THE COMPETITIVE SEMICONDUCTOR MANUFACTURING HUMAN
Second Interim Report
Clair Brown, Editor
11.5 Engineering Innovations
on the Shop Floor
Management of innovation on the shop floor includes ensuring that
shopfloor improvements are controlled and, with approval, become
part of the organizations knowledge base. In semiconductors,
the consequences of employee modifications that are not standardized
or approved can be serious, since the characteristics of the chip
may be affected inadvertently. Making unauthorized modifications
is often referred to as "tweaking" the process. However,
JapanTech and USTech differ in their approaches to controlling replication,
which reflects their timing of and their procedures governing the
transfer of the new product to manufacturing.
Exact Replication and Problem Solving at USTech
USTech wants to tie short-run activities
to long-run planning as they move process technology across fabs.
USTech takes extreme measures to ensure that the fabrication process
is replicated exactly and that any potential employee improvement
goes through appropriate channels before being implemented. USTech
instituted the rule of exact replication of the given specifications
in order to minimize portability problems and eliminate declines
in yields over time. In the past when the technology was transferred
to high volume, the fab initially had high yield; but since many
manufacturing problems had not been solved, the manufacturing engineers
modified the process and caused a fall in yield. As one manager
said, "We had to crush creativity in manufacturing fabs (no
tweaking). We start with top down control, and fabs need approval
for any changes." However, engineering still occurs in manufacturing
(e.g., projects to reduce variations in the process or to reduce
maintenance problems on loading procedures).
Exact replication has reduced complexity in manufacturing, but other
problems arise. Exact replication is a relative concept and makes
the transfer process tricky and expensive. One manager said, "Exact
replication needs common sense: how to measure, and who decides.
We want the fab to copy exactly and then get help with problems.
This requires a lot of training for engineers in the manufacturing
fab." The formal methodology for the hand-off includes using
and updating a "rule book," which gives process development
and characterization, information on running modules, and baseline
performance. The rule book is not comprehensive, but interpretive.
It should be "a living document" with engineers adding
information as the volume increases. Although, as one manager explained,
"Engineers do not like to document, and so updates are hard
to get done. They dont have a customer, so the incentive is
gone." Besides discouraging tweaking, managers often must encourage
more mundane activities. For example, manufacturing engineers may
resist doing the mundane but necessary tasks of writing up the documentation
for process changes. Instead, engineers would rather be doing the
more challenging tasks of solving process problems, which may not
be part of their assignment.
The equipment engineers, who are assigned to oversee daily operations
of specific equipment, are responsible for reaching specified equipment
performance goals. They are also responsible for calibration of
tools across the factory, which is another job that engineers do
not want to do and often hand off to technicians. In hiring and
recruiting manufacturing engineers, USTech looks for the "good,
hands on" engineer. By the time of hand-off, manufacturing
engineers have taken ownership. As one USTech manager said, "We
are trying to combine the development skills of the development
engineer with problem-solving skills of the manufacturing engineer."
A trade-off exists between fostering individual creativity and controlling
innovations, especially for engineers involved in transfer of new
technology to the fab.
Making Improvements and Solving Problems at JapanTech
JapanTech is a conglomeration of many
fabs that operate as separate companies with different cultures.
The parent R&D center must decide how much power to give the
fabs in order to motivate and encourage creativity while the parent
company still maintains control. The manufacturing engineers at
the fabs must be made to feel like they are part of the parent company
while at the same time they strive to make their fab outstanding.
At JapanTech, the development lab does not undertake activities
involved in developing the commercial sample. These activities (e.g.,
improve process and equipment to get reasonable yield, confirm device
reliability, and complete qualification samples) are done at the
mass production fab. Generally, JapanTech purchases only 20% new
equipment for volume production at the receiving fab. For the rest
of the equipment, the development fab issues specifications and
the manufacturing engineers "tune" the existing equipment.
JapanTechs strategy is to make many products on the same line,
which adds depth of knowledge since there are many different variations
for a particular product line.
JapanTech brings high-volume fab engineers to the development center
so that they understand the process and feel some ownership. They
eventually do the work of the development engineers, who are on
loan to the fab for the transfer from the parent company. There
are also formal meetings with engineers and managers, meetings with
only engineers, and informal meetings, phone calls, and the exchange
of data between the development engineers and the manufacturing
engineers. The engineers at the manufacturing fab may suggest changes
and send information about suggested refinements to the R&D
JapanTech encourages competition among its plants on cost performance
and on improvements or refinements to the system. Annually, JapanTech
has targeted goals of price and costs and highlights the fab showing
the best performance. Competition is more spiritual than monetary
("Japanese dont want to lose"). Yet the development
engineers from the parent company support new process introduction,
and there is a need for cooperation between engineers at the plant,
at headquarters, and among plants. Incentives for both competition
and team work are present. For example, communication among engineers
is an important part of coordination across divisions and plants.
