Millimeter-Scale, Micro-Electro-

Mechanical Systems Gas Turbine

Engines
The confluence of market demand for greatly improved compact power sources for por-
table electronics with the rapidly expanding capability of micromachining technology has
made feasible the development of gas turbines in the millimeter-size range. With airfoil

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

Alan H. Epstein spans measured in 100’s of microns rather than meters, these ‘‘microengines’’ have about
Gas Turbine Laboratory 1 millionth the air flow of large gas turbines and thus should produce about one millionth
Massachusetts Institute of Technology the power, 10–100 W. Based on semiconductor industry-derived processing of materials
Cambridge, MA 01239 such as silicon and silicon carbide to submicron accuracy, such devices are known as
e-mail: Epstein@[Link] micro-electro-mechanical systems (MEMS). Current millimeter-scale designs use centrifu-
gal turbomachinery with pressure ratios in the range of 2:1 to 4:1 and turbine inlet
temperatures of 1200–1600 K. The projected performance of these engines are on a par
with gas turbines of the 1940s. The thermodynamics of MEMS gas turbines are the same
as those for large engines but the mechanics differ due to scaling considerations and
manufacturing constraints. The principal challenge is to arrive at a design which meets
the thermodynamic and component functional requirements while staying within the realm
of realizable micromachining technology. This paper reviews the state of the art of
millimeter-size gas turbine engines, including system design and integration, manufactur-
ing, materials, component design, accessories, applications, and economics. It discusses
the underlying technical issues, reviews current design approaches, and discusses future
development and applications. 关DOI: 10.1115/1.1739245兴

Introduction tion cost in large-scale production. Such assemblies are known in

the U.S. as micro-electrical-mechanical systems 共MEMS兲 and
For most of the 60-year-plus history of the gas turbine, eco-
have been the subject of thousands of publications over the last
nomic forces have directed the industry toward ever larger en-
two decades, 关3兴. In Japan and Europe, devices of this type are
gines, currently exceeding 100,000 lbs of thrust for aircraft pro-
known as ‘‘microsystems,’’ a term which may encompass a wider
pulsion and 400 MW for electric power production applications. variety of fabrication approaches. Early work in MEMS focused
In the 1990s, interest in smaller-size engines increased, in the few on sensors and simple actuators, and many devices based on this
hundred pound thrust range for small aircraft and missiles and in technology are in large-scale production, such as pressure trans-
the 20–250 kW size for distributed power production 共popularly ducers and airbag accelerometers for automobiles. More recently,
known as ‘‘microturbines’’兲. More recently, interest has developed fluid handling is receiving attention. For example, MEMS valves
in even smaller size machines, 1–10 kW, several of which are are commercially available, and there are many emerging bio-
marketed commercially, 关1,2兴. Gas turbines below a few hundred medical diagnostic applications. Also, chemical engineers are pur-
kilowatts in size generally use centrifugal turbomachinery 共often suing MEMS chemical reactors 共chemical plants兲 on a chip, 关4兴.
derivative of automotive turbocharger technology in the smaller User pull is predominantly one of electric power. The prolifera-
sizes兲, but are otherwise very similar to their larger brethren in tion of small, portable electronics—computers, digital assistants,
that they are fabricated in much the same way 共cast, forged, ma- cell phones, GPS receivers, etc.—requires compact energy sup-
chined, and assembled兲 from the same materials 共steel, titanium, plies. Increasingly, these electronics demand energy supplies
nickel superalloys兲. Recently, manufacturing technologies devel- whose energy and power density exceed that of the best batteries
oped by the semiconductor industry have opened a new and very available today. Also, the continuing advance in microelectronics
different design space for gas turbine engines—one that enables permits the shrinking of electronic subsystems of mobile devices
gas turbines with diameters of millimeters rather than meters, with such as ground robots and air vehicles. These small, and in some
airfoil dimensions in microns rather than millimeters. These shirt- cases very small, mobile systems require increasingly compact
button-sized gas turbine engines are the focus of this review. power and propulsion. Hydrocarbon fuels burned in air have
Interest in millimeter-scale gas turbines is fueled by both a 20–30 times the energy density of the best current lithium
technology push and a user pull. The technology push is the de- chemistry-based batteries, so that fuelled systems need only be
velopment of micromachining capability based on semiconductor modestly efficient to compete well with batteries.
manufacturing techniques. This enables the fabrication of com- Given the need for high power density energy conversion in
plex small parts and assemblies—devices with dimensions in the very small packages, a millimeter-scale gas turbine is an obvious
1–10,000 ␮m size range with submicron precision. Such parts are candidate. The air flow through gas turbines of this size is about
produced with photolithographically defined features and many six orders of magnitude smaller than that of the largest engines
can be made simultaneously, offering the promise of low produc- and thus they should produce about a million times less power,
10–100 watts with equivalent cycles. Work first started on MEMS
Contributed by the International Gas Turbine Institute 共IGTI兲 of THE AMERICAN approaches in the mid 1990s, 关5–7兴. Researchers rapidly discov-
SOCIETY OF MECHANICAL ENGINEERS for publication in the ASME JOURNAL OF ered that gas turbines at these small sizes have no fewer engineer-
ENGINEERING FOR GAS TURBINES AND POWER. Paper presented at the Interna-
tional Gas Turbine and Aeroengine Congress and Exhibition, Atlanta, GA, June
ing challenges than do conventional machines and that many of
16 –19; 2003, Paper No. 2003-GT-38866. Manuscript received by IGTI, October the solutions evolved over six decades of technology development
2002, final revision, March 2003. Associate Editor: H. R. Simmons. do not apply in the new design space. This paper reviews work on

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 205
Copyright © 2004 by ASME
MEMS gas turbine engines for propulsion and power production.
It begins with a short discussion of scaling and preliminary design
considerations, and then presents a concise overview of relevant
MEMS manufacturing techniques. In more depth, it examines the
microscale implications for cycle analysis, aerodynamic and struc-
tural design, materials, bearings and rotor dynamics, combustion,
and controls and accessories. The gas turbine engine as a system
is then considered. This review then discusses propulsion and
power applications and briefly looks at derivative technologies
such as combined cycles, cogeneration, turbopumps, and rocket
engines. The paper concludes with thoughts on future develop-
ments.

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

Thermodynamic and Scaling Considerations
Thermal power systems encompass a multitude of technical dis- Fig. 1 Simple cycle gas turbine performance with H2 fuel
ciplines. The architecture of the overall system is determined by
thermodynamics while the design of the system’s components is
influenced by fluid and structural mechanics and by material, elec- laxed turbomachinery performance requirements, but it introduces
trical and fabrication concerns. The physical constraints on the additional design and fabrication complexity. Thus, the first de-
design of the mechanical and electrical components are often dif- signs are simple cycle gas turbines.
ferent at microscale than at more familiar sizes so that the optimal How big should a ‘‘micro’’ engine be? A micron, a millimeter, a
component and system designs are different as well. Conceptually, centimeter? Determination of the optimal size for such a device
any of the thermodynamic systems in use today could be realized involves considerations of application requirements, fluid me-
at microscale. Brayton 共air兲 cycle and the Rankine 共vapor兲 cycle chanics and combustion, manufacturing constraints, and econom-
machines are steady flow devices while the Otto, 关8兴, Diesel, and ics. The requirements for many power production applications fa-
Stirling cycles are unsteady engines. The Brayton power cycle vor a larger engine size, 50–100 W. Viscous effects in the fluid
共gas turbine兲 is superior based on considerations of power density, and combustor residence time requirements also favor larger en-
simplicity of fabrication, ease of initial demonstration, ultimate gine size. Current semiconductor manufacturing technology
efficiency, and thermal anisotropy. places both upper and lower limits on engine size. The upper size
A conventional, macroscopic gas turbine generator consists of a limit is set mainly by etching depth capability, a few hundred
compressor, a combustion chamber, and a turbine driven by the microns at this time. The lower limit is set by feature resolution
combustion exhaust that powers the compressor. The residual en- and aspect ratio. Economic concerns include manufacturing yield
thalpy in the exhaust stream provides thrust or can power an elec- and cost. A wafer of fixed size 共say, 200 mm diameter兲 would
tric generator. A macroscale gas turbine with a meter-diameter air yield many more low power engines than high power engines at
intake area generates power on the order of 100 MW. Thus, tens essentially the same manufacturing cost per wafer. 共Note that the
of watts would be produced when such a device is scaled to mil- sum of the power produced by all of the engines on the wafer
limeter size if the power per unit of air flow is maintained. When would remain constant at 1–10 kW.兲 When commercialized, ap-
based on rotating machinery, such power density requires com- plications and market forces may establish a strong preference
bustor exit temperatures of 1200–1600 K; rotor peripheral speeds here. For the first demonstrations of a concept, a minimum tech-
of 300– 600 m/s and thus rotating structures centrifugally stressed nical risk approach is attractive. Analysis suggested that fluid me-
to several hundred MPa since the power density of both turboma- chanics would be difficult at smaller scales, so the largest size
chinery and electrical machines scale with the square of the speed, near the edge of current microfabrication technology, about a cen-
as does the rotor material centrifugal stress; low friction bearings; timeter in diameter, was chosen as a focus of MIT’s efforts.
tight geometric tolerances and clearances between rotating and Performance calculations indicate that the power per unit air
static parts to inhibit fluid leakage 共the clearances in large engines flow from the configuration discussed below is 50–150 W/共g/sec兲
are maintained at about one part in 2000 of the diameter兲; and of air flow 共Fig. 1兲. For a given rotor radius, the air flow rate is
thermal isolation of the hot and cold sections. limited primarily by airfoil span as set by stress in the turbine
These thermodynamic considerations are no different at micro blade roots. Calculations suggest that it might be possible to im-
than at macroscale. But the physics and mechanics influencing the prove the specific work, fuel consumption, and air flow rate in
design of the components do change with scale, so that the opti- later designs with recuperators to realize microengines with power
mal detailed designs can be quite different. Examples of these outputs of as much as 50–100 W, power specific fuel consumption
differences include the viscous forces in the fluid 共larger at mi- of 0.3–0.4 g/w-hr, and thrust-to-weight ratios of 100:1. This level
croscale兲, usable strength of materials 共larger at microscale兲, sur- of specific fuel consumption approaches that of current small gas
face area-to-volume ratios 共larger at microscale兲, chemical reac- turbine engines but the thrust-to-weight ratio is 5–10 times better
tion times 共invariant兲, realizable electric field strength 共higher at than that of the best aircraft engine. The extremely high thrust-to-
microscale兲, and manufacturing constraints 共limited mainly to weight ratio is simply a result of the so-called ‘‘cube-square law.’’
two-dimensional, planar geometries given current microfabrica- All else being the same as the engine is scaled down linearly, the
tion technology兲. air flow and thus the power decreases with the intake area 共the
There are many thermodynamic and architectural design square of the linear size兲 while the weight decreases with the
choices in a device as complex as a gas turbine engine. These volume of the engine 共the cube of the linear size兲, so that the
involve tradeoffs among fabrication difficulty, structural design, power-to-weight ratio increases linearly as the engine size is re-
heat transfer, and fluid mechanics. Given a primary goal of dem- duced. Detailed calculations show that the actual scaling is not
onstrating that a high power density MEMS heat engine is physi- quite this dramatic since the specific power is lower at the very
cally realizable, MIT’s research effort adopted the design philoso- small sizes, 关5兴. A principal point is that a micro-heat engine is a
phy that the first engine should be as simple as possible, with different device than more familiar full-sized engines, with differ-
performance traded for simplicity. For example, a recuperated ent weaknesses and different strengths.
cycle, which requires the addition of a heat exchanger transferring
heat from the turbine exhaust to the compressor discharge fluid, Mechanics Scaling. While the thermodynamics are invariant
offers many benefits including reduced fuel consumption and re- down to this scale, the mechanics are not. The fluid mechanics, for

206 Õ Vol. 126, APRIL 2004 Transactions of the ASME

example, are scale-dependent, 关9兴. One aspect is that viscous
forces are more important at small scale. Pressure ratios of 2:1 to
4:1 per stage imply turbomachinery tip Mach numbers that are in
the high subsonic or supersonic range. Airfoil chords on the order
of a millimeter imply that a device with room temperature inflow,
such as a compressor, will operate at Reynolds numbers in the
tens of thousands. With higher gas temperatures, turbines of simi-
lar size will operate at a Reynolds number of a few thousand.
These are small values compared to the 105 – 106 range of large-
scale turbomachinery and viscous losses will be concomitantly
larger. But viscous losses make up only about a third of the total
fluid loss in a high speed turbomachine 共three-dimensional, tip
leakage, and shock wave losses account for most of the rest兲 so

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

that the decrease in machine efficiency with size is not so dra-
matic. The increased viscous forces also mean that fluid drag in
small gaps and on rotating disks will be relatively higher. Unless
gas flow passages are smaller than one micron, the fluid behavior
can be represented as continuum flow so that molecular kinetics,
Knudsen number considerations, are not important.
Heat transfer is another aspect of fluid mechanics in which Fig. 2 Critical temperature change to cause fracture via ther-
microdevices operate in a different design space than large-scale mal shock
machines. The fluid temperatures and velocities are the same but
the viscous forces are larger, so the fluid film heat transfer coef-
ficients are higher by a factor of about three. Not only is there precision, microfabrication technology applies mainly to silicon.
more heat transfer to or from the structure but thermal conduc- Since Si rapidly loses strength above 950 K, this becomes an
tance within the structure is higher due to the short length scale. upper limit to the turbine rotor temperature. But 950 K is too low
Thus, temperature gradients within the structure are reduced. This a combustor exit temperature to close the engine cycle 共i.e., pro-
is helpful in reducing thermal stress but makes thermal isolation duce net power兲 with the component efficiencies available, so
challenging. cooling is required for Si turbines. The simplest way to cool the
For structural mechanics, it is the change in material properties turbine in a millimeter-sized machine is to eliminate the shaft, and
with length scale that is most important. Very small length scale thus conduct the turbine heat to the compressor, rejecting the heat
influences both material properties and material selection. In en- to the compressor fluid. This has the great advantage of simplicity
gines a few millimeters in diameter, design features such as blade and the great disadvantage of lowering the pressure ratio of the
tips, fillets, orifices, seals, etc. may be only a few microns in size. now non-adiabatic compressor from about 4:1 to 2:1 with a con-
Here, differences between mechanical design and material prop- comitant decrease in cycle power output and efficiency. Hydrogen
erties begin to blur. The scale is not so small 共atomic lattice or was chosen as the first fuel to simplify the combustor develop-
dislocation core size兲 that continuum mechanics no longer applies. ment. This expedient arrangement was referred to as the H2 demo
Thus, elastic, plastic, heat conduction, creep, and oxidation behav- engine. It is a gas generator/turbojet designed with the objective
iors do not change, but fracture strength can differ. Material se- of demonstrating the concept of a MEMS gas turbine. It does not
lection is influenced both by mechanical requirements and by fab- contain electrical machinery or controls, all of which are external.
rication constraints. For example, structure ceramics such as The MIT H2 demo engine design is shown in Fig. 3. The cen-
silicon carbide 共SiC兲 and silicon nitride (Si3 N4 ) have long been trifugal compressor and radial turbine rotor diameters are 8 mm
recognized as attractive candidates for gas turbine components and 6 mm, respectively. The compressor discharge air wraps
due to their high strength, low density, and good oxidation resis- around the outside of the combustor to cool the combustor walls,
tance. Their use has been limited, however, by the lack of tech- capturing the waste heat and so increasing the combustor effi-
nology to manufacture flaw-free material in sizes large enough for ciency while reducing the external package temperature. The rotor
conventional engines. Shrinking engine size by three orders of radial loads are supported on a journal bearing on the periphery of
magnitude virtually eliminates this problem. Indeed, mass- the compressor. Thrust bearings on the centerline and a thrust
produced, single-crystal semiconductor materials are essentially balance piston behind the compressor disk support the axial loads.
perfect down to the atomic level so that their usable strength is an The balance piston is the air source for the hydrostatic journal
order of magnitude better than conventional metals. This higher bearing pressurization. The thrust bearings and balance piston are
strength can be used to realize lighter structures, higher rotation supplied from external air sources. The design peripheral speed of
speeds 共and thus higher power densities兲 at constant geometry, or the compressor is 500 m/s so that the rotation rate is 1.2 Mrpm.
simplified geometry 共and thus manufacturing兲 at constant periph- External air is used to start the machine. With 400 ␮m span air-
eral speed. An additional material consideration is that thermal foils, the unit is sized to pump about 0.36 grams/sec of air, pro-
shock susceptibility decreases as part size shrinks. Thus, materials
such as alumina (Al2 O3 ) which have very high temperature capa-
bilities but are not considered high temperature structural ceram-
ics due to their susceptibility to thermal shock are viable at milli-
meter length scales 共Fig. 2兲. Since these have not been considered
as MEMS materials in the past, there is currently little suitable
manufacturing technology available, 关10兴.

