[Lowell Thomas and Luke Volpe]

Editor's note: This article originally appeared in the Winter 2009 issue of MICROmanufacturing magazine.

As the drive to build smaller and more complex electromechanical devices (microdevices) continues, design engineers struggle to bridge the gap between traditional micromachining techniques, such as milling, laser cutting and EDMing, and MEMS (microelectromechanical systems) technology.

Specifically, design engineers are discovering that many products now in the design stage are too small to be machined with traditional techniques and are not candidates for MEMS technology because of size, cost or material limitations.


Four-layer micro induction coil with 12.5µm lines and spaces, 7µm-thick gold traces and 10µm-thick polyimide dielectric between traces. All images courtesy Metrigraphics.

A combination of two mature technologies—electrochemical metal deposition and micron-level photolithography—have enabled the manufacture of 3-D microstructures that bridge the gap between traditional machining techniques and MEMS technology.

The term “electroforming,” as used in this article, refers to the electrochemical deposition of a variety of conductive metals, including gold, copper, nickel and nickel/cobalt, into a prefabricated mold, usually an ultraviolet-imagable photoresist. The technology forms freestanding or linked multilevel microstructures, sometimes referred to as 3-D micros, with microsized features.

With its ability to create ultrafine features, electroforming has replaced EDMing and conventional machining in some applications. However, the strength of this new technology is not in replacing other techniques but in enabling new part designs.

This article discusses electroforming technology and limitations, specific application areas and development trends.

Technology’s foundation

As stated, electroforming creates 3-D structures by electrochemically depositing a metal into a photoresist mold. Basic process steps are as follows (see illustration below):

1. A carrier plate is prepared by sputter depositing a thin (less than 5,000 angstroms) adhesive/conductive seed metal layer onto a glass blank, or other suitable base material.
2. A photoresist mold of the intended structure is created. The mold is formed by depositing and imaging the X- and Y-plane features of the intended structure into a UV-sensitive photoresist on the seed metal coated glass carrier.
3. The desired metal is electrochemically deposited into the photoresist mold prepared in step 2 above.
4. The photoresist mold is removed from the glass carrier plate.
5. Finally, the completed electroformed microstructure is removed from the carrier plate.

Multilayer structures are formed by repeating the above steps. This process is capable of creating 3-D microstructures with features—posts, channels and grooves—as small as 0.005mm. Funnel-shaped through-holes as small as 0.001mm in diameter are possible.

The X- and Y-plane dimensions are controlled by the photoresist process. The Z-axis dimensions, such as post and wall heights, channel depths and wall profiles, are controlled and restricted by the photoresist aspect ratio. The aspect ratio is equal to the maximum height (Z dimension) divided by the minimum feature width (X or Y dimension).

For example, a 0.010mm-dia. × 0.030mm-high post would have an aspect ratio of 3:1. Maximum X and Y dimensions can be 25mm, and generally are less than 1mm or 2mm.

The current generation of 3-D microstructures made by microforming fall into the dimensional realm described above and are used for a variety of applications, including medical devices, magnetic induction coils for RF coupling, data transfer, microfluidic molds and integrated circuit manufacturing.

To date, electroforming has been successfully implemented in many applications, enabling the development of diverse structures not possible with other micromanufacturing technologies. (See Figures 1, 2 and 3 for examples of electroformed devices and features.)


Figure 1: RF coupling device.


Figure 2: Funnel-shaped orifice hole.

Figure 3: Monofilament/microlens holder.

However, current electroforming technology faces challenges in terms of minimum structure size, maximum structural aspect ratio, number of possible layers, flexibility issues related to multilayer devices and manufacturing cost. Future applications could also require dielectric coatings or interlayers not currently available. (Note: Dielectric isolation has always been required for 3-D microstructures used in electrical applications. Current challenges in applying dielectric coatings on microstructures include coating uniformity and the inability to coat the structure’s carrier side. As structures become smaller, it will be essential to uniformly overcoat the entire 3-D structure.)

Photoresist

Several photoresist materials exist that are capable of resolving 0.010mm structural features with aspect ratios of 3:1 (5:1 aspect ratio structures are possible with some restrictions). Photoresist materials that allow higher aspect ratios for smaller feature sizes could improve the functionality of existing devices and enable new applications.

Future device-design engineers will be looking to build structures with minimum feature sizes in the 0.002mm to 0.005mm range with 5:1 to 10:1 aspect ratios.

Once the electroformed metal deposition has been completed, the photoresist mold must be chemically removed, a process known as stripping, or dissolving. Removability has a significant effect on minimum aperture, or blind-hole size, as well as on isolated cavities in multilevel structures. As blind-holes and isolated cavities become smaller, the ability of solvent chemistry to penetrate holes and cavities becomes an issue.

