Orbital Welding of Semiconductor Process Gas Lines

Orbital Welding of Semiconductor Process Gas Lines

BarBara K. Henon, PH.D.

A discussion of the advances in semiconductor welding that meet standards and contribute to better, cleaner, and more repeatable welds

Introduction


Figure 1. Orbital welding of semiconductor tubing in
a cleanroom with the AMI Model 207 power supply
(left) and Model 107 power supply (right); pipe stands
are used to support tubing assemblies during welding.
(Photo courtesy of Murray Co.)

Beneath the floor of every microchip manufacturer there is a basement or “subfab” with miles of stainless steel tubing transporting gases used in production. Some of these gases are highly toxic at very low concentrations, some are pyrophoric, bursting into flames on contact with air, and some are highly corrosive. Not only must the piping be corrosion resistant and leak free, but the joining technology must maintain the integrity and corrosion resistance of the system. Furthermore, gases at the parts per billion purity range must be able to pass through the piping systems without accumulating particulates, moisture or other contaminates.

Orbital Welding Technology

Orbital welding technology, by consistently delivering welds of the highest quality, is essential for achieving the high yields of semiconductor devices that might otherwise be compromised by the least amount of particulate contamination. Orbital welding, by definition, is “automatic or machine welding of tubes or pipes in-place with the electrode rotating (or orbiting) around the work.” 1 The welding process is called Gas Tungsten Arc Welding (GTAW) or Tungsten Inert Gas (TIG), which for semiconductor applications is a fusion or autogenous weld, but for other applications wire can be added. The tungsten electrode is unconsumed, thus autogenous GTAW is a very clean process in that nothing is added to the weld.

Orbital welding was developed for the aerospace industry in the 1960s and was adopted for use by the semiconductor industry in the early 1980s. Since that time millions of orbital welds have been made to join semiconductor process gas lines. While orbital welding of semiconductor process gas lines reached its peak about 10 years ago, it is still a major application of orbital tube welding (see Figure 1, 2).

SEMI Standards

Prior to 1993, standards for semiconductor gas lines relied primarily on company specifications. Practices or standard operating procedures (SOPs) were developed by installing contractors to meet the stringent purity standards set by the industry. SOPs are written procedures to assure that all welding operations are performed in the same way by welding personnel. However, at that time there was no clearly established practice and no common set of weld criteria that could be applied to the entire industry. The existing Semiconductor Equipment and Materials International (SEMI) standard at that time, SEMI F3-94, mentioned orbital welding but provided no guidelines.

In 1993 and 1994, SEMI introduced two new standards that apply to orbital GTA welding in semiconductor fluid distribution systems. These standards, which replaced SEMI F3-94 are: SEMI F78-0304 Practice for Gas Tungsten Arc (GTA) Welding of Fluid Distribution Systems in Semiconductor Manufacturing Applications, which is a guideline for fabrication, and SEMI F81-1103 Specification for Visual Inspection and Acceptance2, which details weld acceptance criteria.

The two documents were intended to be complementary: if the procedures outlined in the Practice are followed, a weld capable of passing the criteria outlined in the Specification should result. The new SEMI standards provided guidelines for welders and welding operators while also providing a set of criteria by which quality assurance/quality control inspectors could evaluate welds. A common terminology was defined so that welding personnel and QA/QC could effectively communicate with each other.

In keeping with the scope of SEMI F78, this discussion will be limited to autogenous (without filler) GTA butt welds on tubing and components 6 inches or less in diameter.


Figure 2. A Model 9-500 weld head mounted on an ICF Micro-Align Plate; the plate provides for precision alignment of components for welding.
( Photo courtesy of Arc Machines, Inc.)