Overview of a MEMS Gas Turbine Engine Design

Efforts at MIT were initially directed at showing that a MEMS-
based gas turbine is indeed possible, by demonstrating benchtop
operation of such a device. This implies that, for a first demon-
stration, it would be expedient to trade engine performance for Fig. 3 H2 demo engine with conduction-cooled turbine con-
simplicity, especially fabrication simplicity. Most current, high structed from six silicon wafers

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 207
 Fig. 4 Cutaway H2 demo gas turbine chip

ducing 0.1 Newtons of thrust or 17 watts of shaft power. A cut-

away engine chip is shown in Fig. 4. In this particular engine

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

build, the airfoil span is 225 ␮m and the disks are 300 ␮m thick.
The following sections elaborate on the component technolo-
gies of this engine design. It starts with a primer on microfabri-
cation and then goes on to turbomachinery aerodynamic design,
structures and materials, combustion, bearings and rotor dynam-
ics, and controls and accessories. A system integration discussion
then expands on the high-level tradeoffs which define the design
space of a MEMS gas turbine engine.

A Primer on Micromachining
Gas turbine engine design has always been constrained by the
practicality of manufacturing parts in the desired shape and size
and with the material properties needed. As with conventional
metal fabrication, the mechanical and electrical properties of
MEMS materials can be strongly influenced by the fabrication
process.
While an old-school designer may have admonished his team
‘‘Don’t let the manufacturing people tell you what you can’t do!’’,
design for manufacturing is now an important concern in industry. Fig. 5 Si wafer of radial inflow turbine stages
Major decisions in engine architecture are often set by manufac-
turing constraints. Of course this was true in the design of Whit-
tle’s first jet engine, in which the prominent external, reverse flow
combustors reflected the need to keep the turbine very close to the by direct optical or e-beam writing. The photoresist is then chemi-
compressor to control rotor dynamics given that the forging tech- cally developed as though it were photographic film, baked, and
nology of the day could only produce short, small diameter shafts then the exposed areas are removed with a solvent. This leaves
integral with a disk, 关11兴. bare silicon in the areas to be etched and photoresist-protected
Compared to manufacturing technologies familiar at large silicon elsewhere. The etching process is based on the principle
scale, current microfabrication technology is quite constrained in that the bare silicon is etched at a much higher rate, typically
the geometries that can be produced and this severely limits en- 50–100⫻, than the mask material. Many different options for
gine design options. Indeed, the principal challenge is to arrive at making masks have been developed, including a wide variety of
a design which meets the thermodynamic and component func- photoresists and various oxide or metal films. By using several
tional requirements while staying within the realm of realizable layers of masking material, each sensitive to different solvents,
micromachining technology. The following paragraphs present a multi-depth structures can be defined. Photoresist on top of SiO2
simple overview of current micromachining technology important is one example.
to this application and then discuss how it influences the design of The exposed areas of the wafer can now be etched, either
very small rotating machinery. These technologies were derived chemically or with a plasma. The devices we are concerned with
from the semiconductor industry 15–20 years ago, but the busi- here require structures which are 100’s of microns high with very
ness of micromachining has now progressed to the level that con- steep walls, thus a current technology of great interest is deep
siderable process equipment 共known as ‘‘tools’’兲 is developed spe- reactive ion etching 共DRIE兲. In the DRIE machine, the wafer is
cifically for these purposes, 关12兴. etched by plasma-assisted fluorine chemistry for several tens of
The primary fabrication processes important in this application seconds, then the gas composition is changed and a micron or so
are etching of photolithographically defined planar geometries and of a teflon-like polymer is deposited which preferentially protects
bonding of multiple wafers. The usual starting point is a flat wafer the vertical surfaces, and then the etch cycle is repeated, 关13兴. The
of the base material, most often single-crystal silicon. These wa- average depth of a feature is a function of the etch time and the
fers are typically 0.5 to 1.0 mm thick and 100 to 300 mm in local geometry. The etch anisotropy 共steepness of the walls兲 can
diameter, the larger size representing the most modern technology. be changed by adjusting the plasma properties, gas composition,
Since a single device of interest here is typically a centimeter or and pressure. In addition, these adjustments may alter the unifor-
two square, dozens to hundreds fit on a single wafer 共Fig. 5兲. mity of the etch rate across the wafer by a few percent since no
Ideally, the processing of all the devices on a wafer is carried out machine is perfect. One feature of current tools is that the etch
in parallel, leading to one of the great advantages of this micro- rate is a function of local geometry such as the lateral extent of a
machining approach, low unit cost. To greatly simplify a complex feature. This means that, for example, different width trenches
process with very many options, the devices of interest will serve etch at different rates, presenting a challenge to the designer of a
as illustrative examples. complex planar structure. A DRIE tool typically etches silicon at
First, the wafers are coated with a light-sensitive photoresist. A an average rate of 1–3 ␮m per minute, the precise rate being
high contrast black-and-white pattern defining the geometry is feature and depth-dependent. Thus, structures that are many hun-
then optically transferred to the resist either by means of a contact dreds of microns deep require many hours of etching. Such a tool
exposure with a glass plate containing the pattern 共a ‘‘mask’’兲, or currently costs $0.5–1.0 M and etches one wafer at a time, so the

208 Õ Vol. 126, APRIL 2004 Transactions of the ASME

 ure, they are actually slightly tapered from hub to tip. Current
technology can yield a taper uniformity of about 30:1 to 50:1 with
either a positive or negative slope. At the current state of the art,
the airfoil length can be controlled to better than 1 ␮m across the
disk, which is sufficient to achieve high-speed operation without
the need for dynamic balancing. Turbomachines of similar geom-
etry have been produced with blade spans of over 400 ␮m.
The processing of a 4 mm diameter turbine stage is illustrated
in Fig. 7 as a somewhat simplified example. Note that the vertical
scaling in the figure is vastly exaggerated for clarity since the
thickness of the layers varies so much 共about 1 ␮m of silicon
dioxide and 10 ␮m of photoresist on 450 ␮m of silicon兲. It is a
16-step process for wafer 1, requiring two photo masks. It in-

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

cludes multiple steps of oxide growth 共to protect the surface for
wafer bonding兲, patterning, wet etching 共with a buffered hydro-
fluoric acid solution known as BOE兲, deep reactive ion etching
共DRIE兲, and wafer bonding 共of the rotor wafer, #1, to an adjoining
Fig. 6 A 4:1 pressure ratio, 4 mm rotor dia radial inflow turbine wafer, #2, to prevent the rotor from falling out during processing兲.
stage Note that wafer 2 in the figure was previously processed since it
contains additional thrust bearing and plumbing features which
are not shown here for clarity, In fact, it is more complex to
etching operation is a dominant factor in the cost of producing fabricate than the rotor wafer illustrated.
such deep mechanical structures. Both sides of a wafer may be The second basic fabrication technology of interest here is the
etched sequentially. bonding together of processed wafers in precision alignment so as
Figure 6 is an image of a 4 mm rotor diameter, radial inflow air to form multilayer structures. There are several classes of wafer
turbine designed to produce 60 watts of mechanical power at a tip bonding technologies. One uses an intermediate bonding layer
speed of 500 m/s, 关14,15兴. The airfoil span is 200 ␮m. The cylin- such as a gold eutectic or SiO2 . These approaches, however, re-
drical structure in the center is a thrust pad for an axial thrust air sult in structures which have limited temperature capabilities, a
bearing. The circumferential gap between the rotor and stator few hundred °C. It is also possible to directly bond silicon to
blades is a 15 ␮m wide air journal bearing required to support the silicon and realize the material’s intrinsic strength through the
radial loads. The trailing edge of the rotor blades is 25 ␮m thick entire usable temperature range of the material, 关16,17兴. Direct
共uniform to within 0.5 ␮m兲 and the blade roots have 10 ␮m radius bonding requires very smooth 共better than 10 nanometers兲 and
fillets for stress relief. While the airfoils appear planar in the fig- very clean surfaces 共a single 1 ␮m diameter particle can keep

Fig. 7 Simplified processing steps to produce the turbine in Fig. 6 in a wafer stack. „Figure courtesy of N. Miki.…

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 209
 for bonding. A fundamentally different approach is to arrange a
sequence of fabrication steps with all processing done at the wafer
level so that a freely rotating captured rotor is the end product.
The process must be such that the rotor is not free at any time
during which it can fall out, i.e., it must be mechanically con-
strained at all times. There are several ways that this can be ac-
complished. For example, the layer containing the rotor can be
‘‘glued’’ to adjoining wafers with an oxide during fabrication.
This glue can then be dissolved away to free the rotor after the
device is completely fabricated. In one version of the 4 mm tur-
bine of Fig. 6, an SiO2 film bonds the rotor layer at the thrust
bearing pad to the adjoining wafer, before the journal bearing is

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

etched. Another approach employs ‘‘break-off tabs’’ or mechani-
cal fuses, flimsy structures which retain the rotor in place during
fabrication and are mechanically failed after fabrication is com-
plete to release the rotor, 关19兴. Both approaches have been proven
successful.
The last MEMS technology we will mention is that for elec-
tronic circuitry, mainly for embedded sensors and electric machin-
ery such as actuators, motors, and generators. The circuitry is
generally constructed by laying down alternating insulating and
conducting layers, typically by using vapor deposition or sputter-
ing approaches, and patterning them as they are applied using the
photoresist technology outlined above. While the microelectronics
industry has developed an extremely wide set of such technolo-
gies, only a small subset are compatible with the relatively harsh
environment of the processing needed to realize wafer-bonded
mechanical structures hundreds of microns deep. Specifically, the
high wafer annealing temperatures limit the conductor choices to
polysilicon or high temperature metals such as platinum or tung-
sten. The energetic etching processes require relatively thick
masking material which limits the smallest electrical feature size
to the order of a micron, rather than the tens of nanometers used in
state-of-the-art microelectronic devices.
Fig. 8 Complete, five-layer turbine ‘‘stack’’ including bearings Using the above technologies, shapes are restricted to mainly
and fluid plumbing; „a… conceptual cross-section, „b… electron
microscope image of cross-section
prismatic or ‘‘extruded’’ geometries of constant height. Ongoing
research with greyscale lithography suggests that smoothly vari-
able etch depths 共and thus airfoils of variable span兲 may be fea-
sible in the near term, 关20兴. Conceptually, more complex three-
several square centimeters of surface from bonding兲. Thus, a very dimensional shapes can be constructed of multiple precision-
high standard of cleanliness and wafer handling must be main- aligned two-dimensional layers. But layering is expensive with
tained throughout the fabrication process. The wafers to be current technology and 5– 6 is considered a large number of
bonded are hydrated and then aligned using reference marks pre- precision-aligned layers for a microdevice. Since three-
viously etched in the surface. The aligned wafers are brought into dimensional rotating machine geometries are difficult to realize,
contact and held there by Van der Waals forces. The stack of planar geometries are preferred. While three-dimensional shapes
wafers is then pressed and heated to a few hundred degrees for are difficult, in-plane two-dimensional geometric complexity is
tens of minutes. Finally, the stack is annealed for about one hour essentially free in manufacture since photolithography and etching
at 1100°C in an inert gas furnace. 共If a lower temperature is used, process an entire wafer at one time. These are much different
a much longer time will be needed for annealing.兲 Such a stack, manufacturing constraints than are common in the large-scale
well processed from clean wafers, will not have any discernable world so it is not surprising the optimal machine design may also
bond lines, even under high magnification. Tests show the bonds be different.
to be as strong as the base material. The current state of the art is
stacks of 5– 6 wafers aligned across a wafer to 0.5–1.0 ␮m. More
wafers can be bonded if alignment precision is less important.
Note that the annealing temperature is generally higher than de-
vices encounter in operation. This process step thus represents the
Turbomachinery Fluid Mechanics
limiting temperature for the selection of materials to be included
within the device, 关18兴. Figure 8 shows the turbine layer of Fig. 6 The turbomachine designs considered to date for MEMS engine
bonded as the center of a stack of five wafers, the others contain applications have all been centrifugal since this geometry is
the thrust bearings and fluid plumbing. readily compatible with manufacturing techniques involving pla-
A third fabrication technology of interest for microrotating ma- nar lithography. 共It is also possible to manufacture single axial
chinery is that which realizes a freely spinning rotor within a flow stages by using intrinsic stresses generated in the manufac-
wafer-bonded structure. We require completed micromachines turing process to warp what otherwise would be planar paddles
which include freely rotating assemblies with clearances mea- into twisted blades, but such techniques have not been pursued for
sured in microns. While it is possible to separately fabricate ro- high-speed turbomachinery.兲 In most ways, the fluid mechanics of
tors, insert them into a stationary structure, and then bond an microturbomachinery are similar to that of large-scale machines.
overlaying static structure, this implies pick-and-place hand op- For example, high tip speeds are needed to achieve high pressure
erations 共rather than parallel processing of complete wafers兲 and ratios per stage. Micromachines are different in two significant
increases the difficulty in maintaining surfaces sufficiently clean ways: small Reynolds numbers 共increased viscous forces in the

210 Õ Vol. 126, APRIL 2004 Transactions of the ASME

fluid兲 and, currently, two-dimensional, prismatic geometry limita-
tions. The low Reynolds numbers, 103 – 105 , are simply a reflec-
tion of the small size, and place the designs in the laminar or
transitional range. These values are low enough that it is difficult
to diffuse the flow, either in a rotor or a stator, without separation.
This implies that either most of the stage work must come from
the centrifugal pressure change or that some separation must be
tolerated. The design challenges introduced by the low Reynolds
numbers are exacerbated by geometric restrictions imposed by
current microfabrication technology. In particular, the fabrication
constraint of constant passage height is a problem in these high-
speed designs. High work on the fluid means large fluid density

