1. Introduction: Definition of diagnostic problems

TRIZ [1-4] is as an essential ingredient for modern innovation, as quality is a key factor for success in customer satisfaction. Initial research has shown [2-5] that the synergy from the cross-fertilization of TRIZ and quality methods (QM) is very promising. If TRIZ and quality are necessary conditions, then what are goals ? We think that one of the most important targets is design of artificial systems (i.e., products) with 100% reliability. In order to achieve this goal we proposed and tested a new method of diagnostic problem solving (DPS), that is discussed in this paper.

The technical problems are a special type of predicament that occurs in the artificial systems during their design, manufacture, usage or utilization. It is well known that most of inventive technical problems can be resolved within TRIZ [1,2,4]. The diagnostic problems are the particular case of the technical problems that occur during an artificial systems defeat or a possible failure. Hence, the diagnostic problems are connected with discovering (exposing, finding out) causes (reasons) of different phenomena. Simply speaking while solving the diagnostic problems, we have to answer the questions: "Why does such a thing occur ?" (the diagnostic problem of the first kind) or "What might go wrong with this system in future? (the diagnostic problem of the second kind). In order to solve such problems a few methods have been proposed by USSR school of TRIZ [3-7] and by Japanese and USA specialists in quality [8-12].The major efforts in the development were focused on problem solving tools and tools for data analysis as well as organization of team work for Failure Mode and Effect Analysis (FMEA) [8] or Root Cause Analysis (RCA) [9] correspondingly (see the table at the next page).

NOTES: The meaning of marks are: Perfect (++); Good (+); Weak (-)

In the framework of TRIZ the following approaches have been developed [3-7]:

a) Anticipatory Failure Determination (AFD) is the method that has been developed by B.L. Zlotin and A.V. Zusman. It consists of two parts (AFD-I and AFD-II), known in Russia as "Diversionary Analysis" and "Research Problem Solving Method" [3,4]. According to this method, instead of asking: "Why does such a thing occur and what has caused such a result?" or "What might go wrong with this system in future? (RCA, FMEA etc. approaches), we ask: "How can we achieve this result?" The last question transforms a diagnostic problem into an inventive one and enables us to use all TRIZ instruments in order to find the solution. The only limitation is to use exclusively resources of the system only in order to find the solution. Some suggestions for improvement of AFD were discussed in the works of S. Kaplan [6].

b) Failure Prevention Analysis (FPA) is the method that was developed not long ago by G. Yezersky [7]. This method is aimed to improve QM (first of all FMEA). FPA is based on a "failure-free" system model creation and a further functional analysis of this model in comparison with the real system. In contrast, a real system with possible "built-in" failures is analyzed in usual FMEA. The approach can be considered as an application of one of the basic TRIZ concepts of the Ideality [1-4] to the Diagnostic Problem Solving. Only a short description of FPA is currently available in the public domain [7], so it is too early to review this improvement of FMEA, inspite of the very promising ideas that the method is based on. Note that this method can advance so-called Design for Quality approach [10].

In this article we give an executive overview of the Diagnostic Problem Solving method. During the process of solving a diagnostic problem of the second kind it is transformed into one of the first kind. As result this article focuses on the more general solving process of diagnostic problems of the second kind.

The structure of this paper is as follows:

• Some weak points of AFD are outlined in the next section.
• Classification of diagnostic problems is presented in Section 3.
• The short review of DPS is given in Section 4 (we do not present here reviews of TRIZ and QM. There are many other books [1-4,8-12] available about these techniques).
• The algorithm for DPS is presented in Section 5.
• The case study and some conclusions are given in the last two sections.

The aim of Diagnostic Problem Solving (and this work in particular) is to increase efficiency of the Quality Methods by means of clear classification of the diagnostic problems and their connection with the TRIZ and QM tools by a new algorithm.

2. Drawbacks of AFD

One of the weak points of the Anticipatory Failure Determination [3,4,6] is the unsatisfactory connection of this approach with the TRIZ tools. It is not so clear which TRIZ tool has to be chosen in AFD in order to resolve the "transformed problem".

This disadvantage of AFD appears because

- a classification of the diagnostic problems has NOT yet been proposed,

and

- clear rules of the diagnostic problems transformation into inventive situations are NOT determined.

Another weak point of AFD [3,4,6] is the ignorance of Western quality methods for handling the similar problems (like RCA, FMEA, 8-D, etc.) that have been developed in order to predict future failures in the artificial systems [8-12]. On the other hand, most of Western Quality techniques do not have good problem solving tools (like ARIZ or Altshuller's Matrix [1,2,4,13]).

3. Classification of the Diagnostic Problems

According to the Ref. 5 the diagnostic problems after their "transformation" into inventive problems can be divided into four types:

i. Finding ways to perform the action that causes the harmful effect connected with the change of parameters or properties.

ii. Finding ways to perform the action that causes the harmful effect connected with wrong measurement.

iii. Finding ways of intensifying the action that causes the harmful effect.

iv. Finding ways of elimination of the useful effect, function or interaction.

