July 2008 Vol. 141 No. 1

Understanding Surface Finish Is Important

The message is in the molecules

Robert B. Aronson, Senior Editor

To order a hardcopy reproduction of this article, click here. To purchase digital reprints or reproduction licenses, please contact the resource center at service@sme.org or call (800) 733-4763.

One answer to the constant quest for six-sigma quality and performance may be hidden in the surface of your product. The microscopic features of a part's surface may hold the reasons behind a product's failure, reveal the effectiveness of a new process, or validate specs.

For some time, most of industry has been content with R_a readings. But this gives only an average height of surface features. There are a number of other parameters that will tell you more. They include: root mean square roughness R_q, skewness S_k, maximum peak height R_p, maximum valley height R_v, and maximum peak to valley height R_max. Any or all of these can provide critical information about a part. (Parameters are separated into "R" types for instruments that produce 2-D information and "S" types for those producing 3-D data.)

The term surface texture or morphology refers to the primary profile, chieflyroughness and waviness, and other surface attributes such as the direction of the surface features.

The mechanical profilometer is the oldest and least costly of the surface-measuring instruments. In operation, a fine stylus, typically diamond, is pulled across a test specimen. The stylus moves up and down as it travels over the surface and these motions are recorded. The result is a 2-D image of a part's surface along the stylus path. For a 3-D image, a number of parallel passes are made and the results "stitched" together to create a "topographical map" of the surface.

The two main types of this instrument are skidded and skidless. With the skidded type, a washer-like fixture is fitted around the profilometer's stylus tip to provide a reference plan from which measurements are made. This method is used when only shorter special wavelength features are of interest.

The skidless version has a wider travel range, and can give a more complete surface profile. It detects both long (waviness) and short (roughness) wavelength features. A good unit can detect height variations as small as 1 nm. At the other end of the instrument spectrum are the noncontacting profilers. They scan the test subject with a beam, usually laser or white light, and produce 3-D images of a surface.

In operation, the instrument uses white light at an optical wavelength. A beam splitter sends part of the light to a reference surface, and the other is reflected from the test object. When both beams are recombined by an interferometer, fringe patterns are created that are converted to a topographical map of the part's surface. The operator gets readings by focusing on various depths of the part's features. Lenses of different focal lengths are used.

"One of the major trends among manufacturers concerned with surface finishing is a move away from doing their own shop-floor measurements," according to Len Carravallah of Mitutoyo America Corp. (Aurora, IL). "They want to know the condition of a material when it comes in the door, so the suppliers are becoming responsible for surface finish checks."

Mitutoyo offers a wide range of contact profilometers, ranging from hand-held instruments to units designed for laboratory work in both skidded and skidless types. One of the more popular units is the SJ 400. The unit can measure 36 roughness parameters. Measuring range is 80µm with a resolution of 0.000125 µm. The unit has a curved-surface compensation function to use when evaluating cylindrical forms.

"Surface measurements are becoming manufacturing requirements in production for a growing number of companies," says Erik Novak, director of applications and technology for Veeco Instruments optical and stylus division, (Tucson, AZ) "It's now used more for R&D project evaluation, measuring finished products, evaluating surface treatments, and limited failure analysis. However, moving these tests into the standard production operation improves quality and yield and is a differentiator for many precision manufacturers."

As new opportunities open up, manufacturers may have to measure things they have never worked with before. For example, very small screws used to attach repair devices to a patient's bone must have surface characteristics that help promote bone growth. Another unique requirement is evaluating the moving surfaces of replacement knee and hip joints. They usually have to be quite smooth so abrasion generates no infection-causing particles.

Roughness tests usually cover an area 0.8-mm long or less while waviness requires passes in the 0.8–4-mm range. For example, if a rubber sealing ring is being investigated, the waviness check would show variations in the overall surface, such as those that might prevent a tight seal. Roughness would look for flaws in specific areas of the seal.

Research instruments range in size from tabletop units to those designed for a clean lab. They can work with part reflectivity ranging from 0.1 to 99%. Their top of the line units measure with a resolution better than 0.1 nm with an accuracy of a few nm in the vertical axis and the most advanced units can now record and track dynamic processes such as corrosion and vibration. Maximum Z-axis range is 10 mm. These instruments are applied both to one-off objects, such as space shuttle components as well as checking the samples from a high-volume production operation.

Most of industry has been content with R_a readings. But, as the user base gets more sophisticated, there is a greater need for process analysis and full surface and critical dimension characterization. Instruments are moving to the shop floor and 3-D instruments are more prevalent.

Some of Veeco's units go to service companies, but most are for in-house metrology, chiefly for production checks as opposed to analysis. Users often find that having more complete data on a process provides a competitive advantage.

A customer can buy a full microscope for a stand-alone installation or incorporate the metrology head for inline production measurements. Applications include film thickness, cutting tool geometries, and even engine block roughness and component dimensions.

The latest "white light" units have moved from tungsten or halogen illumination to LEDs for improved MTBF and data quality. New systems can now also get data from extremely sloped parts which are generally hard to image. "We are constantly improving the angle at which we can make measurements," says Novak. "It's quite simple to reflect a beam off a flat surface, but a clear image becomes more difficult as the wall angle shifts toward the perpendicular. Currently our instruments can get signals from angles of more than 60°," Novak concludes.

"Our units use focused white light that laterally scans across a part," says Craig Leising, product manager, Micro Photonics (Irvine, CA). Data are collected point-by-point at the rate of 20 mm/sec. Maximum area scanned is 6 x 6" (150 x 150 mm) with a vertical measurement range up to 27 mm.