Some photoresists are available that can resolve higher aspect ratios, but, to date, these materials are limited by solubility characteristics, or stripping issues. However, most manufacturers of photoresists are developing solutions to these challenges.

Imaging method

The imaging method (UV) also has a significant effect on aspect ratio and feature characteristics, such as wall angle and uniformity. There are two traditional imaging methods.

The predominant method is contact printing. With this approach, a photographic mask containing multiple, reverse-polarity images of the structure to be built is placed in intimate contact with the photoresist-coated carrier plate. The carrier plate/photoresist/mask stack is exposed to a collimated UV light source. Depending on the type of resist used, the UV-exposed photoresist will either be cross linked or made soluble so the retained photoresist will form the desired mold cavity.

This system is economical, relatively easy to use and can create a full sheet of hundreds—possibly thousands—of mold cavities with a single exposure. However, UV intensity variation over the mask area and defects caused by mask contact with the photoresist can reduce product yield.

The second method is projection-imaging stepping (repeating). Devices made with this method are used extensively in the integrated-circuit and flat-panel display industries. These systems use a single-structure image mask at some magnification that is projected through a reduction lens onto and into the photoresist plane (base carrier plate coated with photoresist).

The UV source is positioned behind the single-image mask with a shutter mechanism between them. Every time the shutter opens, a single unit structure is exposed. When the shutter is closed, an interferometer-controlled, movable stage repositions the base carrier plate.

The advantages of this system are that higher-resolution, concentrated UV power permits deeper exposure, resulting in a higher aspect-ratio capability and greater image-to-image consistency for UV exposures. The system is also suitable for high-volume manufacturing. Its main disadvantage is that it requires a major capital investment. State-of-the-art projection steppers cost $1 million and up, but some used equipment may be available now, or in the future, at significantly lower cost.

Any discussion relating to high-aspect-ratio imaging systems would be incomplete without mention of LIGA, a process named for the German acronym for lithographie, galvanoformung and abformung (lithography, electroplating and molding). LIGA is based on synchrotron, hard X-ray exposure of polymethyl methacrylate (PMMA). Although this system can resolve extremely high-aspect-ratio images (50:1 and greater), there are a limited number of synchrotron beam lines available and these are extremely costly to operate. (Synchrotron radiation is electromagnetic radiation accelerated to speeds approaching that of light.)

Such systems cost hundreds of millions of dollars, depending on the total size and number of beam lines. There are only four synchrotrons in the U.S., and they are all associated with large institutions, such as national laboratories and large universities.

On a more practical level, continuing laboratory testing of newly developed photoresist materials will enable higher aspect ratios using more conventional exposure and developing techniques.

Multilevel stacking

Based on recent trends, future 3-D microstructures will become more complex, requiring multilayer structures. The current multilevel stack limit is between three and five levels, depending on the criticality of individual stack thickness and flatness specifications, minimum feature size and aspect ratio. (See Figures 4 and 5 for examples of multilayer structures.)


Figure 4: Two-level, magnetically activated switching device. The cross-section of the three serpentine springs is 0.050mm × 0.050 mm.


Figure 5: Positive-locking, two-level contact interconnect.

The critical issues in the stacking process are previous-layer uniformity, photoresist coating uniformity and precision of mask alignment to the previously electroformed layer.

Plating uniformity is a key part of this process. Plating uniformity is a function of the anode and cathode positional relationship (in this system, the cathode is the sheet of 3-D structures being built) and current density uniformity.

The geometry of the structure being plated determines current density uniformity. Practically all 3-D microstructures have varying geometric (X- and Y-plane) shapes and sizes. This X- and Y-plane variation is unique to each particular microstructure pattern.

Unfortunately, one of the significant causes of nonuniform plating is pattern geometry variation. All microstructure-plating layers and thicknesses have some nonuniformity, the overriding multiple-level structure limitation.

Nonuniform electro-deposited thicknesses cause imprecise alignment, and nonuniform plating in subsequent layers exaggerates the problem. As the number of required structural layers increases, structural geometries become distorted and overall thickness tolerances cannot be maintained.

A planarizing technology called chemical mechanical planarization (CMP) combines chemical etching and abrasive polishing. It allows dissimilar materials, such as an electrochemically deposited metal and photoresist, to be planarized, or polished, to microflatness. This planarization process addresses the multilevel limiting issues stated above. CMP is a mature technology but is used sparingly in multilevel microstructure manufacturing because of cost.

As the demand for higher volume and more complex multilevel devices increases, chemical mechanical planarization will be required. In-house CMP capability may become a necessity for companies involved in 3-D microstructure manufacturing.