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

changes. In conventional centrifugal turbomachinery, density
change is accommodated in compressors by contracting or in tur-
bines by expanding the height of the flow path. However, conven-
tional microfabrication technology is not amenable to tapering
passage heights, so all devices built to date have a constant span.
How these design challenges manifest themselves are somewhat Fig. 9 Calculated sensitivity of two-dimensional airfoil loss
with Reynolds number, †9‡
different in compressors and turbines.
A common fluid design challenge is turning the flow to angles
orthogonal to the lithographically defined etch plane, such as at
the impeller eye or the outer periphery of the compressor diffuser. While extensive two-dimensional and three-dimensional nu-
At conventional scale, these geometries would be carefully con- merical simulations have been used to help in the design and
toured and perhaps turning vanes would be added. Such geometry analysis of the micromachines, as in all high-speed turbomachin-
is currently difficult to produce with microfabrication, which most ery development, test data are needed. Instrumentation suitable for
naturally produces sharp right angles that are deleterious to the fluid flow measurements in turbomachinery with blade spans of a
fluid flow. For example, at the 2 mm diameter inlet to a compres- few hundred microns and turning at over a million rpm is not
sor impeller, three-dimensional computational fluid dynamics readily available. While it is theoretically possible to microfabri-
共CFD兲 simulations show that a right-angle turn costs 5% in com- cate the required instrumentation into the turbomachine, this ap-
pressor efficiency and 15% in mass flow compared to a smooth proach to instrumentation is at least as difficult as fabricating the
turn, 关21兴. Engineering approaches to this problem include lower- microturbomachine in the first place. Instead, the standard tech-
ing the Mach number at the turns 共by increasing the flow area兲, nique of using a scaled turbomachine test rig was adopted, 关23兴. In
smoothing the turns with steps or angles 共which adds fabrication this case the test rig was a 75⫻ linear scaleup of a 4 mm diameter
complexity兲, and adding externally produced contoured parts compressor 共sufficiently large with a 300 mm rotor diameter for
when the turns are at the inlet or outlet to the chip. conventional instrumentation兲 rather than the 2– 4⫻ scaledown
common in industry. The geometry tested was a model of a 2:1
Compressor Aerodynamic Design. The engine cycle de- pressure ratio, 4 mm diameter compressor with a design tip speed
mands pressure ratios of 2– 4, the higher the better. This implies of 400 m/sec for use in a micromotor-driven air compressor, 关24兴.
that transonic tip Mach numbers and therefore rotor tip peripheral This design used the thick-blade-to-control-diffusion philosophy
speeds in the 400–500 m/s range are needed. This yields Rey- discussed above. The rig was operated at reduced inlet pressure to
nolds numbers 共Re兲 in the range of 104 for millimeter-chord match the full-scale design Reynolds number of about 20,000. A
blades. The sensitivity of two-dimensional blade performance to comparison of a steady three-dimensional viscous CFD 共FLU-
Re in this regime is illustrated in Fig. 9, which presents the varia- ENT兲 simulation to data is shown in Fig. 11. The simulation do-
tions of efficiency with size for a radial flow compressor and main included the blade tip gaps and right-angle turn at the inlet.
turbine. While this analysis suggests that for low loss it is desir- It predicts the pressure rise and mass flow rate to 5% and 10%,
able to maximize chord, note that the span of the airfoils is less respectively.
than the chord, implying that aero designs should include endwall Tight clearances are considered highly desirable for compressor
considerations at this scale. aerodynamics in general but are a two-edged sword for the thick-
In conventional size machines the flow path contracts to control bladed designs discussed above. Small tip clearance reduces leak-
diffusion. Since this was not possible with established microma-
chining technology, the first approach taken was to control diffu-
sion in blade and vane passages by tailoring the airfoil thickness
rather than the passage height, 关21,22兴. This approach results in
very thick blades, as can be seen in the 4:1 pressure ratio com-
pressor shown in Fig. 10. Compared to conventional blading, the
trailing edges are relatively thick and the exit angle is quite high.
The design trade is between thick trailing edges 共which add loss to
the rotor兲 or high rotor exit angles 共which result in reduced work
at constant wheel speed, increased diffuser loss, and reduced op-
erating range兲.
Although the geometry is two-dimensional the fluid flow is not.
The relatively short blade spans, thick airfoil tips, and low Rey-
nolds numbers result in large hub-to-tip flow variations, especially
at the impeller exit. This imposes a spanwise variation on stator
inlet angle of 15–20 deg for the geometries examined. This can-
not be accommodated by twisting the airfoils, which is not per-
mitted in current microfabrication. The limited ability to diffuse
the flow without separation at these Reynolds numbers also pre-
cludes the use of vaneless diffusers if high efficiency is required, Fig. 10 A 500 mÕs tip speed, 8 mm dia. centrifugal engine com-
since the flow rapidly separates off parallel endwalls. pressor

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 211
 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 12 The influence of heat addition on compressor perfor-
Fig. 11 Sensitivity of compressor pressure rise to tip clear- mance „pressure ratio is ␲, the subscript ‘‘ad’’ refers to the
ance „% span… adiabatic condition…

age flow and its associated losses, but increases drag for designs bine compressors using inlet-corrected mass flow as an indicator
in which the blade tip is at least as wide as the passage. The of size. The efficiency decreases with size but how much of this is
full-scale blading dimensions of the microcompressor tested intrinsic to the fluid physics and how much is due to the discrep-
scaled up was a blade chord of about 1000 ␮m and a span of 225 ancy in development effort 共little engines have little budgets兲 is
␮m. Thus the design minimum tip clearance of 2 ␮m 共set to avoid not clear. 共Note that there is an inconsistency of about a percent in
blade tip rubs兲 represents 0.2% of chord and 1% of span. Figure this data due to different definitions of efficiency, i.e., whether
11 includes measurements of the sensitivity of this design to tip losses in the inlet guide vanes and the exit vanes or struts are
clearance. included.兲
Recent microfabrication advances using greyscale lithography
approaches suggest that variable span turbomachinery may indeed Turbine Aerodynamic and Heat Transfer Design. While
be feasible, 关20兴. This would facilitate designs with attached flow the aerodynamic design of a microfabricated, centimeter-diameter
on thin blades. Compared to the thick blade approach, three- radial inflow turbine shares many of the design challenges of a
dimensional CFD simulations of thin blade compressors with a tip similar scale compressor, such as a constant airfoil span manufac-
shroud show about twice the mass flow for the same maximum turing constraint, the emphasis is different. Diffusion within the
span and wheel speed, an increase in pressure ratio from 2.5 to blade passages is not the dominant issue it is with the compressor,
3.5, and an increase in adiabatic efficiency from 50% to 70%, so the thick blade shapes are not attractive. The Reynolds numbers
关25兴. are lower, however, given increased viscosity of the high tempera-
Isomura et al. have taken a different approach to centimeter- ture combustor exit fluid. The nozzle guide vanes 共NGVs兲 operate
scale centrifugal compressors, 关26,27兴. They have chosen to scale at a Re of 1000–2000 for millimeter-chord airfoils.
a conventional three-dimensional aerodynamic machine with an One 6 mm diameter constant-span engine turbine is shown in
inducer down to a 12 mm diameter for a design 2 g/s mass flow Fig. 14. With a 400 ␮m span it is designed to produce 53 W of
rate and 3:1 pressure ratio. The test article is machined from alu- shaft power at a pressure ratio 共T-S兲 of 2.1, tip speed of 370 m/s,
minum on a high-precision five-axis miller. No test results have and mass flow of 0.28 g/s. The reaction is 0.2 which means that
been reported to date. the flow is accelerating through the turbine. Three-dimensional
Kang et al. 关28兴 have built a 12 mm diameter conventional CFD simulations were used to explore the performance of this
geometry centrifugal compressor from silicon nitride using a rapid design using FLUENT. The calculational domain included the
prototype technology known as mold shape deposition manufac- blade tip gap regions, the discharge of bearing air into the turbine,
ture. It was designed to produce a pressure ratio of 3:1 at 500 m/s
tip speed with a mass flow of 2.5 g/s and an efficiency of 65–
70%. To date, they report testing up to 250 m/s and performance
consistent with CFD analysis.
A major aerodynamic design issue peculiar to these very small
machines is their sensitivity to heat addition. It is difficult to de-
sign a centimeter-scale gas turbine engine to be completely adia-
batic, thus there will be some degree of heat addition through
conduction. An isothermal compressor at fixed temperature exhib-
its behavior close to that of an adiabatic machine with the same
amount of heat added at the inlet, 关29兴. Thus, the influence of the
heat addition shows up as reductions in mass flow, pressure rise,
and adiabatic efficiency. The effect of heat addition on compressor
efficiency and pressure ratio are shown in Fig. 12. These effects
can be quite dramatic at high levels of heat flow. The influence of
this nonadiabatic behavior on the overall cycle performance will
be discussed later.
The ultimate efficiency potential for compressors in this size
range has yet to be determined. Figure 13 plots the polytropic Fig. 13 Variation of engine compressor polytropic efficiency
efficiency of a number of aeroengines and ground-based gas tur- with size

212 Õ Vol. 126, APRIL 2004 Transactions of the ASME

 stable regime 共e.g., the boundary layers should stay attached兲,
关30兴. He then designed film-cooled turbines and analyzed these
designs with CFD simulation.
Based upon the work to date, it should be possible to realize
microfabricated single-stage compressors with adiabatic pressure
ratios above 4:1 at 500 m/s tip speed with total-to-static efficien-
cies of 50– 60%. Achievable turbine efficiencies may be 5–10%
higher.

Combustion
The primary design requirements for gas turbine combustors
include large temperature rise, high efficiency, low pressure drop,

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

structural integrity, ignition, stability, and low emissions. These
requirements are no different for a microcombustor which may
flow less than 1 g/s of air than for a 100 kg/sec large machine, but
the implementation required to achieve them can be. A compari-
son between a modern aircraft engine combustor and a mi-
croengine is shown in Table 1, 关31兴. Scaling considerations result
in the power density of a microcombustor exceeding that of a
large engine. However, the combustor volume relative to the rest
of the microengine is much larger, by a factor of 40, than that of
a large engine. The reasons for this scaling can be understood in
reference to the basics of combustion science, 关32兴.
Combustion requires the mixing of fuel and air followed by
Fig. 14 Silicon engine radial inflow turbine inside annular chemical reaction. The time required to complete these processes
combustor; the flow passages in the NGV’s are for bearing and is generally referred to as the required combustion residence time
balance air and effectively sets the minimum volume of the combustor for a
given mass flow. The mixing time can scale with device size but
the chemical reaction times do not. In a large engine, mixing may
and the right-angle turn and duct downstream of the rotor. These account for more than 90% of combustor residence time. A useful
calculations predict that this design has an adiabatic efficiency of metric is the homogeneous Damkohler number, which is the ratio
about 60%. The remainder of the power goes to NGV losses 共9%兲, of the actual fluid residence time in the combustor to the reaction
rotor losses 共11%兲, and exit losses 共20%兲, 关30兴. These are very low time. Obviously a Damkohler of one or greater is needed for
aspect ratio airfoils 共⬃0.25兲 and this is reflected in the shear on complete combustion and therefore high combustion efficiency.
the endwalls being about twice that on the airfoil surfaces. The One difference between large and microscale machines is the in-
exit losses, the largest source of inefficiency, consist of residual creased surface area-to-volume ratio at small sizes. This offers
swirl, losses in the right-angle turn, and lack of pressure recovery more area for catalysts; it also implies that microcombustors have
in the downstream duct. This implies that 共a兲 the rotor exit Mach proportionately larger heat losses. While combustor heat loss is
number should be reduced if possible, and 共b兲 that the turbine negligible for large-scale engines, it is a dominant design factor at
would benefit from an exit diffuser. microscale since it can reduce the combustor efficiency and lower
High engine-specific powers require turbine inlet temperatures the reaction temperature. This narrows the flammability limits and
共TIT兲 above the 950 K capability of uncooled single-crystal Si. decreases the kinetic rates, which drops the effective Damkohler
The MIT demo engine was designed with a TIT of 1600 K and so number. As an example, Fig. 15, 关31兴, illustrates the viable design
requires turbine cooling. In the demo design the turbine is con- space for an H2 -fuelled, 0.07 cc microcombustor as a function of
ductively cooled through the structure. The heat flow is on the the heat lost to the walls and as constrained by flame stability,
same order as the shaft power, and the resultant entropy reduction structure limits, and cycle requirement considerations. The design
is equivalent to 1–2% improvement in turbine efficiency. Ad- space shown permits a trade between heat loss and stoichiometry,
vanced engine designs may use film cooling. A major issue in this which is especially important when burning hydrocarbons with
case is the stability of a cold boundary layer on a rotating disk narrow stoichiometry bounds.
with radial inflow. While this is, in general, an unstable flow, The design details are dependent on the fuel chosen. The design
Philippon has shown through analysis and CFD simulation that approach first taken was to separate the fuel-air mixing from the
the region of interest for these millimeter-scale turbines lies in a chemical reaction. This is accomplished by premixing the fuel

Table 1 A comparison of a microengine combustor with a large aeroengine combustor

Conventional Micro-
Combustor Combustor
Length 0.2 m 0.001 m
Volume 0.073 m3 6.6⫻10⫺8 m3
Cross-sectional area 0.36 m2 6⫻10⫺5 m2
Inlet total pressure 37.5 atm 4 atm
Inlet total temperature 870 K 500 K
Mass flow rate 140 kg/s 1.8⫻10⫺4 kg/s
Residence time ⬃7 ms ⬃0.5 ms
Efficiency ⬎99% ⬎90%
Pressure ratio ⬎0.95 ⬎0.95
Exit temperature 1800 K 1600 K
Power Density 1960 MW/m3 3000 MW/m3

共Note: residence times are calculated using inlet pressure and an average flow temperature of 1000 K兲

