Sample Manipulation Technical Notes
Many different terms are associated with sample manipulation and the devices used to achieve it: motion feedthroughs, sample transfer, translation devices, xyz manipulation, z-only manipulation, sample rotation, rotary drives, linear drives, etc. All cover different aspects of the same basic requirement— to mechanically move an object that is inside a vacuum chamber and under vacuum. Such devices can provide precise, repeatable movement or coarse positioning. They may move the object just a few microns or a few feet. They may provide rotary motion, linear motion, or a combination of both. The most complex devices give motions in both linear and rotary axes. Many products offer the option of a variety of actuations including manual, pneumatic, and motorized. When selecting travel with motorization option, please note a small amount of travel is lost at either end due to the limit switches. For example, 100mm travel becomes 98mm when motorized.
The mechanical movement may be generated by two basic mechanisms: a vacuum-tight seal on a mechanical device that moves through the vacuum wall or a magnetic coupling that transfers motion from air-side to vacuum-side.
A rotary motion drive is essentially an air-side "handle" that rotates a vacuum-side shaft or tube. At least five types of rotary drive exist, differentiated by actuator mechanism, vacuum sealing, compatible pressure range, and application.
An outer (air-side) rotating ring has a number of strip magnets (the strips parallel to the ring's centerline) mounted so one magnet's poles are opposite in sign to its immediate neighboring magnets. This outer ring magnetically couples through a stainless steel vacuum sheath to a vacuum-side ring with an identical number of strip magnets. A coupling places an outer N pole over an inner S pole.
The inner magnet ring is fixed to a shaft rotating on two (MoS2 impregnated) ball bearings. The lack of mechanical coupling and one-piece construction removes any possibility of leaks. All selected construction materials enable baking to 250°C, making this a rotary drive with high vacuum and UHV compatible. The maximum torque transmitted is determined by the force that decouples the inner/outer magnets and for larger rotary drives this is ~40Nm. If the magnetic coupling is not under high torque, this drive gives very precise rotary motion. It can be mounted in any orientation and has a long life under continuous rotation (max. 500—1000 rpm). The trade name for this popular device is MagiDrive™.
The vacuum seal on this rotary drive is an elastomeric o-ring or "knife-edge" attached to the stationary body and rubs on the rotating shaft in what is called a "dynamic seal." Gas leaks, permeation through the elastomer, and seal wear limit this drive to 10-5 Torr or 10-6 Torr range at best. They are used in applications where rotation is intermittent, rotation speeds are <100 rpm, there is little side-loading on the shaft, poor vacuum conditions are acceptable, but cost mus be low. Any mounting orientation is permitted.
This rotary drive is basically a flanged cylinder with two roller bearings supporting a central rotating shaft. The shaft, a high magnetic-permeability material, is machined in a series of circumferential peaks and valleys (in section it looks like a cross-cut saw). A ring magnet, mounted in the cylinder, surrounds the shaft, creating a small gap between the shaft's peaks and the ring. This gap is loaded with a ferromagnetic fluid—a low vapor pressure fluid in which extremely small magnetic particles are suspended.
The field concentration effect of each peak causes the ferrofluid to form liquid o-rings that can sustain a 70 Torr differential pressure. That is, designs with >11 peaks provide a non-wearing vacuum seal against atmospheric pressure.
In general, ferrofluid outgassing limits this drive to applications above 1 x 10-8 Torr.
Some models are capable of high torque loading and high speed (10,000 rpm) with long life under continuous rotation. They can be mounted in any orientation and the continuous shaft means these drives provide precise rotations. The trade name for one manufacturer is Ferrotec™.
Linear motion devices, like rotary drives, are essentially air-side "handles" that control some motion in the vacuum. Unlike rotary drives, linear drives are differentiated by what moves and how far it moves, rather than the sealing mechanism.
Positioners are bellows-sealed or magnetically coupled rods that move along the rod's axis. The mechanisms are either manually or pneumatically actuated, enabling a push-pull motion between two stop positions. Another method is a precise screw mechanism, with manual or motorized actuation, that can be stopped at any intermediate position between its travel limits.
Positioners are used in applications needing straight-line, fixed distance movement; for example, beam stops, shutter actuators, substrate movements, etc., at all pressures between atmosphere and UHV.
Linear positioners sealed with elastomeric o-rings or "knife-edge" seals are available and perform the same functions. However, like other "dynamic seal" devices, they are compatible with pressures between atmosphere and ~10-5 Torr.