At the second high-volume fab to receive a device, with the transfer
being made from the first high-volume fab, older equipment is expected
to get the same yield. Since the fabs are in competition, this may
affect their cooperation in making the transfer.
In our survey of engineers in Japan and the U.S. on the procedures
they most recently used to solve a technical problem, four-tenths
of the Japanese engineers reported they first approached a team
member with similar expertise and three-tenths discussed the problem
with the team as a whole. (Charts 11.3A and
11.3B.) One-third of U.S. engineers first consulted a team member
with similar expertise, and one-fourth first consulted the team
as a whole. Altogether, the American and Japanese engineers reported
taking similar approaches to problem solving: eight out of ten of
the U.S. and Japanese engineers reported approaching a team member
with similar expertise, seven out of ten reported discussing the
problem with the team as a whole, four out of ten reported approaching
a co-worker with similar expertise on another team, and four out
of ten reported approaching their supervisor in the first four steps
taken. In addition, one-half of the Japanese engineers also reported
consulting someone from manufacturing in the first four steps. Neither
group of engineers reported heavy reliance on team or company documents,
although team documents were consulted by one-fifth of U.S. engineers
and two-fifths of Japanese engineers at some point in the problem
solving process. Journals were consulted by four out of ten Japanese
engineers and three out of ten U.S. engineers.
The tension between rewarding the individual or team to support
the creative process versus maintaining control over the development
of technology has been analyzed by comparing the Japanese and American
employment or HR systems. Since product and process innovations
in both R&D and manufacturing define a company's performance,
the creation and management of knowledge and skills within this
rapidly evolving industry emerges as a key to long-run success.
JapanTech and USTech use different employment systems to manage
their development and transfer process. Both companies have HR systems
made up of consistent and re-enforcing parts, which reflect both
their product and labor market environments. As a producer of logic
devices, USTechs goal is to control the market for their devices
by maintaining a lead over potential competitors in introducing
the next generation. Time to market, but not price competition,
is an important part of strategy. As a producer of memory, JapanTechs
goal is to keep up with their competitors in introducing the next
generation. Since generations are now separated by only two years,
the time to market and price competition are both important parts
of their strategy. JapanTechs labor market institutions include
lifetime employment, annual national wage determination, a company
union, and a higher education system that does not provide much
research or graduate education. USTechs labor market institutions
include decentralized and individualized wage setting in a mostly
nonunionized industry, a mobile labor force, employees use to a
high degree of autonomy and input into the job assignment process,
and a higher education system known for its research and graduate
The resulting HR systems have the important following characteristics:
Job assignment reflects past performance
and expected future performance at USTech, while job assignment
at JapanTech reflects projects requirement for skills
and knowledge already acquired, and the plan to develop the
knowledge of junior engineers. USTech rewards development engineers
for outstanding performance by assigning them more responsibility
on their next project. JapanTech assigns development engineers
to projects on the basis of company needs and requires more
rotation among different types of tasks,including fabrication.
U.S. engineers were more likely to
consult engineers outside the company for technical information
(although not so at USTech), while Japanese engineers were more
likely to use journals and patents.
USTech transfers a new process to
a manufacturing fab after high volume has been accomplished
at the development fab. JapanTech makes the transfer much earlier
and develops the engineering sample at the manufacturing fab,
which solves volume-related process problems. Therefore, JapanTech
depends on process modifications and improvements by the manufacturing
engineers, while USTech prohibits it.
Although engineers at both companies
prefer to do the more challenging development work rather than
the mundane tasks such as documenting modifications and calibrating
equipment, this problem seems to be more widespread at JapanTech
since most engineers do not specialize in development or manufacturing,
and they begin work with a BS degree. At JapanTech, engineers
(excluding the Central Research Labs) rotate among development
and fabrication tasks. USTechs engineers are more specialized,
and their work reflects their education with the development
and research engineers likely to have an advanced degree and
the manufacturing engineers hired with BS degrees.
A confrontational style is practiced
at USTech, but disagreements are put aside after a commitment
to an idea is made. Autonomy and creativity are highly prized
at USTech in development, but not in manufacturing. A consensus
approach is practiced at JapanTech, and teamwork and stability
are highly prized.
Incentive or Compensation Systems
Both companies use a relative performance
ranking system to evaluate their engineers, but the rewards
for performance are different. Pay, especially for the first
dozen or so years while the engineer is in the company union,
is more rigidly set at JapanTech than USTech, which is more
performance oriented. USTech focuses on rewarding an individuals
ideas and efforts. Although both companies pay bonuses, the
bonus at JapanTech mainly reflects national wage setting while
the bonus at USTech reflects performance at the unit, division,
and company levels. Also, USTech engineers can be richly rewarded
with stock options.
