by Dr. Dirk Ortloff
Dirk Ortloff, Jens Popp and Andreas Wagener, Process Relations
It is no secret that the semiconductor industry and other high-tech
industries like the MEMS and NEMS industry have suffered over the past year.
According to figures from the GSA, semiconductor sales were down by 6% in 2008
compared to 2007. Nor is it much of a secret that many industry figures have
championed innovation as the key for companies to ride out the storm and indeed
to prepare for the upturn when it comes. Most high-tech industries, which are
particularly suffering in the current climate, have experienced a downturn in
sales. As such, driving down costs is obviously essential in these tough times.
However, some long term cost savings have been largely unexplored by high tech
companies until now. In particular, the semiconductor, MEMS/NEMS and
photovoltaic industries have the potential to decrease time-to-market and the
cost of product development by investing wisely in tough times.
This paper investigates the challenges, discusses some important
considerations and highlights potential methods for companies to come out of
this crisis stronger. The value of specialised software, called Process
Development Execution Systems (PDES), will be discussed. A PDES helps to cope
with the challenges of growing complexity, shorter time-to-market and
development inefficiencies. The paper will demonstrate that the demands to grow
future markets with increasingly complex products can only be fulfilled through
the intensive use of PDES software like XperiDesk working together with product
engineering methodologies. It will be shown that the tools and methods give
companies a competitive advantage by developing better solutions and shorterning
The economic crisis is currently demanding the lowest possible development
costs. These demands are easy to understand but in practice very hard to
achieve. Tight time limitations and the use of the wrong tools, or no tools
whatsoever, make the situation more difficult to resolve. Added to this, the
complexity of process development increases every day.
More technology options, material choices and suppliers complicate matters
even further. Now, more than ever before, new data management strategies need to
be implemented to help deal with the challenges. Employing intelligent software
support is one potential solution that can offer many benefits. However, the
vast amount of development data is difficult to manage. To avoid getting lost in
an ocean of information, an intelligent solution covering the full development
loop, indicated in Figure 1, is required.
Figure 1. Development
On the other hand, cash is king nowadays. So is the timing really right to
invest into new strategies while old strategies are more or less still
applicable? Some leading industry figureheads have championed innovation as the
key for their organisation to ride out the storm and indeed to prepare for the
upturn when it comes.
It is very difficult to disagree with this viewpoint, innovation is crucial
to any company, recession or not. However, when thinking of innovation, it is
often only the straightforward and tangible implications that are considered.
People often associate innovation with new products or new strategies. What is
often underestimated is that the main challenge for engineers now is the
increasingly complexity of new technologies combined with the need to develop
and deploy these technologies within a much tighter time scale at an
increasingly competitive price. To meet these challenges innovation needs to go
a lot deeper into a company.
The main challenge for engineers is dealing with new technologies which are
increasingly complex, whilst having to commit to developing products within a
much tighter time scale at an increasingly competitive price. Over the past
twenty years, there has been a focus on shortening cycle times in manufacturing.
The widespread use of various statistical tools such as 'Statistical Experiment
Design' has enabled this development to continue.
However, there are physical limits to development time and the semiconductor
industry in particular is fast approaching them. If we want to keep up this
rapid development of ever more intelligent and function rich ('More-than-Moore's
Law'), yet low cost chips, new technologies and materials need to be employed.
The process development stage of thin-film developments has obvious scope for
improvements as introduced in more detail in Ref #2.
For example, by utilising existing knowledge to its fullest, the number of
experiments that need to be performed can be limited. Those that remain can be
simulated first rather than being performed physically immediately. Finally, all
the results gained from the experiments can be automatically collected using a
structured process to develop a knowledge base for future use. Planning with
such techniques can reduce the time spent iterating development steps.
Collaboration between different people and groups inside or outside a company
is imperative for high tech developments. This can include collaboration between
different groups around the globe. Market trends and development costs even
drive the need for collaborative development activities between different legal
Collaboration can improve the time-to-market. However, a proper central
development platform, including communication and electronic knowledge transfer
functionalities, is needed to do so efficiently. Historical data must be
recorded and shared otherwise the mistakes of the past will be repeated.
Experts in semiconductor process development estimate that 10-15% of failed
and retaken experiments could be avoided if previous results were accessible. It
is vital to document everything - every idea, project, experiment, meeting, and
conclusion. If this is done successfully and effortlessly, engineers can draw
optimum results from new results. New and existing data can be correlated and
thus new information can be exposed. For example, PDES can build up a network of
imperative information that will not only benefit current developments, but it
will be useful in future process developments too.
Another challenge often encountered is the transfer of manufacturing
processes from research into production. Once the experiment and development
process has been streamlined, a Manufacturing Execution System (MES) should be
implemented, to assist with the extensive production process. However, it is
only as good as the input it receives. If the program is cluttered with
unnecessary experiments, the cycle time will inevitably increase. A key benefit
of implementing a system such as the PDES XperiDesk is that it enables
manufacturers to run less wafers in the fab. As a result, it frees up valuable
resources, saves money and significantly reduces the probability of equipment
Documenting and reporting the development progress can be tedious at best.