A linear shift is a pair of flanges connected by a bellows. One flange is free to move along its axis in relation to the other. The motion is constrained by a rugged, precise slide mechanism so the flange faces are always parallel. Linear shifts may be viewed as the linear version of the rotary platform; that is, they enable wide diameter devices to be inserted into the vacuum chamber and moved linearly between limits. The action is similar to the Z-only manipulator, but its positioning precision and travel length (and therefore cost) are lower. Linear shifts are actuated by hand-wheels or, for higher precision, stepper motors. They are compatible with high vacuum and UHV, and find applications moving heavier, larger-diameter loads than positioners.
The term sample transfer covers two different motions: (1) Long travel between two chambers or a load lock and chamber, through vacuum isolation valves using equipment called transporters, probes, or transfer devices; and (2) 3-D motion needed to transfer a sample from a motion device or storage area to a sample stage/holder. The latter is more mechanically complex, enabling intuitive XYZ motion and having a "hook" to attach to the sample and some degree of rotation to enable sample alignment. They are called wobble sticks or mechanical hands.
Of the long travel transporters described, three are associated with sample-loading components such as load lock doors or sample entry port. The fourth—radial distribution center—combines long travel transportation with directional choice and is associated with moving samples between chambers rather than just sample entry. Unlike short-travel devices, long-travel linear drives are subject to gravitational "droop" at their horizontal travel limits.
The familiar long-travel transporters produce motion perpendicular to their mounting flanges. A less common variety, called a bi-directional transporter, produces motion parallel to its mounting flange's face in both left and right directions.
The probe—the component that moves—has a "rack" gear along its length. It is supported on two groups off our roller bearings (dry lubricated) and is moved by a "pinion" gear on a rotary drive mounted between the bearings. These gears give the device its common name, linear rack & pinion probe or LRP transporter.
The rotary drive is either wobble-bellows or magnetically coupled for true UHV applications, or Ferrofluidic® type for high torque and heavy-load applications. The travel distance can be precisely controlled (particularly with motorized drives) with good position repeatability. Since the drive's actuator is mounted close to the chamber, in the manual version one person can simultaneously move the probe and observe its progress through a viewport. The normal maximum for travel distances is 44" (112 cm) but examples have been built with travel distances up to 96" (244 cm).
The probe's "far end" (farthest from the chamber) is attached to an in-vacuum magnetic array. Outside the stainless steel vacuum envelope, a cylindrical magnet actuator couples with the internal magnet. Moving the external magnet moves the probe. Unlike the LRP, this probe's magnetic coupling enables rotary motion, too—leading to its common name of linear/rotary magnetic probe (for which the acronym LRM is sometimes used). The probe is UHV compatible and has a maximum travel distance of 60" (152 cm).
The manual magnetic transporter's one disadvantage is that its actuator's location— at the farthest point from the chamber—makes simultaneous movement and probe observation difficult. However, motorization overcomes this.
For maximum load capabilities, "high power" LRMs have a (vertically) magnetic decoupling force up to 310N (180N Standard) and rotary decoupling forces of 4.5Nm. The magnetic transporter's major advantage, particularly for automatic docking/ transfer, is its rotation through small (misalignment) or gross (mounting orientation) angles.
While a tilt mechanism is still a bellows connecting two flanges, directional changes are made by a mechanism not used for XY stages or Y-shifts. The flanges have three threaded rods, parallel to the bellows axis, mounted on their periphery. The rods are locked into the fixed flange and have a nut on either side of the moving flange. Tilt motion is induced by first loosening all the top nuts on the moving flange. The bottom nut on one rod is moved (toward or away from the fixed flange) and all the top nuts re-tightened. The result is, the two flanges are no longer parallel. This port aligner has identical flanges at both ends and is available in 23/4" CF (DN40), 41/2" CF (DN63), and 6" CF (DN100) flange sizes.
The nominal maximum tilt angle depends on the distance between the flanges which, in turn, depends on the flange's O.D.: 23/4" CF ±5°, 41/2" CF ±4°, and 6" CF ±3°. Tilt mechanisms can be baked to 250°C.
Short travel devices used for sample movement rarely involve a sample transfer step. By contrast, long travel transporters almost always require sample transfer from the probe to a sample stage/holder. The wobble stick is a device that transmits simple hand movements through the vacuum wall.
A magnetically coupled drive provides linear/rotary motion while bellows provides "wobble" motion, enabling the wobble stick's end to reach positions in an "action cone." On the vacuum side, the probe terminates in a mechanism to attach/detach the sample such as hooks, forks, screw-thread bosses, pincer grips, etc.
Because the magnetic coupling enables complete rotation, the screw- thread sample attachment system is regarded as the most secure sample transfer available.