JapanTech is struggling with the
aging of its work force and the declining demand for managers
relative to those eligible. In a two-tier management system,
JapanTech is exploring how to provide cost-effective incentives
to older professionals who are specialists and do not supervisor
employees. A younger and faster growing company, USTech only
mentioned this as a potential problem. However, USTechs
performance oriented and flexible compensation system allows
it to deal with changing company needs and employee demographics
more easily than JapanTechs compensation system, which
is more dependent on rigid job grades and career ladders.
Knowledge and Communication Systems
USTech rarely makes public presentations,
publishes papers, or shares information with outsiders (including
vendors), since it believes there is nothing to learn from others
through sharing knowledge. Patent applications are made only
if the knowledge can be learned through reverse engineering.
In contrast, JapanTech depends on public presentations to maintain
its reputation and to announce the introduction of new devices.
Knowledge sharing with suppliers is part of the equipment development
process. Engineers are expected to submit 3-4 patents annually.
The publication of papers, patent applications, and conference
presentations are important for keeping up with the competition
and for the advancement of an engineers career.
When research engineers with advanced
degrees and manufacturing engineers with BS degrees are hired,
often straight out of the university, to work for USTech, they
are assumed to have the research skills necessary to undertake
their own research projects or the manufacturing skills necessary
to oversee the operation of specific equipment. Both research
and manufacturing engineers go to work for JapanTech after graduating
with a BS degree. They are expected to learn on the job through
their team work, continual firm-based training, and job rotation
that usually includes both development and fabrication activities.
Some engineers earn advanced degrees while working, either from
the company or from an affiliated university.
At USTech, junior development engineers
are given major responsibility for developing new technologies.
At JapanTech, major assignments are given to subteams within
a team setting, and new ideas are evaluated by a test group.
Junior engineers are assigned to work with senior engineers
and are expected to learn through their work assignments.
Both companies exchange engineers
between the development fab and the manufacturing fab and between
manufacturing fabs in order to transfer knowledge and to encourage
the ownership of the new technology. However, since USTech transfers
the new technology at a much later stage of development, fewer
engineers need to be exchanged between the development fab and
the manufacturing fab compared to JapanTech.
The decision of when to transfer the
new technology from the development fab to the volume manufacturing,
along with the institutional environment of the labor market and
the development demands imposed by the characteristics of the product
or device, determine the requirements and constraints placed upon
the HR system. A companys strategy for innovation in development
and in manufacturing is intertwined and reflects the relative importance
in maintaining market control (i.e., remaining first to market for
a particular product) versus the relative importance of keeping
up with the competition in delivering new products at competitive
prices. USTech is an example of the former situation and JapanTech
is an example of the latter. USTechs transfer of new technology
occurs after high volume manufacturing has been accomplished. They
manage the potential tension between encouraging new ideas and controlling
the development process by having their development engineers highly
specialized and by assigning major responsibilities for solving
a specific problem to one or two engineers. In addition, the required
design specifications constrain creativity in the development process.
Engineers who are successful are highly rewarded both monetarily
and by their next assignment; those who are not successful are likely
to leave. At the manufacturing stage, USTech has a strict policy
of copying the new technology exactly. Manufacturing engineers,
who have only BS degrees, are involved in the last part of the development
process in order to train them as well as to have them feel like
they have ownership of the process.
JapanTechs transfer of new technology occurs early in the
process after only a few good dice are produced in the development
fab. The group leader decides at an early stage among competing
ideas, primarily based on the results from the test team. The junior
engineers education continues within the company through working
with senior engineers on projects and through formal classes. JapanTech
also believes that job rotations that include fabrication as well
as development assignments are an important part of the education
process. Since an engineers career depends on the teams
presentations at conferences and patent applications, individual
creativity is less important than team outcomes, and the individual
is granted less autonomy and responsibility by the group leader.
In conclusion, a system emphasizing individual autonomy, responsibility,
and reward for development engineers (but not for manufacturing
engineers), along with no knowledge sharing with outsiders, characterizes
USTech. A system of team work, explicit career ladders, and company-based
education for engineers, who do not specialize in development or
fabrication jobs, along with required sharing of knowledge through
required patent applications and presentation of papers, characterizes
JapanTech. Consistent with the external environments imposed by
their product and labor markets, these approaches have resulted
in USTech being a top performer in logic and JapanTech being a top
performer in memory.
Go to Table of Contents for this
Go to Table of Contents for the
CSM-HR Interim Report