Cluttered results storage burdens development engineers with major manual
effort, requiring them to manually collect data from diverse machinery.
Additionally the assembly of the collected result data into reports and the
evaluation can take a major part of engineering time. Industry experts report
that up to 80% of engineering time is spent on data collection and arrangement
rather than on data evaluation. Even if the average is much lower, it is still a
major part of the working time spent on tedious and error prone tasks.
Automising these can inject creativity back into the development process.
Reporting on the development status is often more of a manual assembly of the
reports rather than an automated process. The input data is often not up to date
so that the work in progress (WIP) status is not necessarily precise. The
impacts of these effects are even aggravated by quality assurance and compliance
demands such as ISO 900X, CMMI, SOX etc. Because those apply more and more in
development as well as in production, there is a strong demand to fulfil the
imposed documentation requirements.
To summarise, making cost reductions in process development can be achieved
by following four basic rules:
1. fully utilise existing knowledge,
2. learn in
a different way (e.g. by simulations),
3. gain more information
from your experiments, and
4. transfer all required knowledge to
Introducing Software Support
These rules are key while facing the current economic climate. But the
question arises whether the timing is right to introduce new software to support
the associated tasks now. When developing new products or technologies what many
people fail to consider is the Total Cost of Ownership (TCO). This refers to the
full burden of costs involved in development and innovation. TCO goes far beyond
the usual costs, overheads and staff costs. It also incorporates all of the
costs involved with acquiring new or better machinery, the HR costs of taking on
new staff, the costs of adopting new methodologies and even activities such as
staff training that are often overlooked. This is the depth of innovation this
paper refers to. It is not enough to develop a new product. Rather, in order to
engender long term innovation and commercial success, companies need to be
looking at a much bigger picture.
What is more, there is a huge advantage to dealing with these issues during
an economic downturn: costs are lower during a recession; generally it is a time
when companies will be operating with reduced staff levels; and lastly during a
recession there will be, generally speaking, more bandwidth and resources
available within a company. The findings of a McKinsey study of the 1990-91
recession show how companies can take advantage of these circumstances.
The study found that companies that remained market leaders or became serious
challengers during the downturn had done so by increasing their acquisition,
R.&D., and ad budgets, while companies at the bottom of the pile had reduced
them. Certainly there are already companies that take advantage of these
circumstances; IBM has a stated policy to invest during downturns - in people,
in training and in technology. But it is not necessarily a widespread
philosophy. Most companies understandably view downturns as a time for cost
cutting and a rationalisation of activities. However, to introduce these
disruptive technologies and methodologies internally, there is no better time
than during an economic downturn.
Not only are the costs of new machines, software or technology lower during a
downturn, but also a whole raft of other cost savings can be made. For instance,
during times of reduced staff, introduction costs and hurdles for new systems or
procedures are much lower. Moreover, instituting more staff training during a
downturn, when staff levels are lower, is a significant cost saver. And more
generally in economic downtimes more resources are available for evaluating
current practices and learning about new ones. During an upturn there is very
little bandwidth, in terms staff availability and time pressures, for
introducing these new technologies, procedures or training courses.
Companies need to prepare for the next upturn, and in recent months there
have been more positive economic signs. Future Horizons' latest Global
Semiconductor Report shows that April 2009 had the strongest month-on-month
growth for April since 1996. Ultimately companies have to find ways of doing
more with less (during a market or production rise companies will generally have
fewer engineers than at the peak of the market).
Of course, new products and new ideas are what will ultimately deliver the
commercial success technology companies desire, but where will these products
and ideas come from? These things cannot be left up to chance and investing in
an organisation, its infrastructure and its staff has the potential to give a
significant head start.
This paper gave a brief overview of current process development practices and
challenges with a focus on possible improvements using better software support.
Additionally it highlighted aspects of introducing such software support and the
impacts and timing of the introduction. According to results of McKinsey studies
investigating the impacts and strategies of investing during challenging
economic times, the timing is right to change the development strategies in
tougher times rather than during rising or peak times. Better software support
can be achieved by introducing a Process Development Execution System such as
XperiDesk (a commercially available implementation of a PDES) which supports the
whole development flow - from the first device idea to the transfer of the
resulting recipe into production or to a collaborating partner.
1. K. Hahn, T. Schmidt, M. Mielke, T. Brück, D. Ortloff: Micro
and Nano Product Engineering using Data Management for Silicon-based Fabrication
Process Development. In: Proceedings of the 9th IEEE International Conference on
the Nanotechnology, Geneva, 2009.
2. D. Ortloff, J. Popp, A.
Wagener: Recurring deficiencies in process development support. In: Proceedings
of the 13th International Conference on the Commercialization of Micro and Nano
Systems, Puerto Vallarta, 2008.
Presented at COMS 2009, Copenhagen
Copyright AZoNano.com, MANCEF.org