Essentially, the radial distribution center—RDC is a bi-direction transporter mounted in its own (ovoid-shaped) chamber and allowed to rotate (in the horizontal plane through 360°) so the BDT's direction of travel coincides with the centerlines of 4, 6, or 8 ports spaced around the RDC chamber's circumference. The BTD's flange is mounted in the chamber's top surface with the pumping port directly below in the lower surface.
The RDC transfers samples between multiple chambers attached to its ports, one of which is usually a load lock for sample entry/removal. The completed arrangement is not unlike process chambers surrounding the central robot in a cluster tool.
Most RDCs are UHV compatible and used to connect high vacuum and UHV chambers. One common application is interconnecting chambers designed as/for load lock, plasma etch for sample cleaning, thin film deposition, surface science experiments, and analytical techniques (SIMS or ellipsometry).
Any alignment errors are greatly magnified by long travel transporters. Some of these potential error sources are:
- Flange face not perpendicular to the tube axis on transporter, load lock, gate valve, or chamber port
- Chamber port's centerline not aligned with sample transfer position
- Gravity's action causes the probe to "droop"
Unfortunately, alignment can only be effectively checked when the system is fully assembled and under vacuum. Clearly, to correct these errors, devices were needed that moved the probe's position while the system was still under vacuum. Many such "direction adjustment" devices were developed and became known collectively as port aligners.
Some long travel transporters are required to carry loads that exceeded the capabilities of the precise XY stage. For such applications, heavy-duty Y-shifts were developed. Their basic design is like the XY stage—two flanges (one of which moves) are joined by a bellows and built into a cage. However, travel is restricted to one direction so a much sturdier cage and motion actuator (handwheel or motor) construction is possible. Two versions are available (giving fixed flange O.D. first): 4-1/2" CF to 4-1/2" CF (DN63 to DN63) with a motion of ±7.5 mm (0.295") and 6" CF to 4-/2" CF (DN100 to DN63) with a motion of ±31 mm (1.22").
Although this port aligner is called a Y-shift adjuster, its direction of motion depends on the bolt-hole orientation when the Y-shift is attached to the system. That is, if up-down is not the directional change needed, the shift is rotated 90° before mounting to give left-right motion.
In the XY stage, a short flexible metal bellows joins two flanges. A cage holds the flanges at a fixed distance apart and parallel but enables one flange's axis to move laterally (in the XY plane) with respect to the other flange's axis. While XY stages can have same-size flanges, in most arrangements the fixed flange is larger than the moving flange. Two common sizes are: 4-1/2" O.D. CF to 2-3/4" O.D. CF (DN63 to DN40) and 6" O.D. CF to 4-1/2" O.D. CF (DN100 to DN63). The moving flange's motion is controlled by two vernier drives acting on two cross-roller slides set at 90°. This enables the XY position to be precisely set or re-set.
A device that precisely moves a sample to any point in space and any rotational orientation (within the design travel limits) is called XYZ manipulator or occasionally XYZ translator. There are six degrees of freedom for such movement, one along each X, Y, and Z axis, and three rotations about these axes. For most practical applications, no more than five (X, Y, Z plus two rotations) are necessary. In the description below, the word motions is used only to describe travel along the X, Y, or Z axes (when no specific direction is implied). Similarly, the word rotations covers generalized rotation.
The sample is typically mounted at some central position inside the vacuum volume, enabling access for instrumentation or processes. The actuators and devices controlling the motions and rotations are all outside the vacuum volume—motions sealed by flexible bellows and rotations sealed by wobble bellows, linear-acting bellows, or magnetic couplings.
Accessories/ancillaries used with XYZ manipulators include:
- Sample/substrate holders or stages
- Heating stages (to raise the sample's temperature)
- Cooling stages (for cryogenic studies on samples)
- Tilt devices (to incline the sample's support probe)
- Motorizing motions and rotation
If a manipulator is viewed with its Z-travel bellows vertically in front of the observer and its support structure behind it:
- Y-axis runs through the bellows' center and the support's center
- X-axis runs left-right (through the Z-bellows' axis)
- Z-axis is the bellows' axis
Sample stages/holders at the probe's end may have the sample's surface on the Z-axis or displaced along the Y-axis, either in front of, or behind the Z-axis. This displacement is called offset. However, whatever the position of the sample's surface, if its (flat) surface is vertical and facing the observer standing in front of the manipulator, the rotations around the various axes are called:
- Z-axis: R1-, primary-, polar-, theta-rotation
- Y-axis: R2-, azimuthal-, phi-, alpha-rotation
- X-axis: R3-, tilt-, third-, flip-, beta-rotation
XYZ manipulators are either "single bellows" or "dual bellows". In both cases, the sample is mounted on a probe supported from the top flange of the Z-travel device. That means the distance from probe support point to sample may be long. Since the probe's "moment arm" will amplify vibrations or motions, the probe has a large diameter to increase its stiffness and rigidity.