Types of Orbital Welding Equipment
Orbital welding power supplies for SEMI F-78 specified GTAW include, constant current, DCEN (direct current electrode negative) and electronically controlled with rapid dynamic response capable of 5 Hz (CPS) or greater pulsed welding. Initially the power supplies were large and cumbersome, but when the first compact autogenous welding power supply that could run on 110 VAC was developed, it was immediately seized upon by contractors who could easily move it around a construction site. The early portable power supplies, such as the Arc Machines, Inc. Model 107, featured digit switch controls. Each time the tube size was changed, the operator would have to dial in all the weld parameters for the new weld schedule.

Weld parameters controlled by the power supply include: prepurge time when the enclosed weld head fills with inert gas; rotation delay time, during which the arc is struck and a liquid weld pool is established; primary and background amps sufficient to achieve full penetration; pulse times for high and low current pulses; travel speed (RPM), typically 5 inches (127 mm) per minute; and downslope time during which the arc is gradually extinguished, followed by postpurge; which continues the gas flow until the weld and heat- affected zone (HAZ) have sufficiently cooled to prevent discoloration of the weld.
Autogenous orbital welds are typically completed in a single pass with the weld current gradually decreased in a series of steps called levels. The Model 107 had 4 levels of current while the newer machines, such as the AMI Model 207 and later models, have a virtually unlimited number. Other power supplies with only one level, can achieve penetration by making more than one pass. The first pass serves as a preheat and penetration is achieved on subsequent passes.

Multipass welds with no pulsation were done in some semiconductor applications in an attempt to create a smoother weld bead. However, weld bead smoothness depends at least as much on material properties as on welding technique. Any of the power supplies that have more than one level of current can be programmed to perform either a multipass weld or a single pass weld, and both approaches can produce excellent welds.
While many of the digit switch machines are still in service today, power supplies were soon developed that took advantage of the technology they helped to build. The next generation of power supplies are microprocessor-based and store weld programs in memory. The operator calls up the appropriate program, or writes and enters his own program from a built-in keyboard.

More recent developments include Windows-based units such as the Model 307 with the capability of real time data acquisition and the ability to maintain weld records for QA/QC purposes. When a weld head is specified and the outside (OD) diameter and wall thickness are entered, the units will generate a weld schedule that requires little or no adjustment to produce an acceptable weld.

The latest technology, represented by the AMI Model 205, allows for ease of use. When the starting amperage is entered, the power supply gradually reduces the current to produce a uniform weld bead width around the entire weld joint. The amount of current slope can be adjusted to achieve the desired amount of penetration, and the unit can be programmed in degrees, which assures that the weld head will perform a complete 360 degree weld plus tie-in before beginning the downslope.

Autogenous orbital tube welding is completely automatic in that the welding operator installs the tubing and/or fittings in the weld head, starts the ID purge, initiates the weld sequence by pushing a button and does not make any adjustments during the weld.

Remote operating pendants are typically available for power supplies for use in the field. Operators can initiate the weld from the pendant where the power supply may be located at some distance from the weld joint.

Types of Orbital Weld Heads

The type of weld head used for high-purity and ultra-high purity semiconductor applications is fully enclosed so that the entire weld joint and electrode are protected by the shielding gas. The head is filled with inert gas, usually argon or a mix of argon with no more than 5% hydrogen, prior to striking an arc. A tungsten electrode installed in a rotor moves the electrode around the weld joint while the head and the tube remain stationary.
SEMI F-78 requires that weld heads and fixtures be clean and free of any particulates and excessive discoloration. They must rotate freely and smoothly at all speeds and hold the tubes or fittings firmly in place. There must be a provision in the fixture for viewing the weld joint to insure proper fit up.

The electrodes must be precision ground to the factory specification for the head and weld type. The electrodes may be thoriated, ceriated, or lanthanated3 rather than pure tungsten to facilitate arc strike. However, in our experience, ceriated provide the most reliable arc strike. The tungsten length must be fixed for each size tubing to assure a uniform arc gap set to within 0.002 inches (0.051 mm) that remains the same for every joint of similar outside diameter (OD) and wall thickness.