System advantages are said to include the ability to measure any surface, including fabrics, and sloped surfaces up to an angle of 85°. "We are currently developing units with a higher scanning speed and higher throughput," says Leising.

The introduction of noncontacting surface-measuring instruments has given researchers new insight into material performance. For example, one area of concern in the auto industry was unexplained failure of a transmission under development. It took the introduction of a noncontacting surface evaluation instrument to resolve the problem.

We found that certain parts, chiefly shafts and cylinder bores, were failing even though they met all the required specifications as far as dimensions, surface treatment and hardness," explains Yucong Wang, department manager, GM (Detroit).

It was not until they closely examined the contacting faces of various parts that researchers realized that the relative moving direction of contact surfaces was a big factor.

Although the initial specifications did include an R_a number, the direction of the roughness was not given. The roughness could be very different on the same surface in different directions, which created very different tribological performance. Surface roughness could significantly affect lubrication, which could then affect parts' scuffing resistance. A common example would be a piston skirt or piston ring running against cylinder bore. (For example, with two mating gear, one side of each tooth gets a much higher load than the other side. The wear on the driving side of the gear tooth was five times greater than that on the driven tooth side. This difference creates a wear process called scuffing that can ultimately lead to gear failure.

Scuffing, according to ASM, is localized damage caused by solid-phase welding between sliding surfaces. It is characterized by material transfer between sliding surfaces that is often caused by inadequate lubrication. "This problem can be resolved by revising hardness requirements and the type of lubrication," says Wang.

Another drive-train issue is determining how much friction each contacting element should experience. The simplistic thought is that the smoother the better. That's fine for most parts in the vehicle's engine, but not necessarily for some transmission components.

"Our analysis found that some parts transmit torque more efficiently if they are not perfectly smooth," explains Wang. This involves the "grip" between the parts and the type of lubricant present. A somewhat roughened surface transmits torque more efficiently and becomes more durable when it works with lubricant.

"Our test showed that the direction of grinding when the part was first machined influenced surface morphology, so our production testing had to include specifications for measurement in a specific direction (both X and Y planes of the test-object's surface.)"

"It's therefore critical that we determine which level of friction is appropriate for each part and establish manufacturing and lubrication specifications accordingly.

"Some people take their measurements only with a profilometer using processes based on tradition or convenience. That is not adequate in many cases, particularly if you are dealing with a curved surface," Wang concludes.

Carl Zeiss Industrial Measuring Technology (Maple Grove, MN) has introduced the new Surfcom 2000 and Surfcom 5000 instruments. Both can capture surface and contour values in a single measuring run. Based on a traditional roughness sensor, Surfcom 2000 has a probing range of ±2.5 mm, while the laser-controlled detector and stylus system on Surfcom 5000 has a probing range of up to ±6.5 mm at a resolution of 0.31 nm.

The Surfcom 2000 can be reconfigured into a fully automatic CNC measuring station with CNC table modules. Automatic CNC measuring runs are possible when the unit is programmed with TIMS software.

"When establishing a test program, the operator should make a Repeatability and Reproducibility [R&R] study to help verify the accuracy of that program," explains Don Cohen, managing member of Michigan Metrology LLC (Novi, MI).

Repeatability refers to multiple test results from a single test station. High variations indicate there are problems with the instrument or the operator.

Reproducibility information comes from test results of a group of people testing a product on the same type of testing instrument. Errors here probably mean the test process has problems.

"Stability of the program is critical," says Cohen. "Making a single pass with a contact profilometer as a means of checking works well if the process is stabilized and there have been no major changes since the test program was established."

The Status of Standards

The B46 committee, which has existed since 1940, has the task of reviewing its standards every 5 or 10 years. Revising standards might seem like an oxymoron, but technology advances and our understanding of industry's needs is constantly changing. A standard written in 1950 for a type of electron microscope may no longer be valid. In fact, the instrument may no longer be widely used.

Revisions of standards are needed because high-precision measurements are becoming more essential. Accuracy specifications are getting tougher, plus there is a need for greater understanding of process failure and the effectiveness of new or revised processes.

The currently revised surface-measurement standard may be available by late 2009. It is now being reviewed by industry representatives. The major area of "discussion" is how to make the measurement of surface finish practical for industry and identifying and defining the many new measurement parameters that have been developed "beyond R_a".

An executive summary that defines key measurement processes is available at: ASME B46 Committee.

Sample Report

Sample 1 appears to be peened, possessing an average roughness, R_a, of about 2200 nm. Sample 5 appears to be microfinished, having an R_a of about 61 nm.

From a fatigue point of view, texture parameters indicate areas likely to initiate a crack may be of interest. One such parameter, the skew (R_sk), indicates the symmetry of the surface. A uniformly distributed surface (i.e. peaks and valleys) would tend to have a skew of zero (0). We find that the sample, despite having the highest R_a, has the smallest R_sk (0.09), indicating a very symmetric surface.

The lowest R_a surface, in sample 5, has the most negative _Rsk, indicating the presence of deep anomalous valleys. The graphics do demonstrate that Sample 1 consists of large-amplitude symmetric features as opposed to Sample 5, which appears with a number of deep crevices/pits.

"A report such as this might indicate that different processes that cause friction on a sample may provide a desirable R_a but, at the same time, an undesirable R_sk. For example, a lower roughness reading may result in lower friction, but early fatigue failure."

(R_sk evaluates a profile's symmetry. A negative number can indicate a predominance of valleys while a positive result reveals a predominance of peaks. It's possible for two surfaces to have the same R_a values but key features may be radically different.)

To order a hardcopy reproduction of this article, click here. To purchase digital reprints or reproduction licenses, please contact the resource center at service@sme.org or call (800) 733-4763.