Single Bellows manipulators must enable the probe to travel in X-, Y-, and Z-direction through the one bellows. That is, the bellows inside diameter (I.D.) determines the extent of X- and Y-travel and, for large probe movements, the I.D. must be large. To illustrate how large, consider a probe 3/4" (19 mm) O.D. in a bellows 1-1/2" (38 mm) I.D. The maximum X- or Y-travel is ±3/8" (±9.5 mm). To substantially increase the XY-travel, wide-bore bellows are required. Combining that with long Z-travel means the bellows are not only wide, they are also long. Long, wide-bore bellows are exceptionally expensive.
Dual Bellows manipulators separate Z-travel from X- and Y-travel. The Z-travel uses a small I.D. bellows that is as long as needed for the travel distance. For X- and Y-travel, one short, large I.D. bellows is used. The manipulator is constructed with the Z-travel bellows, its entire support structure, the Z-travel's actuator, any rotary drives for rotating the sample, and the probe, built onto a rugged top carriage. The top carriage, in turn, is mounted on a lower carriage. The manipulator is mounted in the lower carriage and connects to the top carriage by the short, large I.D. bellows for the X- and Y-travel. The top carriage is supported on the lower carriage by rearing slides—mounted at a 90° angle—that precisely moves the top carriage in X- and Y-directions relative to the lower carriage. Despite the two bellows manipulator's complexity, it is a more practical and cost-efficient solution to large X-, Y-, and Z-travel motions than a long, wide-bore single bellows design of identical travel limits.
In some manipulator designs, the R1 rotation is continuous. But heating and cooling options limit the rotation to less than a full circle. R2 (and the more unusual R3) rotation is actuated typically by a push-rod movement and limited, normally, to a few degrees. When considering sample holders and rotation, system designers must ensure sufficient clearance (usually around the primary axis), so the sample holder's sweep radius does not cause it to collide with chamber fixtures during the traverse.
As their name implies, sample holders secure the samples so the manipulator's motion is translated into an identical sample motion. The sample holder can be mounted directly to the XYZ manipulator's probe but is frequently connected to the shaft of a rotary drive, mounted through a hollow probe, to provide R1 rotation. Where R2 motion is needed, the sample holder is designed to include mitre gears that causes rotation (around the pinion's axis) when the rack is moved by a push-rod motion transmitted through the hollow probe.
Heating and cooling options are also sample holder functions. Sample heating can be accomplished by electron bombardment (EB) or thermal radiation (IR). For EB, the sample holder is equipped with a filament capable of thermionic emission. The filament is electrically isolated (or the sample is electrically isolated) and biased to make the filament a high negative potential. This causes electrons to bombard the sample at high energy. For IR, a filament is mounted close to the sample's backside and resistively heated to a high temperature. The IR from the filament heats the sample. Typical maximum temperatures depend on many factors, including the sample material, radiation shielding, sample emissivity, etc., but range from 800°C to 1200°C LN2 cooling is also available as an option.
A tilt device is a bellows connecting two flanges similar to the tilt mechanisms noted under Direction Adjustment. In this device, however, the flanges' non-parallelity is controlled by three vernier screw drives around the flanges' periphery. Moving one or more verniers from the neutral position forces the flanges' faces out-of-parallel. The probe can be pointed in any direction, within a defined cone angle, relative to the manipulator's axis.
To achieve the highest positioning accuracy, resolution, and repeatability, XYZ manipulator motions/rotations are motorized with air-side mounted stepper motors. Motors are also used for long Z-travel-only manipulators, where making long travel movements by hand can be arduous (because of the exceptionally fine gear ratios), and on XYZ manipulators located at inaccessible points on large systems.
Why rotate a substrate? Rotation is often associated with thin film deposition. If the flux of material from a PVD source or the precursor gas's local concentration from CVD "shower head" is nonuniform, the film's growth on a static substrate will have a non-uniform thickness. One solution is to rotate the substrate and expose its entire surface area to the more uniform average flux.
Why heat a substrate? Heating is often, again, associated with thin film deposition. For PVD techniques, the substrate's temperature influences the surface migration of depositing atoms and the film's morphology—determining its porosity and grain structure. For CVD techniques, the substrate's temperature is the reaction driver, converting the volatile precursor into the solid film.