4. Overview of DPS method.

The following three main stages are identified in DPS [5]:

I. "Invention" of possible harmful effects.

The functional analysis is made in order to expose all possible harmful effects and system's failures. Possible harmful effects creation or intensification are considered for every element of each functional group: function itself, function carrier, function object.

In order to "invent" possible harmful effects and/or failures one has to:

1. Make functional analysis of the system;

2. Divide the system into functional groups (each group consists of function carrier, the function itself and function object(s);

3. Determine basic life cycle stages for the analyzed system (manufacture, storage, usage, maintenance, utilization etc.);

4. Determine (invent) for each element of every functional group at every stage of the system's life cycle: a. possible harmful effects and/or failures that might be caused by this element to other elements of the system or environment (including people) . b. possible harmful effects and/or failures that might be caused to this element by other elements of the system or environment (including people).

For either "a" or "b" the following is considered:

1. harm (connected with changing of parameters or properties) creation

2. harm generation (connected with wrong measurement)

3. intensification of harm that already exists (if it exists)

4. useful interactions' elimination.

(At this stage ignore how exactly harm or failure can be caused. The results can be presented in a graphical form, for example, as an Ishikawa's fishbone diagram [9-12]).

At the end of this stage, a simple checklist for invented harmful effect "realization" in the real system, can be used. Find the answers on the following questions:

a. Is the effect realized in the system ?

b. If it isn't - What are the possible conditions in which it can be realized?

This stage should be completed for each stage of the analyzed system's life cycle (manufacturing, work, storage, etc.).

II. Transformation of a diagnostic problem into an inventive one.

Any diagnostic problem can be transformed into one of two types of the inventive situations:

a) One has to perform some function in order to receive the harmful result which then is considered as useful.

b) One has to eliminate an undesirable effect (useful interaction, which then is considered as harmful, or low efficiency of a harmful action, which then is considered as useful).

Now we can switch from these situations to the problems according to the rules of such temporary transition:

For a) we define:

the function;

the object of the function;

the known method (facility) of the function's performance;

the undesirable effect (UDE) which arises if we use this known method.

For b) we define:

the UDE;

the element connected with this UDE;

the function of this element;

the object of the function.

Now we choose the possible directions in order to resolve the problems, such as:

1. performance of the function without use of a known method (system);

2. elimination of the UDE which is connected with use of a known method (system);

Theoretically, the choice of the correct direction is always possible. This choice depends on the condition: "Only the resources of the system can be used in order to find the solution", but in reality both of the directions have to be checked.

III. Elimination of harmful effects and/or conditions in which it can be realized.

We have to solve inventive problems ("usual" for TRIZ [1,2]) or routine engineering problems at this stage.

Note that the regular TRIZ goal of inventive problem solving is to INCREASE the Ideality of a system by increasing the sum of useful functions and decreasing the sum of payment factors. In opposite, in order to DECREASE Ideality of a system in DPS, we shrink the sum of useful functions and increase the sum of payment factors, because we consider future failures and the undesirable effects (UDE) as the major factors. However, in the future improvements of the system, a low efficiency of some function's performance might be considered as an important UDE.

IV. Evaluation, verification, development and implementation of solution

5. Algorithm for the Diagnostic Problems' Solving.

The DPS algorithm is presented here with a few appendixes at the end of this section. The algorithm's skeleton consists of four parts and includes also a few tables and templates (given in the appendixes too).

1. The problem formulation.

1.1. Select the most important diagnostic problems using the quality methods and tools presented in the following table:

1.2. Transform the diagnostic problems into inventive one.

NOTE: Transformation is performed by asking: "How can we get this result?" (the "active" question) instead of the "passive" question: "Why does this occur and what causes such a result?"

The diagnostic problem can be transformed into two types of problem situation:

A: It is necessary to perform any function but the appropriate systems or facilities for this are absent or unknown.

B: The problem situation is connected with an undesired effect (UDE) inside the existing technical system.

The steps for these types of problem situation are presented in the table shown at the next page.

TABLE: Sub-procedures for the Two Types of Problem Situation

2. Overcoming restrictions.

To resolve diagnostic problem we have to deal with many restrictions. There are two types of restrictions:

- restrictions on a substance incorporation;

- restrictions on a field incorporation.

2.1. Restrictions on substance incorporation.

Following are the directions for overcoming restrictions in this case:

2.1.1. Temporal incorporation of a substance.

2.1.2. Incorporation of a substance in the precise place only.

2.1.3. Usage of the substances already existing in the system or environment like the incorporated ones.

2.1.4. Incorporation of the transformed substances of system or environment itself (i.e., usage of the transformed system's/environment's substances as the incorporated ones).

2.1.5. Usage of vacuum, air, foam as an incorporated substance.

2.1.6. Usage of the substances mixture. In this case, different types of mixtures might be used: mixture of various system substances; mixture of system substance and environment; mixture with air, foam and so on.