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 213
 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 15 Design space for Si H2 microcombustor

with the compressor discharge air upstream of the combustor

flame holders. This permits a reduction of the combustor resi-
dence time required by a factor of about 10 from the usual 5–10
msec. The disadvantage of this approach is a susceptibility to
Fig. 16 Measured performance of 0.2 cc, Si microcombustors
flashback from the combustor into the premix zone, which must
using H2 fuel
be avoided. To expedite the demonstration of a micro-gas turbine
engine, hydrogen was chosen as the initial fuel because of its wide
flammability limits and fast reaction time. This is the same ap-
proach taken by von Ohain when developing the first jet engine in
Germany in the 1930s. Hydrogen is particularly attractive because geneous reactions on the surface of a catalyst can widen the flam-
it will burn at equivalence ratios, ␾, as low as 0.3 which yields mability limits and so reduce the combustion temperature. Both
adiabatic combustion temperatures below 1500 K, facilitating the approaches have been demonstrated at microscale. Ethylene
realization of simple premixed designs. 共which has a high reaction rate兲 and propane have been burned in
Microcombustor technology has been developed in several full- the H2 combustors described above. The combustion efficiency
sized 共i.e., micro兲 test rigs which duplicate the geometry of an with ethylene approached 90% while that for propane was closer
engine but with the rotating parts replaced with stationary swirl to 60%. These fuels need larger combustor volumes compared to
vanes, 关33兴. In the Si micromachined geometry of Figs. 3 and 4, to hydrogen for the same heat release. Data for a variety of geom-
reduce heat losses through the walls and therefore to increase etries and fuels is reduced in terms of Damkohler number in Fig.
combustor efficiency, the inlet air wraps around the outside of the 17, which shows that values of greater than 2 are needed for high
0.2 cc combustor before entering through flame holders in a re- chemical efficiency, 关31兴.
versed flow configuration. This configuration is similar to the tra- Catalytic microcombustors have been produced by filling the
ditional reverse-flow engine combustor but scaled down to 0.1– combustor volume of the above geometries with a platinum-
0.3 g/sec air flow rate. The Si liner in this case is conduction coated nickel foam. For propane, the catalyst increased the heat
rather than film-cooled. In this premixed approach, fuel is injected release in the same volume by a factor of 4 –5 compared to the
near the inlet of the upstream duct to allow time for fuel-air mix- propane-air results discussed above. Pressure drops through the
ing without requiring additional combustor volume. This design foam are only 1–2%, 关35兴. Presumably catalytic combustor per-
takes advantage of microfabrication’s ability to produce similar
geometric features simultaneously, using 90 fuel injection ports,
each 120 ␮m in diameter, to promote uniform fuel-air mixing. A
simple hot wire loop provides ignition, 关34兴.
The combustor was tested in several configurations including
variations of flame holder and dilution hole geometry. Combus-
tion efficiencies approaching 100% have been reported with pres-
sure ratios of about 0.95–0.98. The H2 data in Fig. 16 show the
variation of combustor efficiency versus mass flow rate for two
configurations, one purely premixed 共no dilution holes兲 and one in
which dilution holes have been added to the liner creating a dual-
zone combustor, 关31兴. The missing data is due to instrumentation
burnout. The dual-zone configuration, in which the dilution jets
set up recirculation zones within the combustor, extends the oper-
ating range by about a factor of two at a cost of 10–20% in
combustor efficiency. These combustors have been operated at
exit temperatures above 1800 K.
Hydrocarbon fuels such as methane and propane have reaction
rates only about 20% of those of H2 , requiring larger combustor
volumes for the same heat release. They also must react closer to
stoichiometric and therefore at higher temperatures, above 2000
K. For gas phase 共homogeneous兲 combustion designs this requires
a multizone burner 共stoichiometric zone followed by a dilution Fig. 17 Measured microcombustor performance as a function
region兲 as used on most large gas turbines. Alternatively, hetero- of Damkohler number

214 Õ Vol. 126, APRIL 2004 Transactions of the ASME

formance can be improved by a better choice of catalyst 共platinum in the micro-gas turbine, so considerable cooling may be needed.
was selected for H2 ) and a geometry optimized for catalytic rather For these reasons, the first efforts concentrated on designs em-
than gas-phase combustion. ploying electric fields. The designs examined did not appear
Takahashi et al. 关36兴 are developing combustors designed for promising in that the forces produced were marginal compared to
somewhat larger gas turbines, with flow rates of about 2 g/s. De- the bearing loads expected, 关38兴. Also, since electromagnetic bear-
signed for methane, these are a miniature version of can-type ings are unstable, feedback stabilization is needed, adding to sys-
industrial combustion chambers with a convection-cooled liner tem complexity.
and dilution holes. These are conventionally machined with vol- Gas bearings support their load on thin layers of pressurized
umes of 2– 4 cc. The combustion efficiencies of these units have gas. For micromachines such as turbines they have intrinsic ad-
been demonstrated as above 99% at equivalence ratios of about vantages over electromagnetic approaches, including no tempera-
0.37 with a design combustor exit temperature of 1323 K. The ture limits, high load-bearing capability, and relative manufactur-
design residence time is about 6.5 ms. Matsuo et al. 关37兴 con- ing simplicity. At large scale, gas bearings are used in many high-
structed a still larger 共20 cc volume, 16 g/s flow rate兲 convention-

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

speed turbomachinery applications, including aircraft
ally machined combustor burning liquefied natural gas. They re- environmental control units, auxiliary power units, 30–70 kW
port a combustor exit temperature of about 1200 K. ‘‘microturbines,’’ and turbochargers, 关39兴. At smaller scale, gas
Overall, experiments and calculations to date indicate that high- bearings have been used in gyroscopic instruments for many
efficiency combustion systems can be engineered at microscale years. All else being the same, the relative load-bearing capability
and achieve the heat release rate and efficiency needed for very of a gas bearing improves as size decreases since the volume-to-
small gas turbine engines. surface area ratio 共and thus the inertial load兲 scales inversely with
size. Rotor and bearing dynamics scaling is more complex, 关40兴.
Bearings and Rotor Dynamics However, rotor dynamics in this application are somewhat simpli-
fied compared to large engines since the structure is very stiff, so
The mechanical design of gas turbine engines is dominated by
only rigid-body modes need be considered. In the following para-
the bearings and rotor dynamics considerations of high-speed ro-
graphs we will first discuss journal bearings which support radial
tating machinery. Micromachines are no different in this regard.
loads and then consider thrust bearings needed for axial loads.
As in all high-speed rotating machinery, the basic mechanical ar-
The simplest journal bearing is a cylindrical rotor within a
chitecture of the device must be laid out so as to avoid rotor
close-fitting circular journal. Other more complex variations used
dynamic problems. The high peripheral speeds required by the
in large-size machines include foil bearings and wave bearings.
fluid and electromechanics lead to designs which are supercritical
These can offer several advantages but are more difficult to manu-
共operate above the natural resonant frequency of the rotor system兲,
facture at very small size. Thus, the plane cylindrical geometry
just as they often are in large gas turbines.
was the first approach adopted since it seemed the easiest to mi-
Key design requirements imposed by the rotor dynamics are
that mechanical critical 共resonant兲 frequencies lie outside the crofabricate. Gas bearings of this type can be categorized into two
steady-state operating envelope, and that any critical frequencies general classes which have differing load capacities and dynami-
that must be traversed during acceleration are of sufficiently low cal characteristics. When the gas pressure is supplied from an
amplitude to avoid rubs or unacceptable vibrations. The bearings external source and the bearing support forces are not a first order
play an important role in the rotor dynamics since their location function of speed, the bearing is termed hydrostatic. When the
and dynamical properties 共stiffness and damping兲 are a major de- bearing support forces are derived from the motion of the rotor,
terminant of the rotor dynamics. The bearings in turn must support then the design is hydrodynamic. Hybrid implementations com-
the rotor against all radial and axial loads seen in service. In bining aspects of both are also possible. Since the MEMS gas
addition to the rotor dynamic forces, the bearing loads under nor- turbines include air compressors, both approaches are applicable.
mal operation include all the net pressure and electrical forces Both can readily support the loads of machines in this size range
acting on the rotor as well as the weight of the rotor times the and can be used at very high temperatures. The two types of
external accelerations imposed on the device. For aircraft engines bearings have differing load and dynamic characteristics. In hy-
this is usually chosen as 9 g’s, but a small device dropped on a drodynamic bearings, the load capacity increases with the speed
hard floor from two meters experiences considerably larger peak since the film pressure supporting the rotor is generated by the
accelerations. An additional requirement for portable equipment is rotor motion. This can be true for a hydrostatic bearing as well if
that the rotor support be independent of device orientation. The the film pressure is increased with increasing rotor speed, for ex-
bearing technology chosen must be compatible with the high tem- ample if the pressure is derived from an engine compressor. How-
peratures in a gas turbine engine 共or be protected within cooled ever, when the supporting film pressure in a hydrostatic bearing is
compartments兲 and be compatible with the fabrication processes. kept constant, the load capacity decreases slightly with increasing
Early MEMS rotating machines have been mainly microelectric speed. The calculated unit load capacity 共support force per unit
motors or gear trains turning at significantly lower speeds and for area of bearing兲 of plane journal microbearings is compared with
shorter times than are of interest here, so these made do with dry the measured capacity of conventional air foil bearings in Fig. 18.
friction bearings operating for limited periods. The higher speeds The hydrostatic bearing is at a constant pressure. For hydrody-
and longer lives desired for micro-heat engines require low fric- namic bearings the load capacity is a function both of rotational
tion bearings. The very small size of these devices precludes the speed and of bearing length 共L兲 to diameter 共D兲 ratio. Microbear-
incorporation of commercially available rolling contact bearings. ings currently have low L/D’s due to manufacturing constraints,
A microfabricated bearing solution is needed. Both electromag- so their load capacity is less.
netic and air bearings have been considered for this application. The relevant physical parameters determining the bearing be-
Electromagnetic bearings can be implemented with either mag- havior are the length-to-diameter ratio (L/D); the journal gap-to-
netic or electric fields providing the rotor support force. Although length ratio (g/L); and nondimensional forms of the peripheral
extensive work has been done on the application of magnetic Mach number of the rotor 共a measure of compressibility兲, the
bearings to large rotating machinery, work is just beginning on Reynolds number, and the mass of the rotor. For a bearing sup-
magnetic bearings for micromachines. In addition to their com- ported on a hydrodynamic film, the load bearing capability scales
plexity, magnetic bearings have two major challenges in this ap- inversely with (g/D) 5 which tends to dominate the design consid-
plication. First, magnetic materials are not compatible with most erations, 关41兴.
microfabrication technologies, limiting device fabrication options. The design space available for the micro-journal bearing is
Second, Curie point considerations limit the temperatures at greatly constrained by manufacturing capability, especially if the
which magnetic designs can operate to below those encountered rotor and journal structure are fabricated at the same time 共which

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 215
 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 19 Gas journal bearing model

The measured imbalance shown in the figure is ⬃2 ␮m, compared

Fig. 18 Gas bearing radial unit load capacity variation with to the 12 ␮m bearing clearance 共i.e., at 12 ␮m imbalance the rotor
speed. „Figure courtesy of L. Liu.…
would strike the wall on every revolution兲.
We thus have two rotor dynamic design considerations, travers-
ing the critical frequency and ensuring that the frequency for the
onset of instability is above the operating range. For a hydrostatic
bearing the critical frequency simply scales with the pressure in
avoids the need for assembly and so facilitates low-cost wafer-
the bearing. The damping ratio 共mainly viscous damping兲 de-
level manufacturing兲. The most important constraint is the etching
creases with increasing speed. Thus, the maximum amplitude the
of vertical side walls. Recent advances of etching technology
rotor experiences while crossing the critical frequency increases
yield taper ratios of about 30:1 to 50:1 on narrow 共10–20 ␮m兲
with bearing pressure, i.e., the peak in Fig. 20 moves up and to the
etched vertical channels 300–500 ␮m deep, 关15兴. This capability
right with increasing pressure, 关42兴. This suggests the strategy of
defines the bearing length while the taper ratio delimits the bear-
crossing the critical frequency at low pressure and low speed and
ing gap, g. For hydrodynamic bearings we wish to maximize the
then increasing the pressure to stiffen the bearing as the rotor
footprint and minimize gap/diameter to maximize load capacity,
accelerates to increase the speed at onset of instability, 关43兴.
so the bearing should be on the largest diameter available, the
The rotor imbalance is another factor which influences both the
periphery of the rotor. The penalty for the high diameter is rela-
peak amplitude crossing the critical frequency and the onset of
tively high area and surface speed, thus high bearing drag, and
instability. Large rotating machines are usually dynamically bal-
low L/D and therefore reduced stability. In the radial 4000 ␮m
anced by measuring the imbalance and then adding or subtracting
diameter turbine shown in Fig. 6, the journal bearing is 300 ␮m
mass to reduce it. In many micromachines it is possible to avoid
long and about 15 ␮m wide, so it has an L/D of 0.075, g/D of
the need for dynamic balancing because the base material used to
0.038, and peripheral Mach number of 1. This relatively short,
date 共single-crystal silicon兲 is extremely uniform and, with suffi-
wide-gapped, high-speed bearing is well outside the range of ana-
cient care, the etching uniformity is sufficient to produce ad-
lytical and experimental results reported in the gas bearing litera-
equately balanced rotors. Typically, the center of mass is within
ture. It is much closer to an air seal in aspect ratio.
1–5 ␮m of the geometric center. For the turbine in Fig. 6, the
The dynamical behavior of the rotor is of first-order concern
blades must be etched to about ⫾1 ␮m span uniformity across the
because the high rotational speeds needed for high power density
4000 ␮m diameter disk. For rotors made up of several wafers, the
by the turbo and electrical machinery require the rotor to operate
alignment between wafers must also be considered 共also about 1
at rotational frequencies several times the lowest radial resonant
␮m is needed兲, 关44兴. Using the balance measurement capability
frequency of the bearing/rotor system. The dynamics of gas bear-
evident in Fig. 20 and laser etching, it is also possible to dynami-
ings on a stiff rotor can be simply represented by the rotor
mounted on a set of springs and dampers, as illustrated in Fig. 19.
The fluid in the bearing acts as both the springs and the principal
source of damping. It also generates the destabilizing cross-
stiffness forces which cause instability at high speeds. As in many
conventional engines, the rotor must traverse the critical fre-
quency and avoid instabilities at higher speeds. For example, Fig.
20 illustrates the whirl radius versus speed for a 4 mm diameter
turbine with a 12 ␮m wide bearing. Plotted on the figure are
experimental data and a fit of an analytical fluid mechanic spring-
mass-damper model of the system to that data. The resonant peak
amplitude is reached as the rotor crosses a ‘‘rotor critical’’ 共reso-
nant兲 frequency. If the peak excursion exceeds the bearing clear-
ance, then the rotor hits the wall, i.e., ‘‘crashes.’’ A well-known
characteristic of a spring-mounted rotor system 共a so-called ‘‘Jeff-
cott rotor’’兲 is that at speeds below the critical frequency the rotor
revolves around its geometric center, while well above the critical
frequency the rotor revolves around its center of mass. Thus the
dotted line in the figure, the asymptote of the curve fit, is a mea-
surement of the rotor imbalance expressed in terms of radial dis- Fig. 20 Transcritical response of the micro-journal gas bear-
placement of the rotor center of mass from the geometric center. ing in Fig. 6. „Figure courtesy of C. J. Teo.…

216 Õ Vol. 126, APRIL 2004 Transactions of the ASME

 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 21 Geometry of „a… hydrodynamic and „b… hydrostatic
thrust bearings „not to scale…