Each weld head can accommodate a range of tube diameters, which for most heads is accomplished by changing the tube or fitting clamp inserts on either side of the head.
The most commonly used head for semiconductor process gas lines is the Model 9AF-750. This head can weld sizes from 0.250 inch (6.35 mm) OD to 0.750 inch (19.05 mm) OD. It is a very rugged head suitable for use in the field, or in the shop but can be used in the cleanroom as well.

The Model 9-500 welds diameters from 0.250 inch OD (6.35 mm) to 0.500 inch (12.7 mm) OD. It uses a set of tools to configure the head for tube-to-tube, tube-to-fitting or fitting-to-fitting welds. Initially this head did not have water cooling and that limited weld productivity; but eventually, water cooled heads became available and with this feature up to 300 welds per day became possible.

Water cooling is particularly recommended for tube diameters of 3 to 4 inches OD and larger. While most semiconductor gas lines are 0.250, 0.375, or 0.500 inches in diameter, sizes up to a 6 inch pipe and larger have been done with orbital welding, however, a different type of equipment is required for tube sizes greater than 6 inches
(152.4 mm) OD.

End Preparation
A precision fit between components being welded is essential for achieving repeatable high quality welds. In order for orbital welding to be successful in high purity or ultrahigh purity applications, the weld ends must be machined with tolerances of 0.003 inches (0.762 mm) from a plane perpendicular to the centerline of the tube with an OD/ID burr of less than 0.005 inches (0.127 mm). No oils or lubricants are permitted and the cut curl must be controlled so that metal particles do not enter the tubing or scratch the ID surface.

For tube cutting or parting, wheel type cutters or orbital cutoff saws are allowed, but this must be followed by the specified cleaning procedure. If end tolerances are not met by these techniques, follow-up end preparation with a facing tool is recommended. Facing, bending and cutting operations are performed with a positive purge.

Inert Gas Purging

Purging is an essential part of the GTAW process in which inert gas protects the tungsten electrode and weld pool from discoloration caused by oxidation during the weld, and while the weld cools to ambient temperature. A purge on the ID surface is particularly critical as discoloration is associated with loss of corrosion resistance, and oxidized metal may particulate from the surface and place the semiconductor product in jeopardy.


Figure 3. Top: Prior to orbital welding, a Magnehelic gauge is connected to the weld components
via a tee, and the flow rate adjusted to give the appropriate pressure at the joint.
Bottom: Once the flow rate is set, the Magnehelic and tee are removed from the circuit,
the weld head installed and the weld is completed. (Courtesy of Arc Machines, Inc.)

SEMI F78 12.6.1 states that all welds must use a positive and repeatable form of ID purge pressure control. The technique of using a Magnehelic pressure gauge to control the flow rate during welding is shown in Figure 3. Table 1 in SEMI F78 details the flow rate, purge pressure and size of flow restrictor for each tube diameter and wall thickness from 1/16 inch (1.588 mm) up to 6 inch (152.4 mm). The purge gas must be measured and controlled with separate flowmeters used for both the ID and the OD purge. This technique is useful for controlling the weld bead profile and preventing weld blow-out that may occur in field welds with excessive ID pressure. The weld assembly must be kept under a continuous purge until all welding is complete.The purge gas apparatus must be stainless steel tubing with face seal fittings for high purity and UHP applications. Less critical applications permit PFA plastic tubing for the final run.

The ID purge gas must be certified to less than 3 parts per million (ppm) total moisture, oxygen and other contaminants. While color is the final weld criterion, oxygen analyzers are frequently used to show that gas
exiting the ID purge is of the same level of purity as that directly from the source. Purifiers are used in the purge lines to remove any trace amounts of water or oxygen from the purge gas.