Material variation

The most common electrochemically deposited metals are gold, nickel, nickel/cobalt and copper. Some combinations of these metals work well, such as copper and gold or nickel/cobalt and gold.

Overplating the entire 3-D structure is problematic but possible, depending on the overcoat material. Gold over copper or nickel is possible. Platinum, palladium and rhodium are desirable overplating materials but can present plating issues related to the specific plating bath.

Magnetic metal deposition baths are available and may be considered for those applications requiring magnetic properties. However, complex and exotic solutions like this make the plating process more difficult.

Future 3-D microstructure applications will likely require at least one of the above metal or coating options. For example, pure gold overplatings may be essential on copper-based RF coupling devices where biocompatibility is an issue. Similarly, dielectric overplating will be essential in applications where controlled dielectric characteristics are critical. (See Figure 6 below for an example of a device that has been overplated.)


Figure 6: Two-level solid gold implantable capillary device.

The enabling technologies mentioned previously are expected to make electroforming an even more commercially viable process. For example, projection imaging will enable the creation of electroformed features with resolutions of less than 1µm and aspect ratios of 10:1 or higher. Chemical mechanical polishing will support smaller multilayer structures and enable precision alignments of less than 1µm.

Electroforming has already had a positive impact in several critical applications, such as semi-intrusive and permanent medical implants, telemetry and telecommunication devices and military devices. As the technology is developed further, it will enhance the use of 3-D microstructures in existing applications and enable penetration into new applications. µ
------------------

About the authors: Lowell Thomas is senior staff engineer and Luke Volpe is director of engineering (retired) for Dynamics Research Corp., Metrigraphics Division, Wilmington, Mass. Telephone: (978) 658-6100. Web site: www.drc.com/metrigraphics.
-------------------

The quest for smaller microelectronic devices

One of the critical issues in the quest for smaller, more complex medical and telecommunication devices is the integrated circuit chip interconnect system. Because of size, biocompatibility and flexibility requirements, traditional flex circuit and PC board technologies are typically not suitable for these devices.

The current generation of medically implantable telemetric and telecommunication microdevices are based on extreme-resolution, microflexible-interconnect (ERMF) circuits. This technology depends on microphotolithography, thin-film sputter-deposited base metals, electrochemically deposited metal (usually noble) traces, biocompatible polyimide substrates and dielectric interlayers.

Minimum signal trace and space dimensions are 3µm to 5µm. For applications requiring significant current density, high-aspect ratio trace cross sections allow higher current-carrying capacity and minimize overall circuit area.

ERMF circuits may have a single conductor level or as many as six conductor levels, depending on device complexity.

To date, ERMF circuits have been successfully used in several intrusive and implantable medical applications, such as ultrasonic angioplasty probes, blood glucose monitoring systems and developmental retinal implants.

Other, more complex ERMF circuit-based medical devices are being developed, such as permanently implanted RF coupling (data transfer) monitoring systems and more complex cardiac monitoring probes.

The success of these and other still unidentified medical and telecommunication applications may depend on yet unresolved or identified ERMF manufacturing issues. These issues relate to materials and manufacturing technology.

Liquid-state polyimide is generally considered the material of choice for ERMF circuits. This material, when cast and completely cured, typically has adequate dielectric strength in thicknesses above 0.015mm. As the circuits become more complex and require more trace layers, the flexibility of the stack can become an issue. One possible solution includes improving the intrinsic dielectric properties of the cured polyimide, allowing for thinner dielectric interlayers. Thinner dielectric layers will improve the flexibility of the cured material.

One method of manufacturing utilizes UV photo-imagable polyimide to create via holes. This process can create an entire sheet of hundreds, possibly thousands, of holes with a one-photo process. Using this process, however, imaged via hole wall angles may be difficult to control and the process can produce random, poor or nonexistent continuity interlayer interconnects.

Current ERMF manufacturing processes specify laser drilling to create via holes. This process creates clean, slag-free holes as small as 0.015mm in diameter with cured and planarized deposited polyimide predrilling. Laser-drilled holes also have positively tapered wall angles that ensure complete metal coverage and reliable continuity. However, this process may not be cost effective without semi-automated or fully automated equipment.

Investments in manufacturing development are being evaluated that could help ERMF technology reach its full potential. The following issues are currently under evaluation:

• The standard 6-sq.-in. panel size restricts high-volume production; 12-sq.-in. or round panels will be required.
• Chemical mechanical planarization will be required to guarantee flatness and planarity on each layer of multilayer structures. This process is only suitable with laser via drilling.
• I/O bonding pads must accommodate processes such as pick-and-place assembly and flip chip/solder reflow.
• The circuit-manufacturing process must be automated to reduce the cost per circuit for high-volume manufacturing operations.

— L. Thomas and L. Volpe