Fig. 22 Hybrid hydrodynamic „spiral grooves… and hydrostatic

„orifices… 0.7 mm dia. thrust bearing
cally balance a microrotor. It is unclear at this time whether
dynamic balancing or manufacturing uniformity is a superior
approach. mum rotational speed needed to develop sufficient pressure to
Hydrostatic bearings are stable from zero speed up to the sta- eliminate rubbing between the stationary and rotating parts. Fig-
bility boundary. However, centered hydrodynamic bearings are ure 22 is a 700 ␮m diameter hydrodynamic thrust bearing with 1.5
unstable at low rotational speed but stable at high speeds. Com- ␮m deep spiral grooves that was tested on the microturbine of Fig.
monly, such bearings are stabilized by the application of a unidi- 6. It lifts off at about 80,000 rpm. Such a bearing at 106 rpm will
rectional force which pushes the rotor toward the journal wall, as dissipate about 0.2 watts, about the same as that of a hydrostatic
measured by the eccentricity, the minimum approach distance of bearing of equal load capacity, 关46兴.
the rotor to the wall as a fraction of the average gap 共0⫽centered,
1⫽wall strike兲. At conventional scale, the rotor weight is often the Structures and Materials
source of this side force. At microscale, 共1兲 the rotor weight is
Structural considerations for the design of a MEMS gas turbine
negligible, and 共2兲 insensitivity to orientation is desirable, so a
are in many ways similar to those of conventional engines. The
scheme has been adopted which uses differential gas pressure to
design space is defined by the requirements of the thermodynam-
force the rotor eccentric. Extensive numerical modeling of these
ics 共which require high stress and high temperatures兲, the proper-
microbearing flows has shown that such a rotor will be stable at
ties of the materials, and the manufacturing capabilities. The ma-
eccentricities above 0.8 –0.9, 关45兴. For the geometry of the turbine
terial properties, in turn, are very much dependent on their
in Fig. 6, the rotor must thus operate between 1–2 ␮m from the
processing. This section reviews materials selection, structural de-
journal wall. This implies that deviations from circularity of the
sign features, high-temperature structures, analysis of such micro-
journal and rotor must be small compared to 1 ␮m, an additional
structures, and packaging 共installation兲 technology.
manufacturing requirement.
A rotor must be supported against axial as well as radial loads Materials. Materials for gas turbine engines must exhibit
and so requires thrust bearings in addition to the radial bearings high specific strength 共strength/density兲 at high temperatures.
discussed above. Both hydrostatic and hydrodynamic approaches High-temperature operation also requires creep and oxidation re-
have been demonstrated. In either approach, the bearing must sup- sistance. Other properties of interest include fracture toughness,
port the axial loads and remain stable. The devices built to date modulus, and resistance to thermal shock 共Table 2兲. MEMS pro-
have been designed for subcritical thrust-bearing operation so that cessing technologies are much more mature for silicon than for
bearing behavior traversing the critical frequency is not an issue. other materials so it is the first material a MEMS engineer con-
Hydrostatic thrust bearings meter external air through supply siders 共not so for a gas turbine designer兲. In terms of strength at
orifices onto the bearing surface. The 400 ␮m diameter thrust pad temperature, single-crystal Si is the equal of common nickel al-
at the rotor center of the 4 mm diameter turbine in Fig. 6 rotates loys and, because it has only 1/3 the density, its specific strength is
relative to a stationary thrust bearing surface of similar diameter. much higher, as illustrated in Fig. 23. It is quite oxidation-resistant
The stationary bearing surface is perforated with a circular array and has thermal conductivity approaching that of copper, so it is
of 12 ␮m diameter nozzle orifices fed from a plenum which sup- resistant to thermal shock. On these grounds, it is not a bad ma-
plies the gas lubricating film between the bearing surfaces. A cross terial for gas turbine engines. However, at temperatures below
section is shown in Fig. 21. At a rotor-stator gap of 1.2 ␮m, a flow about 900 K, Si is brittle, so usable strength is very much a func-
of 10 sccm at 2–5 atm is needed to provide sufficient load capac- tion of the details of the processing. Structural life must be as-
ity 共0.5 N兲 and axial stiffness (2⫻105 N/m). Stiffness is maxi- sessed with statistical methods.
mized when the pressure drop through the supply orifices equals Chen et al. 关47兴 have reported room temperature strengths up to
that of the radial outflow from the orifice discharge to the bearing 4 GPa for micromachined Si specimens. Moon et al. 关48兴 have
edge. measured the strength and creep rate of Si at temperatures up to
Hydrodynamic thrust bearings use viscous drag, often enhanced 1000 K. From these measurements and a detailed model of the
with shallow spiral grooves, to generate a pressure gradient in the creep behavior of the material, it appears that long-lived structures
bearing which increases toward the rotor center. The pressurized can be designed for stress levels up to about 500 MPa at 850 K.
gas film provides the bearing load capacity and stiffness. This Oxidation is another concern for high temperature structures.
self-pumping eliminates the need for an external air supply and Conductively cooled Si combustor tests were run at exit tempera-
simplifies the manufacture since bearing air supply plumbing is tures up to 1800 K, 关33兴. The thickness of the oxide layers grown
not required, reducing the number of wafers needed. It also adds were in agreement with standard models of Si oxidation. These
an additional design consideration—rotor liftoff, i.e., the mini- imply that uncoated Si airfoils can have a life of a few hundred

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 217
 Table 2 Design considerations and material properties of interest for gas turbines

Ni-Based
Super Titanium Micro Micro
Alloys Alloys SiC Silicon
Centrifugal stress 330 420 670 1000
关 冑␴ f / ␳ 兴 (m/s)
Thermal stress 2.7⫻10⫺3 1.2⫻10⫺3 1.1⫻10⫺3 0.9⫻10⫺3
关 ␣ E/ ␴ f /y 兴
Stiffness ⬃26 ⬃25 ⬃95 ⬃70
关 E/ ␳ 兴 (MPa/Kgm⫺3 )
Max temp.共°C兲 ⬃1000 ⬃300 ⬃1500 ⬃600
共life limit兲 共creep兲 共strength兲 共oxidation兲 共creep兲

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

hours. Longer lives may require coatings. Si nozzle guide vanes 关49兴. The increased temperature capability of a turbine like that in
run for 5 hrs at 1600 K in a microcombustor exhaust show little Fig. 6 increases with the thickness of the SiC insert, Fig. 25.
degradation, Fig. 24. Another approach being pursued is reaction sintering of powdered
Silicon carbide has about 600 K more temperature capability SiC to form parts such as turbine rotors, 关50兴.
than Si, but the SiC microfabrication technology is much less Additional structural materials of interest for MEMS gas tur-
mature. SiC is available in single-crystal wafers and can be pre- bines include glasses for thermal and electrical isolation, and very
cision etched but SiC wafers cost 100⫻ more than Si at the mo- high temperature materials such as sapphire. There is considerable
ment and etch rates are about 10⫻ slower. An alternative to direct microfabrication experience with glass but very little with the re-
SiC etching is to etch a female mold in Si and then fill the mold fractory ceramics because these have not been considered as
共for example, by chemical vapor deposition, CVD兲, and dissolve MEMS materials in the past.
away the Si, leaving an SiC precision structure. The challenges
here are realizing SiC with the needed mechanical properties and Structural Design Considerations. Structural design of a
dealing with the intrinsic stresses induced by the combination of MEMS gas turbine has many of the same considerations as the
the high temperatures of the CVD process and the difference in design of large machines: basic engine layout is set by rotor dy-
coefficient of thermal expansion between the two materials. A namic considerations, centrifugal stress is the primary rotor load,
variation on this approach is to use CVD to fill cavities in Si stress concentrations must be avoided, and hot section life is creep
wafers with SiC, bond another Si wafer over the filled cavity, and and oxidation-limited. Some large engine concerns do not exist at
then process the pair as though it were a standard Si wafer. This micron scale. For example, the microrotors are very stiff so that
yields SiC-reinforced silicon structures which have more tempera- backbone bending is not a concern; thermal stress from tempera-
ture capability than Si but are easier to manufacture than SiC, ture gradients is not important at these sizes; maintenance is not a

Fig. 23 Material properties relevant to high speed, high temperature rotating machinery

Fig. 24 200 ␮m high, Si turbine blades new and after 5 hrs at 1600 K gas temperature in a
microcombustor exhaust

218 Õ Vol. 126, APRIL 2004 Transactions of the ASME

 avoid unsafe conditions such as over-speed, over-temperature, or
surge. Such a control system consists of sensors 共speed, pressure,
temperature, etc.兲, a feedback controller with a suitable set of
control laws 共now implemented in a digital computer兲, actuators
such as a fuel control valve 共often called a fuel management unit
or FMU兲, and compressor stability devices as needed 共bleed
valves, fences, variable stators兲. Engine accessories include an
ignition system, fuel pump, lubrication system, and starter. All of
this functionality is needed for a millimeter-scale MEMS gas tur-
bine and all must fit within a micro chip if the accessories are not
to dwarf the engine. Following in the tradition of large engine
development, the controls and accessories have received less at-

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

tention to date than the major engine subsystems such as the com-
pressor, turbine, and combustor but they are no less important to
the ultimate success of the concept.
Engine Controls. The simplest engine control would consist
of a single sensor feeding a digital controller which commands the
fuel flow rate valves. The functional requirements for the valving
and the sensors stem from the engine dynamics as represented in
the control laws and from the engine environment. There is a very
rich literature on MEMS sensors and valves, literally thousands of
papers, and some units are commercially available. As for large
Fig. 25 Usable strength of SiÕSiCÕSi hybrid structure in ten- engines, however, the combination of harsh environment, high-
sion. „Figure courtesy of H.-S. Moon.… frequency response, high accuracy, and high reliability means that
sensors and actuators for MEMS-scale gas turbines must be spe-
cifically engineered for that environment.
design issue; and fasteners do not exist here so the engineering Engine control laws are generally based on reduced order mod-
details involved with bolting, static sealing, etc., do not exist, 关51兴. els of the engine dynamics. The dynamics of millimeter-scale en-
Although many of the design considerations are independent of gines are, of course, much faster than larger engines and can also
size, the engineering values are not, of course. Airfoils need fillets include phenomena not seen in large engines. The additional dy-
at the roots to avoid stress concentrations with radii of 10–30 ␮m. namics arise if there is significant heat transfer from the hot sec-
Surface finish is important with roughness measured in nanom- tion into the compressor, 关55兴. In this case the heat transfer de-
eters. Forced response excitation of blade rows must be avoided grades the compressor performance so that the pressure ratio and
with blade-bending frequencies on the order of megahertz rather mass flow are a function of hot section temperature as well as
than kilohertz, the rotor once-per-rev frequency is 20 KHz rather shaft speed. Since the heat transfer has a time constant not much
than 200 Hz. For the turbine of Fig. 6, the lowest blade mode is faster than the rotor acceleration, it alters the dynamics of the gas
2.5 MHz while the blade-passing frequency is 0.9 MHz. turbine from that of a first-order system 共as large engines are兲 to a
Below 850–900 K, silicon is a brittle material so that probabi- second-order system, requiring additional sophistication in the
listic analysis is a preferred method for failure analysis. Such control law design. The best way of avoiding this complexity is to
techniques applied to the turbine rotor geometry of Fig. 6 at pe- thermally isolate the hot and cold sections of the engine, which is,
ripheral speeds of 500 m/s predict failure probabilities of 10⫺10 to of course, desirable for improved thermodynamic performance.
10⫺8 , depending upon the flaw population assumed, 关52兴. In a Sensors. Large-scale gas turbines use compressor pressure ra-
rotor constructed from single-crystal Si, all of the flaws are likely tio and/or rpm as the primary input to the fuel control system.
to be surface flaws. In a rotor of CVD SiC, volumetric flaws may Sensor selection for a MEMS engine is a trade among observabil-
also exist. In either case, the flaw population and thus the usable ity of the state 共dynamical information represented by the sensed
strength of the material is a strong function of the manufacture, as quantity兲, response time, difficulty of fabrication, and environ-
it is at any scale. A variation of a factor of four in strength has mental compatibility. Liu 关55兴 used a dynamic model of a MEMS
been reported for deep-etched Si depending upon post-etch sur-
engine to evaluate the suitability of various sensor options includ-
face treatment, 关53兴.
ing rpm and compressor discharge pressure or temperature. Of
Large engines use standard tubing fittings and electrical con-
these, rpm is the most sensitive and temperature the least. Sensing
nectors to pass fluids and electrical signals to the outside world.
is complicated by the high rotational frequencies 共1,000,000 rpm兲
These do not exist at microscale. Most computer chips and
and high temperatures 共600 K at the compressor discharge兲 in a
MEMS devices do not require fluid connections and those that do
very small engine. In principle, sensors can be fabricated inte-
operate not much above room temperature. For these applications,
grally with an engine or located remotely, with each approach
there are a variety of adhesives and polymer systems. A micro-gas
presenting challenges. A sensor remote from the gas path suffers
turbine can have a surface temperature above 700 K and require
reduced frequency response, which is already a challenge at the
fluid connection at pressures of 10 atm or more, so that high-
MEMS scale. In addition, it is difficult to be very remote from the
strength high-temperature packaging approaches are needed. One
approach which has proven successful is an adaptation of the her- gas path in an 0.5 mm long engine. Integral sensors must with-
metic package technology used for military electronics. This joins stand the high temperature of the gas path. Even for sensors fab-
Kovar 共a nickel alloy兲 tubing to silicon using glass as the bonding ricated in the cold sections, they must be capable of withstanding
agent. The joining is done in a furnace above 1100 K to melt the a wafer-processing environment that exceeds 1000–1400 K. This
glass. Such joints can withstand pressures above 200 atm, 关54兴. effectively precludes the use of low temperature materials such as
polymers and most metals in the device design. Integral sensors
have several advantages, however. They can be very small and
Engine Controls and Accessories thus have high-frequency response, and many can be fabricated in
All gas turbine engines require control systems to insure safe parallel for low-cost redundancy.
operation. Typically, the control system adjusts the fuel flow to One integral solution was developed by Tang 关56兴 who adapted
deliver the requested power, and monitors engine operation to a hot-film-type sensor to this application 共Fig. 26兲. Designed for

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 219
 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 26 A 50 ␮m sq hot film RPM and temperature sensor

placement on the wall above a rotor blade tip, this 50 ␮m square

sensor is a heated, serpentine, polysilicon resistor positioned over
a trench for thermal isolation. The sensor and its leadouts are all
Fig. 27 A 1 mm dia. fuel control valve on Si beam springs
polysilicon which is selectively doped to adjust its resistivity
共high for the sensor, low for the leads兲. Polysilicon has the advan-
tage that it is compatible with most semiconductor fabrication
techniques and can withstand high temperatures. Simulations con- Starter-Generator. Microelectrical machinery is required for
firmed by shock tube testing showed this approach to have suffi- power generation and electric starting, if desired. There is an ex-
cient sensitivity and frequency response to respond to the flow tensive literature on microelectric motors, which is not reviewed
perturbations above a compressor blade tip as predicted by a here, but little work on generators. The requirements for the de-
three-dimensional CFD simulation. With a total thickness of less vices of interest here differ from previous work in that the power
than 1 ␮m, such sensors could be fabricated on the casing above densities needed are at least two orders of magnitude greater than
the compressor blade tips. This type of resistor has also been that of conventional size and previous micromachines. Also, the
shown to be usable as an igniter. thermal environment is much harsher. Integrating the electric ma-
chine within the engine offers the advantage of mechanical sim-
Fuel Control Valves. Very small engines are the topic of this plicity in that no additional bearings or structures are required
discussion so the fuel control valves should be equivalently small. over that needed for the fluid machinery. There is also a supply of
If integrated within the engine, the valve design must then be fully cooling air available.
compatible with the fabrication and operating conditions of the Both electric and magnetic machine designs can be considered
gas turbine. This choice strongly constrains the valve design and, to first order, both approaches can yield about equivalent
space. For example, the high processing and operating tempera- power densities. Since the magnetic machines are material
tures prohibit the use of polymers, so a hard valve seat must be property-limited at high temperature and because of the chal-
used. The principal design requirements are flow rate, pressure, lenges of microfabricating magnetic materials 共which are not
frequency response, very low power consumption and leakage, compatible with standard semiconductor manufacturing tech-
and high temperature capability. niques兲, electric designs were first explored. Power density scales
Yang et al. 关57兴 developed MEMS fuel-metering valves for gas- with electric field strength squared, frequency, and rotational
eous hydrocarbon fuels such as propane. The design is a simple speed. The micromachinery of interest here operates at peripheral
silicon spring-mounted plunger opened by electrostatic forces and velocities 1–2 orders of magnitude higher than previously re-
closed by a combination of the spring and fluid pressure forces. ported micromotors, and so yields concomitantly more power.
The electrostatic approach has the advantages that very little Electric machines may be configured in many ways. Here an in-
power is needed to open a valve 共40 nW兲 and the electrical ma- duction design was chosen since it requires neither electrical con-
terials 共polysilicon兲 are compatible with high temperature semi- tact with the rotor nor knowledge of the rotor position.
conductor fabrication technology. Such a valve is shown in Fig. The operation of an electric induction machine can be under-
27. The 2 mm square valve has a 1000 ␮m diameter plunger stood with reference to Fig. 28, 关58兴. The machine consists of two
which rises 3 ␮m off the seat when actuated. An 8-inch wafer of components, a rotor and a stator. The rotor is comprised of a 5–20
valves would contain about 5000 individual units. The valve ␮m thick good insulator covered with a few microns of a poor
opens against 10 atm pressure and, when open, flows 35 sccm of conductor 共200 M⍀ sheet resistivity兲. The stator consists of a set
N2 at a pressure drop of 0.5 atm. Frequency response is several of conductive radial electrodes supported by an insulator. A trav-
hundred hertz. Cyclic testing of the valve has demonstrated a eling electric potential is imposed on the stator electrodes with the
50,000⫹ cycle life for the units tested. This design is on-off. The aid of external electronics. The resulting rotating electric field
actuation-pressure scaling laws favor small valves so the intent is then induces an image charge on the rotor. Depending on the
to use a parallel array of 20 on-off valves, each with a capacity of relative phase between the motion of rotor charges 共set by the
5% of the maximum fuel flow, to meter the fuel. All 20 valves rotor mechanical speed兲 and that of the stator field 共set by the
would consume less than 1 mW total power and operate at tem- external electronics兲, the machine will operate as a motor, genera-
peratures approaching 1000 K so they can be embedded in an tor, or brake. Torque increases with the square of the electric field
engine chip. Many other arrangements are possible, such as loga- strength and frequency. The maximum electric field strength that
rithmic spacing of the valve orifice sizes to give finer fuel flow an air gap can maintain without breakdown is a function of the
control. gap dimension. In air, the breakdown field is a maximum at a gap

