].
Development work to extend SiGe applications continues at a rapid pace. For instance, the use of monomethylsilane for adding carbon to SiGe HBT device layers is being developed to reduce boron diffusion after implantation. As these and other new applications become commercially important, demand for germane and safe germane packaging technologies can be expected to experience significant growth.
Environment, health, and safety
While an essential chemical in semiconductor manufacturing, germane gas has hazards that must be carefully considered to ensure its safe handling and use. Germane is a colorless gas with a pungent odor that is not detectable by some people. Its immediate health hazards include inhalation of the poison gas and potential for thermal burns.
While not pyrophoric [4], germane is a flammable gas that may form mixtures with air that are flammable or explosive. The lower flammability of germane in air has not been established. Similar to acetylene, germane's upper flammability limit is 100%; this material can propagate a decomposition flame without the presence of air through a deflagration process. Deflagration is defined as the chemical reaction of a substance in which the reaction front advances into the unreacted substance at less than sonic velocity.
The threshold limit value (TLV) and time-weighted average (TWA) for germane is 0.2ppm, as defined by the American Conference of Governmental Industrial Hygienists. Appropriate hydride gas monitoring systems should be considered for any installation using this material.
It is well known that pure germane gas, in the presence of an ignition source, can decompose to germanium and hydrogen [5] via an exothermic reaction. The energy released during this decomposition in a closed system results in a significant temperature increase along with a considerable pressure rise. This pressure surge is due not only to heating, but also to the formation of two molecules of free hydrogen for each molecule of germane decomposed.
On one occasion, a cylinder of pure germane overfilled by current standards and assumed to be contaminated with digermane, ruptured unexpectedly causing injuries. The root cause of the accident was deemed to be decomposition and deflagration of the cylinder contents, generating pressures exceeding the burst pressure of the cylinder. To minimize this risk, Voltaix fills cylinders of 100% germane using this criterion: If all the germane in a container were to decompose adiabatically, the final pressure would not exceed the working pressure of the cylinder. While deflagration has never occurred in a Voltaix product cylinder, this prudent approach is used to assure safe product packaging.
Work done in Japan and verified by Voltaix demonstrated that germane deflagration can be inhibited by dilution or reduction of the pressure inside a cylinder [6]. Based on these findings, another approach to safe handling of germane is to provide the product as a mixture in N2, H2, Ar, or He. Germane is widely used for both semiconductor and photovoltaic applications in safe mixture concentrations ranging from parts per million to 20%.
In summary, the nature of germane dictates either the use of dilute mixtures or reduced fill cylinder packaging of 100% gas. However, there are cost, safety, and process benefits associated with providing 100% germane in full cylinders for many applications.
Subatmospheric gas delivery
Packaging hazardous gases used for the manufacture of ICs is experiencing a significant revolution. Conventional gas cylinders contain gases in a compressed, sometimes liquefied form. A new generation of gas packages uses chemical or physical means to reduce the risks associated with high pressure gas release [7]. For example, adsorption-based gas packaging technologies rely on physical adsorption of gases on microporous media at pressures <1atm. During use, the inventory of adsorbed material is drawn from the cylinder by a vacuum pump.
Adsorption-based chemical storage and delivery was pioneered in the mid-1990s [8] with the introduction of adsorbed-phase SDS cylinders. Designed specifically for ion implantation applications, these packages provided considerable safety and productivity enhancements. The use of adsorption-based gas cylinders recently expanded to chemical vapor deposition (CVD) processes [9]. Initial applications focused on the delivery of neat (100% concentration) subatmospheric gas enhanced (SAGE) phosphine for phosphorus doping of silicon-based thin films. Recently, use of SAGE has expanded to other materials, such as trimethyl silane and germane.
Adsorption-based gas sources enable the use and storage of hazardous materials in fab areas previously not contemplated due to safety considerations. Using traditional gas cylinders, best practices often recommended the use of diluted mixtures or remote installation of the hazard-containing package. SDS and SAGE gas sources reduce the number of cylinder changeouts and enable location closer to the tool. These advantages normally translate into reduced cost of piping and ventilation requirements and longer component lifetimes brought about by the low-pressure service. 100% gas sources can offer process benefits due to consistency.