Weld Criteria SEMI F81
SEMI F81 provides visual inspection and acceptance criteria for GTA welds on stainless steel and other corrosion resistant alloys in fluid distribution systems for semiconductor manufacturing applications. A good weld is fully penetrated with good alignment, and with a uniform well-spaced weld bead; it has a flat inner and outer weld bead with minimal concavity or convexity; there are no visible defects such as cracks, porosity or slag islands, except a small slag inclusion at the end of the downslope—less than 10% of the wall thickness may be acceptable; and there is no ID discoloration due to oxidation when viewed under a bright fluorescent light without magnification for HP and UHP applications. Slight oxidation of the weld OD is permitted. There must be a downslope sufficient to prevent a crater from forming at the end of the weld. Any tack welds must be completely consumed, that is, the arc has to re-melt the tack weld so that it is not visible in the weld.

Weld Coupons and Qualified Welding Procedure


Figure 4. A good weld is fully penetrated with a uniform weld bead, no visible defects and virtually no discoloration from
oxidation. (Photo courtesy of Therma, Inc.)

While borescopic or direct visual examination of welds may be practical for assemblies of small diameter tubing in the clean-room, it is not practical for examining welds on closely spaced small diameter tubing in the field. The semiconductor industry has used couponing as a method of weld examination, based on the assumption that the orbital GTAW process is a repeatable process. So, if the sample weld is good, the subsequent welds made on the same tubing with the same weld parameters and purge conditions will also be good.

Couponing is done in the context of a qualified weld procedure which is detailed in Fig. 1 of SEMI F81 Welding Procedure Flow Chart. All materials must be verified; the weld procedure and welding operators qualified to ASME Section IX of the Boiler and Pressure Vessel code or AWS (American  Welding Society) B2.1.

Prior to the start of production welding on a particular size, wall thickness and alloy, a primary sample weld is made and visually examined on the OD for weld bead appearance, joint contamination, joint soundness, surface oxidation, discoloration, pitting, cracking, and defects of fit-up and workmanship. The sample weld is then cut open (sectioned) to show the entire ID weld bead and analyzed at the job site. The sample weld is checked against the weld criteria in SEMI F81 for penetration, bead concavity, bead variation and oxidation. Once the primary standard sample has been deemed acceptable, the primary sample becomes the on-site work sample against which other weld coupons are evaluated.

Sample welds are made periodically at the beginning and end of a shift, when weld parameters are changed, when the material heat number4 is changed, for a change in tube size or wall thickness, a change in ambient temperature of +/- 20 degrees F, or when the power source or welding machine, weld head, electrode, etc. is changed. The ID purge source used for couponing must be the same as that used for production welds.

Production welding is permitted only after the sample weld has been accepted. The sample weld may be a production weld only when it is possible to examine the ID with a borescope or sight tube. The welding procedure details the conditions under which welds may be repaired or when they must be cut out.

Each weld has a unique number and is cross referenced to a specific drawing. Traceability is assured by a weld log which must be maintained for all welds, including coupon welds.

Operator training

Operator training is essential. Welding operators must not only be able to operate the welding equipment, but must understand and be able to carry out welding procedures as described in SEMI F78. Training is provided by the welding equipment manufacturers, and training classes in orbital welding and cleanroom protocol are offered for apprentices by the United Association of Journeymen and Apprenticeships in the plumbing and pipefitting industry of the United States and Canada5.

Materials

AISI Type 316L stainless steel has been the material of choice for semi-conductor process gas lines. Other corrosion resistant alloys, such as hastelloy C-22, have been used to a lesser extent with corrosive gases. Refinements in steel production technology have lead to materials relatively free of inclusions and other contaminants , and when welded, have exceptionally smooth weld beads.

The Effect of Sulfur on Welding
One of the initial problems in welding Type 316L was the wide range in sulfur content in the AISI (American Institute of Steel and Iron) specification that specified a maximum sulfur content of 0.030 wt.%, but no minimum.

Sulfur has a dynamic effect on weld pool characteristics. The direction of fluid flow is based on surface tension, (i.e. the Marangoni effect) which has a positive correlation with temperature when sulfur is present and a negative correlation with temperature at low concentrations.