220 Õ Vol. 126, APRIL 2004 Transactions of the ASME

 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 28 Fields and charges in a microscale electric induction
motor-generator

of a few microns so that micromachines can potentially realize Fig. 30 A 4-pole stator for a 4 mm dia. magnetic motor-
generator. „Figure courtesy of M. Allen.…
higher power density than large machines of the same design.
Frequency is constrained by external electronics design and by
fabrication constraints on the stator electrode geometry. Current
technology is limited to about 300 volts and 1–2 MHz. This is et al. 关62兴. A four-pole stator for a 4 mm diameter, induction mo-
consistent with a 6 mm diameter machine producing about 10 tor generator, designed to be functionally equivalent to the electric
watts at a 3 ␮m air gap. A 4 mm diameter, six-phase, 131-pole machine in Fig. 29, is shown in Fig. 30. More recent versions of
共786 electrodes, each 4 ␮m wide兲 stator for such a machine is this stator include a laminated magnetic return path to reduce eddy
shown in Fig. 29, 关59兴. Note that such an electric motor generator current losses. The efficiency of this magnetic generator is calcu-
occupies less than 20 ␮m thickness at the surface of the rotor and lated to approach 60%. It has the additional advantage that its
stator 共mainly the insulator thickness兲. Thus the power density of external low-frequency low-voltage electronics are easier to engi-
this machine 共excluding the external electronics兲 is many times neer and more compact than the high-frequency high-voltage elec-
that of a conventional magnetic motor generator, on the order of tronics of electric machines. Current stator materials are limited to
100 MW/m3. Fréchette et al. 关60兴 have reported a similar design only 500 K, however. Advanced materials may increase the oper-
run as an electric motor. The torque produced by these devices has ating temperature to 800 K. The rotor requires several hundred
agreed with theoretical predictions but high power operation has microns of iron for the magnetic path, which presents structural
yet to be reported. design challenges at 1,000,000 rpm. Magnetic machines will re-
To maximize power output, induction machines such as these quire careful thermal management when embedded within a
require the spacing between the rotor and stator to be on the order MEMS gas turbine engine.
of the stator pitch. The electrical torque produced scales with the
square of the rotor-stator spacing, a few microns in this case. Engine Design Trades, Component Integration, and De-
However, in these high-speed machines, the rotor periphery is at sign Evolution
sonic velocity so the viscous drag in a gap of only 2–3 microns is Gas turbine engines are more than a set of components bolted
extremely large. Indeed, this drag is the major loss mechanism for together. Rather, a successful gas turbine is a highly integrated
such an electrical machine. Thus, there is a basic design tradeoff system engineered to meet specific requirements, often with artful
for the electric motor generator between power density and effi- compromises between conflicting demands of the fluid, thermo,
ciency. While it may be possible to alter the local geometry to structural, and manufacturing engineering. A MEMS gas turbine is
reduce the drag somewhat, 关24兴, the drag still makes up about half no different in this respect. The two dominant design consider-
the total loss and limits the efficiency 共shaft to net electrical兲 of ations are the fabrication complexity and the thermodynamic
these designs to 40–50%. cycle requirements. The principal challenge is to arrive at a design
A magnetic induction machine has many fewer poles so that the which meets the thermodynamic and component functional re-
optimum rotor-stator spacing is much larger 共30–50 ␮m兲 and the quirements while staying within the realm of realizable microma-
drag concomitantly lower. Koser and Lang 关61兴 designed such a chining technology.
machine based on the microplating technology developed by Park For any gas turbine, maximizing net engine output power in-
cludes maximizing both power per unit mass flow 共specific
power兲 and mass flow. Specific power is sensitive to component
efficiency and pressure ratio, especially at the low-pressure ratios
under consideration here, 2:1 to 4:1. High-pressure ratio in a
single-stage centrifugal machine implies high wheel speeds; 500
m/s peripheral speed was chosen as the maximum compatible
with geometrically simple Si construction. This also requires bear-
ings and rotor dynamics capable of such high-speed operation.
High efficiency implies optimal airfoil design and tight clear-
ances. Tight clearances, in turn, imply high manufacturing preci-
sion, careful design for centrifugal and thermal growth, and ro-
bustness to the occasional high-speed rub. The other key to high
specific power is high turbine inlet temperature 共TIT兲. This re-
quires high temperature materials for the turbine or cooling or
both. High mass flows require high through-flow Mach numbers
Fig. 29 A 131-pole, 6-phase, 4 mm dia. electric induction sta- and large flow areas. With these requirements in mind, let’s ex-
tor amine how these trades influence the design of a MEMS gas tur-

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 221
bine. Specifically, the next section discusses why the demo engine ing’’ 共it is really a film insulation兲 is the most effective cooling
design in Figs. 3 and 4 is so configured. The principal geometric technique here since, in this case, the heat is prevented from en-
constraint on the design is that imposed by the current state of the tering the solid.
art in precision 共micron-level accuracy兲 etch-depth capability of As with large engines, improving performance at the MEMS
300–500 ␮m. We will start with the turbomachinery. scale will require some combination of more complex designs,
The engine design of Fig. 3 requires six wafers: two form the better materials, and improved manufacturing technology. For ex-
rotor 共and the annular combustor volume兲, and two are needed on ample, cycle pressure ratios much over 4:1–5:1 will require mul-
each side of the rotor for the hydrostatic thrust bearings and their tiple spools 共probably not concentric兲. Higher turbine inlet tem-
associated plumbing. Replacing the hydrostatic bearings with hy- peratures will require improved cooling and fabrication
drodynamic ones would eliminate one wafer. Adding a generator technology for such materials as silicon carbide, silicon nitride,
mounted on a rotating compressor tip shroud would add one wa- and sapphire. Improved propulsive efficiency can be realized with
fer. Adding control valves would add an additional wafer above bypass engine designs. Better compression system performance
the compressor. Such a complete, self-contained 共other than elec- will require more attention to thermal isolation. These are all

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

tronics兲 gas turbine generator could thus be built with seven or physically possible but will require considerable investment in
eight wafers, not including external electronics. both the disciplinary and microfabrication technologies.
In a centrifugal compressor and turbine of geometry similar to
those in Figs. 10 and 14, turning at 400–500 m/s tip speed, tur-
bomachine design is constrained to a blade span of 400–500 ␮m
by fabrication technology and bending stress at the blade roots.
Ongoing Technical Developments
Given the blade height, air flow increases with rotor diameter. As of this writing, no one has yet reported an operating gas
Rotor diameter is constrained by the number of engines required turbine engine at this scale. Many of the turbomachinery efforts
per wafer and by etching uniformity 共if a design goal is to avoid appear to be currently wrestling with rotor dynamics issues. At
the necessity of dynamically balancing the rotor兲. Etch uniformity MIT, the requisite component technologies for the H2 demo en-
scales inversely with rotor area and blade height 共i.e., mass flow is gine of Figs. 3 and 4 have been demonstrated at the component
constrained by etch uniformity technology兲. Another factor influ- level—turbine, bearings, combustor, etc.—but a complete device
encing uniformity is rotor/stator airfoil count. Etch rate is a func- fabricated to the design specification has yet to be tested. The
tion of the local geometry, so for best circumferential uniformity challenge appears to be one mainly of stringent process control in
there should be an even multiple of rotor and stator airfoils. This, a long, complex fabrication sequence, which is different in detail
of course, is deleterious to long-term vibrational life for the air- but in many ways as challenging as that for a microprocessor.
foils. If the rotor is fabricated from more then one wafer and Meanwhile, work is ongoing on improving component aerody-
etched with multiple masks, then mask and wafer alignment is an namic performance, hydrocarbon combustors, thermal isolation,
issue which favors the fewest possible wafers and masks. Rotor high temperature materials 共SiC兲, bearing system robustness, and
diameters up to 10 mm with 400–500 ␮m blade spans, fabricated electric and magnetic generators.
from 1–3 wafers, are consistent with current capabilities. MEMS gas turbines differ not only in design and manufacture
Bearing placement is another first-order concern. The bearings from conventional engines, they differ in development process as
must support the load and remain in their stable operating regime. well. Specifically, since the engines are monolithic blocks of sili-
The load capability of air bearings scales with the bearing area. con, they cannot be disassembled, reworked, and then reas-
The bearing length is constrained by current etching technology to sembled during development. Instead, a new engine must be built
300–500 ␮m 共and widths of 10–25 ␮m兲, so load capacity scales from scratch. This is a process which takes about three months
with bearing diameter. For hydrostatic journal bearings of con- since even a seemingly minor change, for example increasing
stant length, the optimum width of the bearing for maximum bearing clearance, may require some process development. Con-
stable operating speed scales inversely with diameter. So as the ceptually, the parallelism inherent in the MEMS manufacture
bearing diameter is reduced, the air required to operate the bearing could be an advantage in development. For example, if the opti-
goes up 共since the flow per unit area grows faster than the area mum clearance was unknown, a wafer could be built with engines
shrinks兲 and the load capacity goes down. This implies that bear- of various clearances. To date, this approach has not proven pro-
ing diameter should be maximized, i.e., on the rotor diameter. ductive at MIT since the majority of the problems encountered
Placing the bearing here also eliminates the need for a seal on the have largely been unanticipated or attributable to inadequate pro-
rotor, since in this case the leakage air is the air bearing fluid, but cess control.
this approach increases bearing drag. Future developments can be considered in three broad catego-
The cycle needs turbine inlet temperatures of at least 1400– ries: evolution of MEMS gas turbines in both technology and
1600 K to meet the application goals. These relatively high tem- application, other approaches to gas turbine engines in the
peratures are needed due to the relatively high losses in the com- centimeter-size range, and additional applications for millimeter-
ponents and the secondary systems such as bearings. For example, to-centimeter-sized devices based on the technologies discussed
the shaft mechanical losses are 4 – 8% of the shaft work for the above.
The current MEMS engine design has a projected performance
design in Figs. 3 and 4, ten times that of engines in the 5,000–
level comparable with gas turbines of the 1940s 共Fig. 31兲. At these
30,000 lb thrust range. These temperatures are above the 900–950
low-cycle pressure ratios, small improvements in pressure ratio
K maximum operating temperature of a single-crystal Si rotor. In
and component efficiency have disproportionate returns of effi-
the engine design of Fig. 3, the turbine rotor is conductively
ciency and output power 共Fig. 32兲. So it is likely that the perfor-
cooled through the compressor disk to the compressor air, which
mance of these devices can be increased, perhaps up to that of the
has a quite deleterious effect on the cycle performance and thus on
1950s. Some of the improvements will come from component
the net engine output. Improving performance requires thermally
evolution, others from new configurations or more complex
isolating the turbine from the compressor which can be accom-
cycles. The approaches of recuperated engines and combined
plished by hollowing the shaft between them 共100–200 ␮m of the
cycles have been explored.
turbine disk in Fig. 14兲. When this is done, the temperature capa-
bility of the turbine rotor must be increased. As for large engines, Combined Cycles. Cycles with heat exchangers are attractive
some combination of increased material capability and improved at microscale because heat exchangers scale favorably as size is
cooling is needed. Improved material capability can be provided reduced. MEMS heat exchangers have the advantage of high ef-
by reinforcing the Si turbine disk with SiC, with the temperature fectiveness and their repetitive structure is readily producible by
capability increasing with fraction of SiC. So-called ‘‘film cool- microfabrication techniques. A micrograph of a high-temperature

222 Õ Vol. 126, APRIL 2004 Transactions of the ASME

 Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025
Fig. 31 Specific core power versus turbine rotor inlet tempera-
ture „after Koff…

heat exchanger is shown in Fig. 33, 关63兴. There is now a large

literature on MEMS heat exchangers, including both single and Fig. 33 Micron-scale counterflow heat exchanger. „Figure
multiple-phase devices, and their behavior, 关64兴. courtesy of J. Brisson.…
Given the high temperatures and poor efficiencies of these very
small gas turbines, their exhaust contains relatively large amounts
of high quality heat. Thus, performance is greatly improved when manufacturing cost. However, there are many low-cost precision
this waste heat is productively utilized. Calculations suggest that manufacturing technologies for small parts and a need such as this
both MEMS-combined cycles and cogeneration cycles are fea- may foster new ones.
sible. One such steam-combined cycle system increases the net
system output by about 50% over that of the gas turbine alone, Additional Devices. The technology needed for a MEMS gas
关65兴. Cogen systems providing cooling or condensing water are turbine is based on microscale high-speed rotating machinery in-
obvious applications when the gas turbine is used for portable cluding disciplinary foundations in aerodynamics, combustion,
power generation. materials, electromechanics, controls, and bearings. The same
technologies can be applied to other microscale systems. One ex-
Alternate Approaches. Several teams are working to build ample is a motor-driven air blower or compressor, 关60兴. Another
gas turbines in the few grams/sec mass flow range using a wider example is a micro-bipropellant liquid rocket motor which is un-
set of manufacturing technologies than MEMS, including conven- der development, 关68兴. This propulsion system includes a regen-
tional metal-forming techniques, 关37,66,67兴 and mold shape depo- eratively cooled silicon thrust chamber and nozzle 共Fig. 34兲, tur-
sition manufacture, 关28兴. Such an approach has its own set of bopumps 共Fig. 35兲, and liquid control valves, 关69,70兴. The
advantages and disadvantages. One major advantage is the lack of propellant flow rate is about 5 g/s. The silicon thrust chamber is
an upper size constraint imposed by microfabrication technology, designed to operate at chamber pressures of 125 atm and tempera-
so that engines in the hundreds of watts may be feasible. Its prin- tures of 3000 K. The pressure in the liquid cooling jacket around
cipal challenge compared to the MEMS approach may be one of the combustion chamber is above 200 atm. These devices operate
at much higher power densities than those of the current gas tur-
bine components and demonstrate that very high pressure, high
power density silicon structures and devices are feasible. Prelimi-
nary tests of the cooled thrust chamber and turbopump are prom-
ising. In support of the engineering design of these devices, fun-
damental studies have been conducted on cavitation in
micropumps, 关71兴, and the cooling behavior of a variety of rocket
propellants in microgeometries, 关72兴.