GeH4 has been successfully offered for several years in the SAGE package. Consistent with the Voltaix practice, ATMI currently supplies germane in adsorbed packages using the criterion that if all the germane were to decompose, it would not exceed the working pressure of the cylinder. Testing work has demonstrated an additional safety benefit in that the SAGE package inhibits germane deflagration.
Germane deflagration studies
The safety aspects of storing germane gas in a SAGE package were investigated relative to conventional storage methods (adsorbent free). The tests were performed at Voltaix in consultation with Chilworth Technology Group, an independent company specializing in industrial safety. Deflagration of germane was induced with adsorbent loadings of 0%, 30%, and 90%/unit volume.
 Figure 1. Experimental setup and apparatus. |
null
In these tests, controlled deflagration reactions were induced in a double-contained vessel (Fig. 1). Germane and adsorbent were contained in a 25.9¥1.3cm stainless steel tube with an internal volume of 0.05 liters. This primary reactor included thermocouples to measure the temperature at the center of the vessel at three equidistant points along its length (top, middle, and bottom). A high-accuracy, high-speed transducer was connected to a data collection instrument capable of scanning at a 1 msec frequency to measure time dependent changes in pressure and temperature. The secondary container was a 4.5 liter 316-stainless steel autoclave reactor (127mm inner diameter, 305mm long) with a maximum working pressure of 5500psig.
The primary vessel was fitted with a 1000psig burst pressure rupture disk. The secondary container included a thermocouple, a pressure relief valve venting to a large volume ballast reservoir, and a 5000psi burst pressure rupture disc. A heating blanket wrapped around the body of the outer vessel could control temperature in the primary reactor.
Ultra high purity grade germane (99.999%) was used in all experiments. The adsorbent was proprietary high-density microporous media used in all SAGE products.
Deflagration of germane inside the primary reactor was induced by applying 6 VDC to a 0.25mm nichrome fuse wire for a pre-set period of time. The voltage output from the power supply was recorded to validate duration of the igniter activation and to calculate the total energy input. After each experiment, unreacted material in the headspace was analyzed.
 |
null
Testing Deflagration
The first tests involved pure germane and no adsorbent. Approximately 5g of germane were loaded into the reactor to a pressure of 300torr at 21°C. The ignition source was activated and ~0.63 sec later, the deflagration reaction began resulting in pressure and temperature increases. The pressure initially rose at a rate of 19psi/sec and then 61psi/sec, reaching a maximum pressure increase of 27.2psia from baseline. Thermocouples placed at the top and middle of the reactor recorded maximum temperatures of 131°C and 83°C, respectively. Analysis of the gas after the event confirmed that 93.8% of the germane had decomposed.
 Figure 2. Deflagration results of pure germane at 650torr. |
null
Then, ~15g of pure germane were loaded (with no adsorbent) into the reactor to a stable pressure of 650torr at 21°C. The deflagration reaction started after 0.55 sec. A maximum pressure of 70.2psig was observed ~0.6 sec after the initial deflagration process began. Thermocouples placed at the top and center of the reactor recorded maximum temperatures of 213°C and 88°C, respectively. Time-dependent temperature and pressure plots are depicted in Fig. 2. Gas analysis of the reaction product indicated that ~98.4% of the germane had decomposed.
Next, data was gathered using pure germane on a 30% adsorbent fill-in insert, with full germane loading. Approximately 15cm3 of adsorbent was loaded into the reactor, occupying 30% of the total available volume. Germane gas was loaded onto the adsorbent to reach an equilibrium pressure in the reactor of 650torr. In total, ~3.3g (0.043 moles) were introduced into the reactor. Roughly 4% of germane was estimated to be in the headspace of the insert while the remainder was adsorbed onto the adsorbent.
To represent the generally accepted worst-case temperature conditions, the reactor was heated to 65°C using the heating blanket. This temperature rise resulted in an expected desorption of some of the germane, increasing the pressure to 53.2psia and the amount of free germane in the headspace to 12.5%. With ignition, in ~0.2 sec the pressure in the system began to rise at a rate of 71psi/sec, reaching a maximum of 56psig. The maximum temperature recorded in the headspace, 91°C, was significantly higher than the temperatures measured at the center and bottom of the reactor, namely 66°C and 68°C.
