Iouri Belski

Department of Communication and Electronic Engineering,

Royal Melbourne Institute of Technology, Melbourne, Victoria, Australia.

Fax: +61 3 9662 1060; E-mail: iouri.belski@rmit.edu.au

**Keywords:** TRIZ, QFD, problem solving, invention, ideal result.

Abstract

Anyone who has ever tried to solve a technical or scientific task, has had to
make decisions on which directions to take. This paper introduces the **Method
of the Ideal Result (MIR).** This method is based on the Russian TRIZ concept
of the Ideal Final Result. MIR can be effectively used in finding the best
direction to the solution. It also evaluates the level of the proposed and
existing solutions. A success story of the development of a new measuring
technique with MIR is presented. Implementation of the technique made a profit
for Russian industry of more than 1.5 million roubles ( US$2.2 million).

The function of the imagination is not to make strange things settled, so much as to make settled things strange.

G.K. Chesterton

If necessity is the mother of invention, who is the father?

It happened in Russia in mid eighties. Mr Shokin, Russian Minister for the Electronics Industry had just announced at the Communist Party Forum that Russian electronics had already overtaken the US counterpart. Being “ahead of the US again”, industry started trial production of TV sets with Surface Acoustic Wave (SAW) filters instead of the traditional ones. SAW filters were designed by the research institute PHONON, where I worked then as a research scientist. During the trial period PHONON produced filters in small numbers, but as soon as the industry made the final decision in favour of SAW filters, two big factories took over production. Their first results were dramatic, with the good to bad ratio being close to zero. PHONON was in trouble.

A simple SAW filter consists of two similar transducers (for input and output) deposited on the surface of a piezoelectric material (substrate) (Figure 1a).

Each transducer looks like two combs inserted into one another to establish
pairs of overlapping electrodes (fingers) of opposite polarity (see Figure 1b).
The electrical input signal, when applied to the input transducer, generates a
surface acoustic wave in the substrate. The wave propagates along the surface to
the output transducer, which converts the acoustical signal back to an
electrical one. Using different geometries of transducers, sophisticated signal
processing can be achieved. The central frequency of a SAW device, *f*_{0
depends mainly on the SAW speed in the substrate Vphase and the period d
of the interdigital transducers (see Figure 1b) and is given by:}

( 1 )

The frequency response characteristic of an ordinary SAW filter is shown on Figure 2.

The width of the main lobe of the filter characteristic, 2D*f,* is
inversely proportional to the number of pairs of fingers *N* in the
interdigital transducer:

( 2 )

About 30 SAW filters for TV sets were fabricated on one wafer of LiNbO3 50.8
mm in diameter by depositing a thin aluminium film using masking lithography.
The production factories utilized PHONON’s mask and the same substrate
material as the research institute. Nevertheless, the difference in the central
frequency *f*0** **among filters was much higher than the maximum
allowed variation (1% rather than 0.1%). The reason was quickly ascertained. The
problem was due to a variation of the wave speed from one wafer to another. To
increase the SAW filter yield it was decided to produce six masks, each with
slightly different transducer finger periods

The SAW velocity had to be measured in the wafer before the lithography. An appropriate non-destructive technique for velocity measurement was needed. Surface acoustic wave could be generated into the piezoelectric substrate by a transducer located not exactly on the surface of the substrate, but at a short distance from the surface. So, to measure the velocity non-destructively, a testing structure consisted of three transducers deposited on the thick piece of glass (non-piezoelectric) was designed. The operator put the wafer on the structure, SAW was generated in it and the wave propagation time from the input transducer to two output transducers was measured. The SAW speed was then calculated by substituting the measured propagation time into the appropriate formula.

The unit for the velocity measurement was implemented quickly. The yield was
increased to about 60%. This was a vast improvement, but still far from ideal.
At that time my research was related to measurements of characteristics of new
materials for SAW devices. This was probably the reason why I had been chosen to
continue the work on the good to bad ratio. My task was to implement a
non-destructive unit for more accurate velocity measurements having in mind *“at
least 90% yield, if you ever want to come back to your research.”*
Necessity was definitely the mother.

A journey of thousand miles begins with a single step.

My first thought was to implement the unit as soon as possible using the most expensive equipment and the most precise technique for velocity measurements which were suitable for production conditions with success assured, I could then return back to my research. I started a survey of available measuring techniques. The existing unit measured the group velocity [1]. The central frequency depended on the phase velocity. It was assumed that SAW velocity in LiNbO3 is independent of frequency, but even slight dispersion (velocity dependence on a frequency) could “help” me to stay out of my research for a long time. The first decision was made - unit must measure the phase velocity. An analysis of available methods of phase velocity of the SAW revealed that a method with the required accuracy did not exist. My research prospects were disappearing on the horizon.