Fig. 32 Simple cycle performance variation for low pressure

ratios, ␲ „␩Äadiabatic efficiency, T t 4 Äturbine inlet temperature,
T t 2 Äcompressor inlet temperature…. „Figure courtesy of M.
Monroe.… Fig. 34 A 15 N „3.3 lb… thrust bipropellant liquid rocket engine

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 223
 Table 3 Harnessing available energy

Practical Whr/kg**
Energy Potential* Based on Conversion
Source Whr/kg Device Efficiency
Lithium battery 1400 175 (LiSO2 )
300 (LiSOCl2 )
TNT 1400 NA
Methanol 6200 1500–3100
Diesel fuel 13,200 1320–5000
Hydrogen 33,000 1150–23,000

*Based on enthalpy
**共System conversion efficiency兲⫻共Energy available in fuel兲

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

environments. Highly redundant, distributed power for actuation,
Fig. 35 A 2.5 gÕsec turbopump rotor „the pump is the inner sensors, etc., may be attractive in many applications.
blade row, the turbine the outer… Competing with a 50– 60% efficient central power plant re-
quires much higher performance. Even including cogen use of the
waste heat, a local gas turbine must have an efficiency of 20–
Pacing Technology. Microfabrication is the pacing technol- 30%, a level which looks many years away at this time. Should
ogy in the development of MEMS gas turbine engines. Fortu- these levels be approached, then the redundancy and extreme qui-
nately, the rate of progress and innovation in MEMS fabrication etness and compactness may make an array of millimeter-scale
dwarfs those of the more mature technologies familiar to gas tur- machines an attractive solution. Emergency power applications
bine developers. Much of this progress stems from the wide prom- would not require as high performance as base power and can be
ise that MEMS offers for a variety of applications, justifying large an attractive application for these small machines if the capital
research investments across a broad set of approaches, 关73兴. In- costs are sufficiently low.
deed, several fabrication limitations that existed at the start of this In addition to compactness and redundancy, one advantage that
effort in the mid 1990s have since disappeared. Thus, many of the millimeter-scale gas turbines do offer for many applications is that
engineering design constraints now attributable to the microfabri- they are very, very quiet, even in large arrays. This stems from
cation process limitations are likely be relaxed as new technology their high frequencies 共blade passing is several hundred kilohertz,
is developed. beyond the audible range兲 and short length scales 共millimeter-
diameter exhaust jets, which mix rapidly兲. The high frequency
Economics and the Future of Millimeter-Scale Gas Tur- sound that is produced is relatively easy to muffle and quick to
attenuate naturally.
bines Propulsion is an obvious application for very small engines
Whittle and Von Ohain were successful in their early jet engine since cubed-square scaling means that they can have very high
developments because they set engineering goals based on the thrust-to-weight ratios and be extremely compact. The U.S. De-
requirement of going fast, significantly faster than the piston- fense Department is investing in reconnaissance airplanes with
powered aircraft of the day, 关74兴. In contrast, the gas turbine de- gross takeoff weights as low as 50–100 g. These aircraft have
velopers who based their requirements on competing on range lift-to-drag ratios on the order of 5 so that an 0.1 N class thrust
with piston-powered propeller aircraft were unsuccessful at that MEMS gas turbine such as that in Fig. 3 is an attractive power
time because their goal required much higher levels of perfor- plant. It is much better than a battery-powered electric solution,
mance, performance that was many years away. In this sense, we
can ask what is required to make millimeter-scale gas turbines
real—real in that machines are in production and making a dis-
cernable impact on society. There are two answers, one technical,
the other economic. Many of the detailed technical issues were
discussed above, but some must be considered in concert with the
economics of power production. In the broadest sense, to be more
than a curiosity, these very small gas turbine engines must fulfill
societal needs. What are the possible applications and what levels
of performance are needed for each? The answers must consider
the alternative engineering solutions to each potential application.
The two major applications of these very small gas turbines
mirror those of large engines, power production and aeronautical
propulsion. Power production in the short term is aimed at por-
table applications where the very much larger energy density of a
hydrocarbon fuel compared to a lithium battery chemistry means
that even a very inefficient gas turbine can be attractive 共Table 3兲.
In the short term, a 5–10% overall system efficiency 共chemical to
net electric power output兲 is sufficient to make a gas turbine en-
gine solution an attractive alternative to a battery. Figure 36 is a
concept of an engine packaged in the form factor of a high per-
formance 50 W military battery 共which costs about $100兲. Most of
the volume is fuel which implies that, in this application, specific
fuel consumption is more important than specific power 共since
multiple engines could be packaged together兲.
A somewhat different compact power application is auxiliary Fig. 36 Concept of a MEMS gas turbine engine packaged as a
power for flight vehicles, especially ones with high temperature standard military battery

224 Õ Vol. 126, APRIL 2004 Transactions of the ASME

and IC engines in production are ten times too large, very noisy, Acknowledgments
and have poor fuel consumption. For subsonic flight applications,
Much of the work summarized herein is the intellectual accom-
thrust and propulsive efficiency can be increased by adopting by-
plishment of a team at MIT including current and former research
pass engine configurations. High-speed flight may be another at- engineers and faculty—A. Ayon, K. Breuer, J. Brisson, F. Ehrich,
tractive application. Here the intrinsic high temperature capability R. Ghodssi, Y. Gong, S. Jacobson, R. Khanna, J. Lang, H. Li, C.
共900 K兲 of even a silicon engine cold section means that the Livermore, Y. Peles, M. Schmidt, S. Senturia, Z. Spakovszky, M.
propulsion system can operate at high ambient temperatures. Con- Spearing, C. Tan, S. Umans, I. Waitz, X. Zhang—and a large
figurations other than a gas turbine may also be realized, such as number of extraordinarily talented and hardworking graduate stu-
a ramjet or pulse detonation engine. Independent of the speed dents, post-docs, and technicians. D. Park prepared the manuscript
regime, clearly several engines can be used to realize increased 共and many, many others兲. It has been a great pleasure working
thrust levels for larger vehicles. How large a thrust level is a with them all.
function of how many engines can be practically assembled to- Special thanks goes to Drs. R. Paur and T. Doligalski of the US
gether 共large phased array radars may use 105 modules兲 and the

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

Army Research Office who were among the first to recognized the
performance of the millimeter-scale engines relative to that of potential value of this technology and have nurtured it over the
large engines. Since it is unlikely that engines of this size will years.
approach the fuel economy of large engines, their use may be The author would also like to thank Dr. J. L. Kerrebrock for 35
restricted to short duration applications use such as lift engines, years as a mentor, colleague, and friend.
where very high thrust-to-weight ratio, compactness, and redun- The research at MIT cited herein has been supported by the
dancy may command a premium. U.S. Army and DARPA.
Both gas turbine engines and complex semiconductor devices
are very expensive to develop and there is no reason to believe
that a marriage of the two will prove substantially less so. Devel-
opment costs can be amortized over many units. Manufacturing References
cost, however, is always a major issue for mass-produced devices. 关1兴 Gerendas, M., and Pfister, R., 2000, ‘‘Development of a Very Small Aero-
Engine,’’ ASME Paper 2000-GT-0536.
There is vast experience on the production cost of semiconductor 关2兴 Nakajima, T., Fukikawa, Y., Goto, T., and Iio, M., 1995, ‘‘The Development of
wafers in large-scale production, much less on MEMS. For ex- the Micro Gas Turbine Generator,’’ 1995 Yokohama International Gas Turbine
ample, a standard 200 mm diameter silicon CMOS wafer from a Congress, Yokohama, Japan.
foundry may cost $600–1000 in large-scale production quantities 关3兴 Senturia, S., 2001, Microsystem Design, Kluwer Academic Publishers, Boston,
MA.
共the industry is currently moving to 300 mm diameter wafers to 关4兴 Jensen, K. F., 2001, ‘‘Microreaction Engineering—Is Small Better?’’ Chem.
reduce costs兲. A specialized wafer produced in small lots 共such as Eng. Sci., 56, pp. 293–303.
GaA’s used for radar and RF兲 may be ten times more expensive. 关5兴 Epstein, A. H., and Senturia, S. D., 1997, ‘‘Macro Power From Micro Machin-
ery,’’ Science, 276, p. 1211.
The cost per chip is also a function of the number of chips per 关6兴 Epstein, A. H., Senturia, S. D., Al-Midani, O., Anathasuresh, G., Ayón, A.,
wafer 共i.e., the size of the chip兲. Thus, a geometrically large, state- Breuer, K., Chen, K.-S., Ehrich, F. E., Esteve, E., Fréchette, L., Gauba, G.,
of-the-art microprocessor chip may cost $200 to manufacture, Ghodssi, R., Groshenry, C., Jacobson, S., Kerrebrock, J. L., Lang, J. H., Lin,
while a small watch chip is only tens of cents. From this we infer C.-C., London, A., Lopata, J., Mehra, A., Mur Miranda, J. O., Nagle, S., Orr,
D. J., Piekos, E., Schmidt, M. A., Shirley, G., Spearing, S. M., Tan, C. S.,
that a wafer of MEMS gas turbines might produce several kilo- Tzeng, Y.-S., and Waitz, I. A., 1997, ‘‘Micro-Heat Engines, Gas Turbines, and
watts of power and cost several thousand dollar to manufacture, Rocket Engines,’’ Paper No. AIAA 97-1773.
putting the specific cost in the $0.5–5 per watt range. On the low 关7兴 Groshenry, C., 1995, ‘‘Preliminary Design Study of a Micro-Gas Turbine En-
end this cost approaches that of large power plants. gine,’’ M.S. thesis, MIT Department of Aeronautics and Astronautics, MIT,
Cambridge, MA.
If the more optimistic projections of their price and perfor- 关8兴 Park, D.-E., Lee, D.-H., Yoon, J.-B., Kwon, S., and Yoon, E., 2002, ‘‘Design
mance are achieved, then it is possible that these millimeter-scale and Fabrication of Micromachined Internal Combustion Engine as a Power
machines and the technologies on which they are based may begin Source for Microsystems,’’ Technical Digest of MEMS 2002, 15th IEEE Inter-
to transform the mechanical infrastructure of society in a manner national Conference on Micro Electro Mechanical Systems, Las Vegas, NV,
IEEE, Piscataway, NJ, pp. 272–275.
similar to the changes in information handling wrought by the 关9兴 Jacobson, S. A., 1998, ‘‘Aerothermal Challenges in the Design of a Microfab-
microprocessor. They will augment and even displace large cen- ricated Gas Turbine Engine,’’ AIAA 98-2545, 29th AIAA Fluid Dynamics
tral facilities. Distributed power, water treatment, and chemical Conference, Albuquerque, NM.
processing may be realized, providing a more robust society. 关10兴 Spearing, M. S., 2000, ‘‘Materials Issues in Microelectro-Mechanical Systems
共MEMS兲,’’ Acta Mater., 48, pp. 179–196.
Many of the more significant changes this technology will engen- 关11兴 Hawthorne, W. R., 1994, ‘‘Reflections on United Kingdom Aircraft Gas Tur-
der are likely unrecognized at this time. bine History,’’ ASME J. Eng. Gas Turbines Power, 116, pp. 495–510.
关12兴 Madou, M., 1997, Fundamentals of Microfabrication, CRC Press, Boca Raton,
FL.
关13兴 Ayón, A. A., Lin, C. C., Braff, R., Bayt, R., Sawin, H. H., and Schmidt, M.,
1998, ‘‘Etching Characteristics and Profile Control in a Time Multiplexed
Inductively Coupled Plasma Etcher,’’ 1998 Solid State Sensors and Actuator
Workshop, Hilton Head, SC.
Concluding Remarks 关14兴 Lin, C. C., 1999, ‘‘Development of a Microfabricated Turbine-Driven Air
Bearing Rig,’’ Ph.D. thesis, MIT Department of Mechanical Engineering, MIT,
Over the past eight years, research has started on very small gas Cambridge, MA.
turbine engines and related technologies. It is quite clear from the 关15兴 Lin, C. C., Ghodssi, R., Ayón, A. A., Chen, D. Z., Jacobson, S., Breuer, K. S.,
work to date that millimeter-scale gas turbine engines and similar Epstein, A. H., and Schmidt, M. A., 1999, ‘‘Fabrication and Characterization
high-speed rotating machinery, combustion, and energy conver- of a Micro Turbine/Bearing Rig,’’ Presented at MEMS ’99, Orlando, FL.
关16兴 Mirza, A. R., and Ayón, A. A., 1999, ‘‘Silicon Wafer Bonding for MEMS
sion systems are technically feasible. One approach to realizing Manufacturing,’’ Solid State Technol., 42, pp. 73–78.
devices at these scales utilizes the semiconductor industry-derived 关17兴 Miki, N., Zhang, X., Khanna, R., Ayón, A. A., Ward, D., and Spearing, S. M.,
micromachining technology known as MEMS. The economic im- 2002, ‘‘A Study of Multi-Stack Silicon-Direct Wafer Bonding for MEMS
pact of these devices will be dependent on the performance levels Manufacturing,’’ Fifteenth IEEE International Conference on Micro Electro
Mechanical Systems (MEMS2002), Las Vegas, NV, IEEE, Piscataway, NJ, pp.
and the manufacturing costs, both of which have yet to be proven. 407– 410.
It is certainly possible, however, that MEMS gas turbines may one 关18兴 Mehra, A., Ayón, A. A., Waitz, I. A., and Schmidt, M. A., 1999, ‘‘Microfabri-
day be competitive with conventional machines in a cost per in- cation of High Temperature Silicon Devices Using Wafer Bonding and Deep
Reactive Ion Etching,’’ J. Microelectromech. Syst., 8, pp. 152–160.
stalled kilowatt. Even at much higher costs, they will be very 关19兴 Fréchette, L. G., Jacobson, S. A., Breuer, K. S., Ehrich, F. F., Ghodssi, R.,
useful as compact power sources for portable electronics, equip- Khanna, R., Wong, C. W., Zhang, X., Schmidt, M. A., and Epstein, A. H.,
ment, and small vehicles. 2000, ‘‘Demonstration of a Microfabricated High-Speed Turbine Supported on