These results are consistent with the deflagration of only a portion of the free germane in the headspace. Complete decomposition of germane would have resulted in pressures exceeding 1000psig, depending on the heat transfer rate between the gas and the adsorbent. Quantitative gas analysis of the headspace gas indicated that 0.00283 moles of hydrogen were formed. Based on the molar balance, 31% of the germane in the headspace and only 3.3% of the total GeH4 volume loaded in the reactor deflagrated.
Finally, full germane loading on ~45cm3 of adsorbent occupying 90% of the available volume was tested. The prepared adsorbent was loaded with 9.7g (0.127 moles) of germane to an equilibration pressure of 650torr. As with the previous test, the system was heated until the average adsorbent bed temperature reached 67°C, resulting in a pressure of 66.7psia. The volume of free germane in the headspace in these conditions was ~5%.
 Figure 3. Deflagration results of pure germane on a 90% adsorbent fill-in insert. |
null
The reactor temperature and pressure varied only slightly from the initial conditions after the ignition source was activated (Fig. 3). Assuming that all the germane loaded onto the adsorbent had deflagrated, the pressure in the insert would have exceeded 1000psig and the rupture disc would have burst. Analyses of the headspace gas showed that 0.000288 moles of hydrogen was generated corresponding to 0.000144 moles of germane decomposition. Hence, 2.3% of the headspace free germane and only 0.2% of the total GeH4 decomposed.
Conclusion
Overall, the temperature and pressure parameters measured during the deflagration of free germane were in agreement with theoretical and previously reported data (see the table on p. 86). The results demonstrated that when GeH4 is adsorbed in a porous media, deflagration is significantly inhibited. The time dependent experimental physical and chemical data provides enough information to propose the mechanisms responsible for the limited decomposition: First, as the initial germane began to decompose, the adsorbent cooled the expanding gases, quenching further deflagration. Then, the pressure increase resulting from the decomposition caused additional germane in the headspace to be adsorbed. This event dampens the amplitude of the pressure wave in addition to reducing the amount of material available for reaction. Additional testing is planned to support the commercialization of "full fill" SAGE cylinders of 100% germane.
Josep Arnó, Stephen P. Griffing, Edward Sturm, ATMI, Danbury, Connecticut
Michael A. Pikulin, Voltaix Inc., North Branch, New Jersey
Acknowledgments
Leisl Dukheden-Lalla and German Shekk conducted the experimental studies reported here; this article is based on a report of those studies written by Dukheden-Lalla. John de Neufville provided useful editorial suggestions. ATMI and Voltaix jointly supported this work. SDS and SAGE are registered trademarks of ATMI.
References
1. B. Meyerson, IBM J. Res. Develop., Vol. 44, No. 3, May 2000.
2. D. Lammers, Electronic Engineering Times, November 11, 2002.
3. K. M. Kuhn, et al., Proceeding of the 2002 IEDM, December 9, 2002.
4. R.G. Aivazyan, Kinetics and Catalysis, Vol. 39, No. 2, pp. 158–161, 1998.
5. S. Horigushi, et al., Koatsu Gas, Vol. 28, No. 4, pp. 270–279, 1991.
6. S. Horiguchi, et al., J. of High Pressure Gas Safety Inst. of Japan, Vol. 30, pp. 799–809, 1993.
7. K. Olander, SESHA Journal, Vol. 15, #3/4, pp. 9–15.
8. K. Olander, et al., Semiconductor Fabtech, 7th Edition.
9. M. Donatucci, et al., Semi Workshop on Gas Distribution Systems, 1998.
Josep Arnó received his PhD in physical chemistry from Texas A&M University. He is director of research at ATMI Inc., 7 Commerce Dr., Danbury, CT 06810; ph 203/794-1100, fax 203/794-8150, [email protected].
Stephen P. Griffing received his BA from Muhlenberg College. He is product manager for SAGE at ATMI.
Edward Sturm received his BS in chemistry from State University of New York's College at Buffalo. He is a senior research engineer at ATMI.
Michael A. Pikulin received his BS in chemistry from the University of Cincinnati. He is senior VP at Voltaix Inc.