A few months earlier, I came across the book about the Theory of Inventive Problem Solving (TRIZ). It was entitled “The Algorithm of the Invention” [2]. I read the book with a great interest and was waiting for a real task to which TRIZ could be applied. At that time I was not familiar with all principles and levels of TRIZ. The system for making inventions looked unrealistic, but a few TRIZ ideas and concepts were very impressive. One of them was the concept of the Ideal Final Result (IFR).

Anyone who has ever tried to solve any technical or scientific task, has had
to make a decision on which direction to take. Which way to go to solve the
problem faster and more effectively? Analysis of the solution often gives
unexpected results. Most of the time is spent on wrong approaches. Trial and
error prevails in 99% of cases. If the direction to the right solution is known,
any task can be solved faster and on a higher technical level. The IFR concept
helps one to find the best direction. The road towards a good solution opens up
by formulating the Ideal (unreachable in most cases) Final Result. When the
Ideal Final Result is formulated in the correct way, the desired result appears
to happen by itself, without the user’s involvement and effort. **The ideal
system performs a required function without actually existing**. The Russian *“I
wish”* is very useful when formulating the IFR. A possible structure of the
IFR looks as follows:

I wish : **( the result to be achieved )**

IFR can sound strange at the very beginning, nevertheless, it shows the right way to go.

TRIZ used the IFR as a helping concept in the Algorithm of Inventive Problem Solving (ARIZ). At that time I was not ready to use ARIZ. The algorithm contained too many steps and seemed to be very time-consuming. However the concept of the Ideal Final Result impressed me. To save time I decided to start with it.

The IFR for my case was formulated as: “I wish:** a wafer shows its phase
SAW velocity by itself, without my involvement.**” Many questions arose
immediately. How close the solution can be to the IFR? How could a wafer show
the velocity? Sudden magical digits on its surface? It did not sound realistic,
but the idea of IFR was interesting.

I tried to use a realistic approach. The actual situation and all available
resources were decided to be taken into consideration and IFR to be reformulated
into a practical **target-task**. The solution can then be found by solving
this target-task. (Later it became clear that by adding to the IFR concept a
flavour of a reality, the independent method of solution was created. I named it
the Method of the Ideal Result (MIR). *MIR* in Russian means *Peace*)*.*
IFR had to appear when using the measurement unit. Taking into account the
situation, IFR was reformulated into the target-task: “**When an operator
puts a wafer on a testing structure of the measuring unit, the velocity of
interest is revealed.**” The existing unit measured the difference in time
delays of SAW signals with different propagating lengths. To find the velocity,
it was necessary to substitute the measured time difference into the appropriate
formula. Was it possible to know the velocity without extra mathematical
calculations? What **extra resources** were available? Could any **available**
device be used to reveal the velocity numerically?

The only device which could show numerical results was a frequency counter.
My target-task was reformulated again: “**When an operator puts a wafer on a
testing structure of the measuring unit, the frequency counter shows the
velocity of interest.**” It meant that every wafer must have a particular
frequency associated with it, numerically the same as the phase velocity, and
the measurement technique must be based on the frequency measurements. Frequency
was a good choice, as it could be measured with very high accuracy. The
available methods were analysed again. The most convenient was the interference
method [3]. This method uses a SAW structure consisting of three transducers:
one input and two output transducers, located at a different distances from the
input transducer. When two signals of the same frequency and different lengths
of propagation are added together, interference occurs. In the frequency domain
it appears as a number of peaks and hollows on the frequency response
characteristic of the SAW device, as shown on the Figure 3.

An expression for the frequency *fn* of the minimum *n *is
as follows:

( 3 )

where *Vphase* is the phase velocity, *DL* is the difference in
length of propagation, *n* is the number of the selected interference
minimum.

The existing method included measurements of the frequencies *fn* and *fn+*1**
**of two adjacent minima

( 4 )

This method was not precise enough to be implemented in the measurement unit
because of the low accuracy of the measurement of frequency difference (*fn+*1*
- fn*). It was ten times worse than needed. A single frequency, but not the
frequency difference had to be used to satisfy the necessary precision. Was it
possible? Closer analysis of the formula (3) showed that it could be rewritten
as follows:

( 5 )

Only the number of the selected minimum was needed to find the velocity with
very high accuracy. I felt I was very close to the solution. Once the distance
between two adjacent minima (*fn+*1 - *fn*) could be measured, the
number of a particular minimum could be found by evaluating:

( 6 )

and rounding the result for *n *to the nearest integer.

I was unable to find the number of the minimum using this idea, as the minima were not equidistant. The distance between adjacent minima changed with frequency due to a slight dispersion (velocity dependence on a frequency). MIR and the IFR concept seemed to be failing. A wafer cannot reveal the velocity by itself! It was not possible to associate a minimum with its number. One frequency could not “show” the velocity. I began to lose faith with TRIZ.