When the sulphur concentration is at the mid- to high-range for the AISI specification, the direction of fluid flow in the weld pool is from the edges of the weld pool inward creating a weld with deep penetration. At the low end of the sulfur specification, typically less than 0.005 wt.%, fluid flow in the weld pool is reversed so that, in this case, the fluid flow is from the center of the weld pool outwards creating a wide, shallow weld bead. Other elements such as oxygen and selenium have similar, but less noticeable effects.

Since sulphur combines with manganese to form inclusions, the surface finish of the higher sulfur heats is unacceptably rough. The semiconductor industry elected to specify sulphur concentrations at the lower end of the range. SEMI F206 specifies an upper limit of sulfur of 0.012 wt.% for general purpose tubing and 0.010 wt.% for high purity and ultra- high purity tubing.

Sulfur mismatch, which occurs when one heat of material2 is significantly higher in sulfur than the other component being joined, can result in the weld pool shifting towards the low sulfur heat. This is less likely to occur when both components are in the lower sulfur range, but welding of very low sulfur materials requires support of the weld puddle during welding by pressurization to achieve an acceptable weld bead profile. This must be done in a controlled manner to achieve consistent results as specified by SEMI F78.

Conclusions

Orbital GTA welding has been of fundamental importance in the development of semiconductor fabrication technology. There have been advances in the welding technology, improvements in welding equipment, increases in gas purity, and developments in materials science that have all contributed to better, cleaner, more repeatable welds. This would not have been possible without orbital welding.

The SEMI Standards, particularly SEMI F78 Practice, reflects a tremendous amount of technical expertise by end users, installing contractors, welding operators, weld equipment manufacturers, and quality assurance personnel. Their combined experience has gone into the development of practical welding specifications for this technologically demanding industry.

References
1. SEMI F78-0304 Practice for Gas Tungsten Arc (GTA) Welding of Fluid Distribution Systems in Semiconductor Manufacturing Applications
2. SEMI F81-1103 Specification for Visual Inspection and Acceptance that details weld acceptance criteria.
3. Thorium, cerium and lanthanum are types of tungsten differentiated by the dopant that is added to the pure tungsten tomake it more conductive and thus facilitate arc strike.
4. Different melts of steel, even of the same type of steel, are differentiated by a “Heat Number”. This is traceable back to the steel mill and provides a list of the base metal, alloying elements and some trace elements in that specific melt of steel. The percentage of alloying and trace elements affect the weldability of each heat of material to some extent.
5. United Association of Journeymen and Apprentices of the Plumbing and Pipe Fitting of the United States and Canada www.ua.org
6. SEMI F20 Specification for 316L Stainless Steel Bar, Forgings, Extruded Shapes, Plate, and Tubing for Components use in General Purpose, High Purity and Ultra-High Purity Semiconductor Manufacturing Applications.

Barbara K. Henon, Ph.D., is a contract employee for Arc Machines, Inc., 10500 Orbital Way, Pacoima, California 91331. Dr. Henon joined Arc Machines in 1984. She was an Instructor in the Arc Machines Training Department and wrote training manuals for welding operators. From 1984 to 1994 she performed orbital welding training for operators in the field for semiconductor, biopharmaceutical and other industries that specify orbital tube welding. Dr. Henon joined the ASME Bioprocessing Equipment (BPE ) Standards Committee in 1989 and served for two terms as Vice Chair of the Main Committee. She is currently a member of the BPE Materials Joining and Sur face Finishes subcommittees and the Subcommittee on General Requirements. She is the official Liaison between the ASME BPE and ASME B31.3 Process Piping Committee which is currently writing a High-Purity chapter. She was a member of the task force that wrote the SEMI F79 and F81 Standards for Orbital Welding in the semi conductor industry. She also serves on the American  Welding Society (AWS) D10 Committees and D18 Committees. Henon can be found on the Arc Machines, Inc. website, www.arcmachines.com under Applications. She can also be reached at barbara.henon@arcmachines.com or at 206-546-9601.