What is a transfer stage? In some applications, the chamber is vented and opened to load the next substrate on the rotation/heating stage. For R&D applications and some (low base pressure) process applications, however, multiple substrates are stored under vacuum in the main chamber or an adjacent load lock. Substrates are exchanged while still under vacuum and, of necessity, the rotation/heating stage's design must allow such transfer.
There are countless design variations of substrate stages. The simplest are fixed mounts without rotation, heating, or transfer. The next level provides rotation for a single disk about its axis. They are often called platens or turntables. Other platens carry multiple substrates that rotate about the platen's axis or, with suitable gearing, about their own axis, as well as the platen's axis. The basis of this last design is carried forward with hemisphere or dome "planetaries" carrying multiple disk substrates. The substrate not only rotates about its own axes, but also rotates through different spatial positions on the dome. These are used in production batch coating tools for coating many, relatively small diameter substrates. Almost all rotating platens are driven by an externally mounted, motorized rotary drive.
Because heating and transfer only add to the confusion of variations, the characteristics below focus on one of the more sophisticated substrate heater-rotator-transfer stages commercially available—EpiCentre™.
Substrate rotation is variable up to 80 rpm using a DC motor or stepper motor. The former is preferred for smooth continuous rotation, and the latter if stage indexing or substrate positioning is required for in situ RHEED measurements. The motor is connected through the vacuum wall to the rotating torque tube by a magnetically coupled rotary drive. The absence of gears and use of well-proven ball races ensures hydrocarbon-free operation with negligible particulate generation. The DC motorized rotary drives are available as single, double, or triple stack arrangements to match the specific cradle inertias. The complete stage is bakeable to 250°C, and is compatible with UHV, high vacuum, and reactive gas environments.
Making a continuously rotating electrical connection that does not arc under vacuum is considered a huge engineering challenge. Disk platens are heated by adding stationary resistive heating elements or high-temperature quartz-halogen lamps to the "backside" of the platen structure. This form of vacuum heating involves IR radiation heat transfer. A sample's reflection, absorption, and transmission in the IR region are critical to its equilibrium temperature for a given heat input— that is, a "maximum temperature" quoted without the material's characteristics must be treated with caution. Based on a molybdenum block sample, the maximum temperature can be as high as 1200°C.Heater Modules
The choice of heater module depends on the application. Pyrolytic graphite coated graphite heaters are robust, reliable, reach high temperatures, and provide good temperature uniformity. Pyrolytic boron nitride coated graphite heaters have a considerably smaller area of exposed graphite, which minimizes "doping" for those applications sensitive to carbon. Pyrolytic boron nitride heaters have a layer of (grown) PBN covering an etched (patterned) graphite film deposited on a PBN disk, and are designed for long life, low current heating.
An interesting modification is the quartz-enclosure heater, which has its element surrounded by a vacuum-sealed quartz "box" that completely separates the heater's environment from the process environment. The "box" seals against an internal water-cooled flange, creating the secondary vacuum enclosure, which is separately pumped via a port on the stage's service collar. This heater is used where particulate cleanliness is critical, corrosive gases are present, or where high O2 partial pressures are necessary. Because the element is farther from the substrate and radiates through the quartz, the maximum sample temperature is reduced to 1000°C for a molybdenum block.Temperature Measurement
Selecting the correct thermocouple type is essential to prevent corrosion or process contamination. The common types used in substrate heating are Type "C" (tungsten/rhenium)—non-magnetic, high maximum temperature, and is generally more resistant to chemical attack and Type "K" (Chromel/Alumel)—costs less, but lower maximum temperature, and is magnetic. The temperature across the central 90% of a Si wafer has nominally ±2.5°C variation.
Placing a single linear shift between the substrate stage and the chamber mounting flange enables the entire stage, including wafer cradle and heater, to move as a unit along the stage's axis (normal to the substrate's surface). This motion is often used to vary the throw distance between substrate and a (PVD) source.
Placing a single linear shift between the heater's service collar and the substrate stage's body enables the heater to be moved away from the substrate cradle, providing a gap through which the substrate can be manually extracted.
By installing both shifts, the stage may be manipulated in a typical "wafer hand-off" routine. The heater is first retracted to provide access to the substrate and then the entire stage lifted and/or lowered as needed by the substrate hand-off motions. In addition to manually operated substrate hand-offs, the shifts may be motorized or even pneumatically actuated, enabling automated substrate exchange using PLCs.
The substrate holder, also called a substrate cradle, is made from refractory metal with various designs to hold a standard wafer 25mm—300mm (1"—12" diameter) or multiple substrate disks. The holders can be isolated to enable substrate bias up to 2kVDC (and a lower level of RF bias).