A few days passed. To solve the task, it was necessary to find the number of
the particular interference minimum. What ways of finding of the number were
available?. What methods could be utilised? Suddenly I understood that the MIR
could be applied again. The IFR was formulated for finding the number of the
minimum in the following way: “I wish: **The minimum revels its number
automatically, by itself**”. In other words the number of the minimum is
known when it is seen. Could this be implemented? I started to analyse the real
situation and resources. What is the obstacle to discrimination between
different minima and their numbers? There are **too many** minima. **If
there were only one minimum inside the main lobe of a filter characteristic and
its number were known it would be easy to do so.** I imagined the main lobe of
a filter frequency response characteristic which contained only one minimum
(Figure 4).

Was it possible to know the number of this minimum? Yes, indeed! The solution
was direct: **There is only one minimum inside the main lobe of the
characteristic and its number is known and is the same for every wafer.**

The implementation was straightforward. The SAW test structure has to be organised to:

make the distance between interference minimums greater than the width of the
main lobe (2D*f*) (to allow only one minimum to appear inside the main lobe
at any time):

** **( 7 )

“force” the minimum with the known number to be inside the main lobe:

( 8 )

A short mathematical exercise with above formulae gave the final conditions
for the length difference *DL* which was “required for the minimum to
reveal its number by itself”:

( 9 )

and

( 10 )

where the integer* **m* represents the number of the minimum
inside the main lobe, *d* is the period of the transducer, and *N* is
the number of pairs of fingers of the transducer.

The main part of the task was solved with only the frequency counter being made to show the phase velocity remained. It was easy to accomplish by rewriting the formula (5) for the velocity as follows:

( 11 )

with *DL* equal to (*m*+1/2)*d*

( 12 )

Provided *d* = 10, 100, 1000,... the phase velocity will contain the
same numbers as the frequency of the minimum. The IDT’s period of *d* =
100 mm was chosen to measure the velocity at the frequency, close to the central
of the TV SAW filter (about 40 MHz for the Russian standard) and to reduce an
influence of the dispersion on the yield.

A few more days were needed to implement the new method and a bit more time to prepare an application for the Russian Patent [4]. I was back to my research. PHONON was out of the trouble.

You never know what you can do until you try.

My first experience with TRIZ was more than simply positive. I invented a new measuring method, helped my employer “smile again”, provided a profit for factories of more than 1.5 million roubles (US$2.2 million) and became wealthier due to the royalty payments. The outcome has found wide application, because of the measuring method’s high accuracy, as other Russian scientists use it in their research.

On the top of it a new method of solution, MIR, was created and applied
successfully. It had been actually utilised twice in the solution process: first
for the main task itself (a substrate reveals the velocity of SAW in it by
itself) and then for the task inside the main (the interference minimum
discloses its number by itself) and seemed to be very effective and quick. The
target-task: “**When an operator puts a wafer on a testing structure of the
measuring unit, the frequency counter shows the velocity of interest**” was
fully accomplished, with the readings of the frequency counter representing the
SAW velocity in the wafer (*Vphase* =3978m/s for the frequency of the
minimum of 39.78MHz).

Since then I studied TRIZ more carefully and deeply. I became familiar with other effective ideas and concepts, and applied them often enough to receive about thirty patents. I still use TRIZ because of its power, but more often I employ MIR. It is simple and powerful enough to provide an excellent solution for the time and effort invested. My Method of the Ideal Result was modified over the years. It is really helpful not only for the problem solving. MIR also enables one to quickly evaluate the level of the proposed solution, as well as to judge how good existing solutions are.

The application of the Method of the Ideal Result to problem solving was presented in this paper. To use MIR for the assessment of the level of the proposed or existing solution it is necessary to:

formulate the IFR in as general way as possible,

write down the list of all available resources,

formulate a target-task, utilising as many available resources as possible,

and then analyse how far the solution of interest is from the target-task._{ }

The distance between the target-task and the solution of interest will be the measure of the level of the solution; the closer they are, the better the solution is.

Years in research and industry taught me that in most of cases the simplest solution is the best solution. Method of the Ideal Result always leads to a simple solution.

Acknowledgment

The author wishes to thank Dr C. Gray and Mr. R. Lord for their interest and useful discussions.

References

[1] McSkimin, H. J. (1961). Pulse superposition method for measuring
ultrasonic wave velocity in solids. *The Journal of the Acoustic Society of
America, *vol. 33, No. 1, pp12 - 16.

[2] Altshuller, G. S. (1973). *The Algorithm of the Invention*. Moscow:
Moscow Worker, (in Russian).

[3] Temmyo, J., Kotaka, J., & Inamura, T. (1980). Precise measurement of
SAW propagation velocity in LiNbO3. *IEEE Transactions on Sonics and
Ultrasonics,* vol. SU-27, No. 4, pp218 - 219.

[4] Belski, I., & Sorokin, V. (1987). A method of SAW-velocity
measurement. *USSR Patent* No. 1298549 (in Russian).