Journal of Engineering for Gas Turbines and Power APRIL 2004, Vol. 126 Õ 225
 Gas Bearings,’’ Hilton Head Solid-State Sensor & Actuator Workshop, Hilton Acting Thrust Bearing for High Speed Micro-Rotors,’’ Technical Digest of
Head Island, SC, pp. 43– 47. MEMS 2002, 15th IEEE International Conference on Micro Electro Mechani-
关20兴 Ghodssi, R., 2003, private communication. cal Systems, Las Vegas, NV, IEEE, Piscataway, NJ, pp. 276 –279.
关21兴 Mehra, A., Jacobson, S. A., Tan, C. S., and Epstein, A. H., 1998, ‘‘Aerody- 关47兴 Chen, K.-S., Ayón, A., and Spearing, S. M., 1999, ‘‘Controlling and Testing the
namic Design Considerations for the Turbomachinery of a Micro Gas Turbine Fracture Strength of Silicon at the Mesoscale,’’ to be published in the J. Am.
Engine,’’ Presented at the 25th National and 1st International Conference on Ceram. Soc. .
Fluid Mechanics and Power, New Delhi, India. 关48兴 Moon, H.-S., Anand, L., and Spearing, S. M., 2002, ‘‘A Constitutive Model for
关22兴 Mehra, A., 1997, ‘‘Computational Investigation and Design of Low Reynolds the Mechanical Behavior of Single Crystal Silicon at Elevated Temperature,’’
Number Micro-Turbomachinery,’’ M.S. thesis, MIT Department of Aeronau- Mater. Res. Soc. Symp. Proc., Materials Research Society, Warrendale, PA,
tics and Astronautics, MIT, Cambridge, MA. 687, Paper No. B9.6.
关23兴 Shirley, G., 1998, ‘‘An Experimental Investigation of a Low Reynolds Num- 关49兴 Choi, D., Shinavski, R. J., Steffier, W. S., Hoyt, S., and Spearing, S. M., 2001,
ber, High Mach Number Centrifugal Compressor,’’ M.S. thesis, MIT Depart- ‘‘Process Development of Silicon-Silicon Carbide Hybrid Micro-Engine Struc-
ment of Aeronautics and Astronautics, MIT, Cambridge, MA. tures,’’ Materials Research Society Symposium Proceedings, Materials Re-
关24兴 Fréchette, L. G., 2000, ‘‘Development of a Microfabricated Silicon Motor- search Society, Warrendale, PA, 687, Paper No. B5.44.
Driven Compression System,’’ Ph.D. thesis, MIT Department of Aeronautics 关50兴 Sugimoto, S., Tanaka, S., Li, J.-F., Watanabe, R., and Esashi, M., 2000, ‘‘Sili-
and Astronautics, MIT, Cambridge, MA. con Carbnide Micro-Reaction-Sintering Using a Multilayer Silicon Mold,’’

Downloaded from [Link] by Vardhaman College Of Engineering user on 25 January 2025

关25兴 Sirakov, B., 2003, private communication. IEEE Transducers, pp. 775–780.
关26兴 Isomura, K., Murayama, M., Yamaguchi, H., Ijichi, N., Asakura, H., Saji, N., 关51兴 Spearing, S. M., and Chen, K. S., 1997, ‘‘Micro-Gas Turbine Engine Materials
Shiga, O., Takahashi, K., Kawakubo, T., Watanabe, T., Yamagata, A., Tange, and Structures,’’ Presented at 21st Annual Cocoa Beach Conference and Ex-
H., Tanaka, S., Genda, T., and Esashi, M., 2002, ‘‘Design Study of a Micro- position on Composite, Advanced Ceramics, Materials and Structures.
machined Gas Turbine With Three-Dimensional Impeller,’’ Presented at the 关52兴 Chen, K.-S., Spearing, S. M., and Nemeth, N. N., 2001, ‘‘Structural Design of
9th International Symposium on Transport Phenomena and Dynamics of Ro- a Silicon Micro-Turbo-Generator,’’ AIAA J., 39, pp. 720–728.
tating Machinery, Honolulu, HI. 关53兴 Chen, K.-S., Ayón, A. A., Lohner, K. A., Kepets, M. A., Melconian, T. K., and
关27兴 Isomura, K., Murayama, M., Yamaguchi, H., Ijichi, N., Asakura, H., Saji, N., Spearing, S. M., 1998, ‘‘Dependence of Silicon Fracture Strength and Surface
Shiga, O., Takahashi, K., Tanaka, S., Genda, T., and Esashi, M., 2002, ‘‘De- Morphology on Deep Reactive Ion Etching Parameters,’’ Presented at the MRS
velopment of Microturbocharger and Microcombustor for a Three- fall Meeting, Boston, MA.
Dimensional Gas Turbine at Microscale,’’ ASME Paper GT-2002-30580. 关54兴 Harrison, T. S., London, A. P., and Spearing, S. M., 2001, ‘‘High Temperature,
关28兴 Kang, S., Johnston, J. P., Arima, T., Matsunaga, M., Tsuru, H., and Prinz, F. B., High Pressure Fluid Connections for Power Micro-Systems,’’ Mat. Res. Soc.
2003, ‘‘Micro-Scale Radial-Flow Compressor Impeller Made of Silicon Ni- Symp. Proc., Materials Research Society, Warrendale, PA, 657, p. 654.
tride, Manufacturing and Performance,’’ ASME Paper GT-2003-38933. 关55兴 Liu, C., 2000, ‘‘Dynamical System Modeling of a Micro Gas Turbine Engine,’’
关29兴 Gong, Y., 2002, private communication. M.S. thesis, MIT Department of Aeronautics and Astronautics, MIT, Cam-
关30兴 Philippon, B., 2001, ‘‘Design of a Film Cooled MEMS Micro Turbine,’’ M.S. bridge, MA.
thesis, MIT Department of Aeronautics and Astronautics, MIT, Cambridge, 关56兴 Tang, D., 2001, ‘‘Rotor Speed Microsensor for the MIT Microengine,’’ M.S.
MA. thesis, MIT Department of Mechanical Engineering, MIT, Cambridge, MA.
关31兴 Spadaccini, C. M., Mehra, A., Lee, J., Lukachko, S., Zhang, X., and Waitz, I. 关57兴 Yang, X., Holke, A., and Schmidt, M. A., 2002, ‘‘An Electrostatic, On/Off
A., 2002, ‘‘High Power Density Silicon Combustion Systems for Micro Gas MEMS Valve for Gas Fuel Delivery of a Microengine,’’ Solid-State Sensor,
Turbine Engines,’’ ASME Paper GT-2002-30082. Actuator and Microsystems Workshop, Hilton Head Island, SC.
关32兴 Waitz, I. A., Gautam, G., and Tzeng, Y.-S., 1998, ‘‘Combustors for Micro-Gas 关58兴 Nagle, S. F., and Lang, J. H., 1999, ‘‘A Micro-Scale Electric-Induction Ma-
Turbine Engines,’’ ASME J. Fluids Eng., 120, 109-117. chine for a Micro Gas Turbine Generator,’’ 27th Annual Meeting of the Elec-
关33兴 Mehra, A., Zhang, X., Ayón, A. A., Waitz, I. A., Schmidt, M. A., and Spadac- trostatics Society of America.
cini, C. M., 2000, ‘‘A 6-Wafer Combuston System for a Silicon Micro Gas 关59兴 Ghodssi, R., Fréchette, L. G., Nagle, S. F., Zhang, X., Ayón, A. A., Senturia, S.
Turbine Engine,’’ J. Microelectromech. Syst., 9, pp. 517–527. D., and Schmidt, M. A., 1999, ‘‘Thick Buried Oxide in Silicon 共TBOS兲: An
关34兴 Zhang, X., Mehra, A., Ayón, A. A., and Waitz, I. A., 2002, ‘‘Development of Integrated Fabrication Technology for Multi-Stack Wafer-Bonded MEMS Pro-
Polysilicon Igniters and Temperature Sensors for a Micro Gas Turbine En- cesses,’’ Transducers 99, IEEE, Piscataway, NJ, pp. 1456-1459.
gine,’’ Technical Digest of MEMS 2002, 15th IEEE International Conference 关60兴 Fréchette, L. G., Nagle, S. F., Ghodssi, R., Umans, S. D., Schmidt, M. A., and
on Micro Electro Mechanical Systems, Las Vegas, NV, IEEE, Piscataway, NJ, Lang, J. H., 2001, ‘‘An Electrostatic Induction Micromotor Supported on Gas-
pp. 280–283. Lubricated Bearings,’’ IEEE 14th International Micro Electro Mechanical Sys-
关35兴 Spadaccini, C. M., Zhang, X., Cadou, C. P., Miki, N., and Waitz, I. A., 2002, tems Conference, MEMS 2001, Interlaken, Switzerland.
‘‘Development of a Catalytic Silicon Micro-Combustor for Hydrocarbon- 关61兴 Koser, H., and Lang, J. H., 2000, ‘‘Modeling a High Power Density MEMS
Fueled Power MEMS,’’ Technical Digest of MEMS 2002, 15th IEEE Interna- Magnetic Induction Machine,’’ Proc. Fourth International Conference On
tional Conference on Micro Electro Mechanical Systems, Las Vegas, NV, Modeling and Simulation of Microsystems, Hilton Head, SC, Transducer Re-
IEEE, Piscataway, NJ, pp. 228 –231. search Foundation, Cleveland, OH, pp. 286 –289.
关36兴 Takahashi, K., Murayama, M., Isomura, K., and Tanaka, S., 2002, ‘‘Develop- 关62兴 Park, J. W., Cros, F., and Allen, M. G., 2002, ‘‘A Sacrificial Layer Approach to
ment of a Methane Fueled Combustor for Micro-Scaled Gas Turbine,’’ Tech- Highly Laminated Magnetic Cores,’’ MEMS 2002 IEEE International Confer-
nical Digest of Power MEMS 2002, Tsukuba, Japan, IEEE, Piscataway, NJ, pp. ence, Las Vegas, NV, IEEE, Piscataway, NJ, pp. 380–383.
44 – 46. 关63兴 Sullivan, S., Zhang, X., Ayón, A. A., and Brisson, J. G., 2001, ‘‘Demonstration
关37兴 Matsuo, E., Yoshiki, H., Nagashima, T., and Kato, C., 2002, ‘‘Development of of a Microscale Heat Exchanger for a Silicon Micro Gas Turbine Engine,’’
Ultra Micro Gas Turbines,’’ Technical Digest of Power MEMS 2002, Tsukuba, Transducers 01, IEEE Piscataway, NJ, pp. 1606 –1609.
Japan, IEEE, Piscataway, NJ, pp. 36 –39. 关64兴 Harris, C., Kelly, K., Wang, T., McCandless, A., and Motakef, S., 2002, ‘‘Fab-
关38兴 Mur Miranda, J. O., 1997, ‘‘Feasibility of Electrostatic Bearings for Micro rication, Modeling, and Testing of Micro-Cross-Flow Heat Exchangers,’’ J.
Turbo Machinery,’’ M. Eng. thesis, MIT Department of Electrical Engineering Microelectromech. Syst., 11, pp. 726 –735.
and Computer Science, MIT, Cambridge, MA. 关65兴 Cui, L., 2003, ‘‘Design and System Analysis of Micro-Scale Rankine Cycle
关39兴 Walton, J. F., and Hesmat, H., 2002, ‘‘Application of Foil Bearings to Turbo- Power Systems,’’ M.S. thesis, MIT Department of Aeronautics and Astronau-
machinery Including Vertical Operation,’’ ASME J. Eng. Gas Turbine Power, tics, MIT, Cambridge, MA.
124, pp. 1032–1039. 关66兴 Isomura, K., Murayama, M., and Kawakubo, T., 2001, ‘‘Feasibility Study of a
关40兴 Spakovszky, Z. S., 2003, ‘‘Scaling Laws for Ultra-Short Hydrostatic Gas Jour- Gas Turbine at Micro Scale,’’ ASME Paper 2001-GT-101.
nal Bearings,’’ 19th Biennial Conference on Mechanical Vibration and Noise, 关67兴 Isomura, K., Murayama, M., Yamaguchi, H., Ijichi, N., Saji, N., Shiga, O.,
ASME Paper No. DETC2003/VIB-48468. Tanaka, S., Genda, T., Hara, M., and Esashi, M., 2002, ‘‘Component Devel-
关41兴 Breuer, K., Ehrich, F., Fréchette, L., Jacobson, S., Lin, C.-C., Orr, D. J., Pie- opment of Micromachined Gas Turbine Generators,’’ Technical Digest of
kos, E., Savoulides, N., and Wong, C.-W., 2003, ‘‘Challenges for Lubrication Power MEMS 2002, Tsukuba, Japan, IEEE, Piscataway, NJ, pp. 32–35.
in High Speed MEMS,’’ Nanotribology, Hsu and Ying, eds., Kluwer Academic 关68兴 London, A. P., Epstein, A. H., and Kerrebrock, J. L., 2001, ‘‘A High Pressure
Press, Dordrecht, The Netherlands. Bipropellant Microrocket Engine,’’ J. Propul. Power, 17, pp. 780–787.
关42兴 Liu, L., Teo, C. J., Miki, N., Epstein, A. H., and Spakovszky, Z. S., 2003, 关69兴 Deux, A., 2001, ‘‘Design of a Silicon Microfabricated Rocket Engine Tur-
‘‘Hydrostatic Gas Journal Bearings for Micro-Turbomachinery,’’ to be pre- bopump,’’ M.S. thesis, MIT Department of Aeronautics and Astronautics, MIT,
sented at the 19th Biennial Conference on Mechanical Vibration and Noise. Cambridge, MA.
关43兴 Ehrich, F. F., and Jacobson, S. A., 2003, ‘‘Development of High Speed Gas 关70兴 Jamonet, L., 2002, ‘‘Testing of a Microrocket Engine Turbopump,’’ M.S. the-
Bearings for High-Power-Density Micro-Devices,’’ ASME J. Eng. Gas Tur- sis, MIT Department of Aeronautics and Astronautics, MIT, Cambridge, MA.
bines Power, 125, pp. 141–148. 关71兴 Pennathur, S., Peles, Y., and Epstein, A. H., 2002, ‘‘Cavitation at Micro-Scale
关44兴 Miki, N., Teo, C. J., Ho, L., and Zhang, X., 2002, ‘‘Precision Fabrication of in MEMS Fluid Machinery,’’ ASME Paper IMECE 2002-33328.
High-Speed Micro-Rotors Using Deep Reactive Ion Etching 共DRIE兲,’’ Solid- 关72兴 Joppin, C., 2002, ‘‘Cooling Performance of Storable Propellants for a Micro
State Sensor, Actuator and Microsystems Workshop, Hilton Head Island, SC. Rocket Engine,’’ M.S. thesis, MIT Department of Aeronautics and Astronau-
关45兴 Piekos, E. S., and Breuer, K. S., 1998, ‘‘Pseudospectral Orbit Simulation of tics, MIT, Cambridge, MA.
Non-Ideal Gas-Lubricated Journal Bearings for Microfabricated Turboma- 关73兴 National Research Council, 2002, Implications of Emerging Micro- and Nano-
chines,’’ ASME Paper No. 98-Trib-48, presented at the Joint ASME/STLE technologies, The National Academies Press, Washington, DC.
Tribology Conference, Toronto, Canada; also to appear in ASME J. Tribol. . 关74兴 Golley, J., Genesis of the Jet, 1996, Gunston, B., tech. ed., Airlife Publishing,
关46兴 Wong, C. W., Zhang, X., Jacobson, S. A., and Epstein, A. H., 2002, ‘‘A Self- Shrewsbury, England.

226 Õ Vol. 126, APRIL 2004 Transactions of the ASME