2.1.7. To use a field instead of incorporated substance.

2.2. Restrictions on field incorporation.

Following are the directions for overcoming restrictions in this case:

2.2.1. Temporal incorporation of a field.

2.2.2. Incorporation of a field in the precise place only.

2.2.3. Usage of the fields already existing in the system or environment like the incorporated ones.

2.2.4. Incorporation of the transformed fields of system or environment itself (i.e., usage of the transformed system's/environment's fields as the incorporated ones).

2.2.5. Usage of vacuum as an incorporated field.

2.2.6. Usage of the combinations of fields. In this case some different combinations might be used: mixture of various system fields and environment; "field vacuum".

3. Ideal Final Result.

3.1. Indication of the time and place of the Ideal Final Results [1-4] demands realization.

3.2. In order to solve either A or B types of problems we can formulate the Ideal Final Results for each type correspondingly according the rules:

A): The function is performed without function carrier:

by object of function itself or

by other elements of the system, or

by environment.

This has to be realized:

without CHANGING of the system,

without violation of the restrictions, and

at desired time and space.

or:

B): The UDE is removed:

by object of function itself or

by other elements of the system, or

by environment.

This have to be realized:

without CHANGING of the system,

without violation of the restrictions, and

at desired time and space.

NOTE: Use the Appendix 1 if you cannot formulate the Ideal Final Results .

4. Choose the appropriate TRIZ tools for solution.

4.1. In order to find appropriate TRIZ tool one can use templates or the table.

• Template for technical contradiction removal

( see the Appendix 2 );

• Template for physical contradiction removal

( see the Appendixes 3 and 5);

• Template for effects

( see the Appendix 4 ).

Notes to the table "TRIZ Tools for DPS" :

a) The number in the first column corresponds to the group of basic inventor's hints that are appropriate for the suitable problem (see the Appendix 5 );

b) We use the original Altshuller's numbering in the recommended principles' column (see Ref. [1,2]);

c) We use the numbering from the short register given in the Appendix 4 for the column "Recommended physical effects".

4.2. Once again check all restrictions (see the part 2 of this algorithm).

In conclusion of this section we would like to note that the presented algorithm is preliminary, more sophisticated DPS algorithm is one of on-going research activities of TRIZ Experts.

Algorithm's Appendixes

6. Case Study

Let us consider a device for shadow termination in sputtering machine. The function of this device is to press a metal mask to a wafer.

The device consist of a base plate 1, bridges 2, press-strips 3 and a metal mask 4. The mask is pressed to the wafer 6 with the help of screws 5 and the press-strips 3 (see the figure). The device is made of stainless steel. The bridges are welded to the base plate. Every bridge has two slots. The press-strips are inserted into these slots and can move upwards and downwards inside them.

The wafer looks like a sandwich consisting of alumina ceramic strips which are stuck with special glue to a glass carrier.

A worker puts the mask upon the wafer and then inserts this "sandwich" into the device. Then he or she performs alignment of mask position under a microscope and presses the mask to the wafer with the help of the press-strips and screws. Then a number of devices is put upon pallet and loaded into the sputtering machine. The wafer's parts uncovered by the mask are coated with Chromium (Cr) and later with Copper (Cu) in a deep vacuum at the temperature of about 300^oC in the machine. The principle of work of this sputtering machine is based on plasma bombing of the Cr, Cu or other metal targets. The atoms "bombed out" from these targets fall down and coat the uncovered parts of the wafer. Then the wafers are taken out of devices. At the end of the work-cycle the devices are cleaned from Cr and Cu in special chemical solutions.

Let us perform, first of all, short functional analysis of our system (it is simple enough) and then for every "function group". Let us "invent" harmful effects which can appear at every stage of life cycle of the analyzed device:

manufacture,

work,

maintenance,

storage,

utilization.

We will consider a potential harm for every element of a functional group. The harm could be caused by any element of the system or environment at every stage of the system life cycle. Four types of possible diagnostic problems will be taken into account:

• harm creation (connected with change of parameters or properties of the analyzed element)
• harm creation (connected with wrong measurement)
• intensification of harm that already exists
• useful interactions' elimination

1. The function of the base plate is to hold the wafer.

Functional group: plate - holds - wafer;

2. The functions of the bridges are:

a. to hold and direct the press-strips.

Functional groups: bridge - holds - strip; bridge - direct - strip;

b. to press press-strip to mask (together with screws).

Functional groups: bridge - presses - press-strip to mask; screw - presses - press-strip to mask;

3. The function of the press-strip is to press the mask to the wafer (together with base plate).

Functional groups: press-strip - presses - mask to wafer; plate - presses - wafer to mask;

4. The function of welding is to join the bridge and the base plate.

Functional group: welding - joins - bridge and plate.

Let us take, for example, the functional group: press-strip -- presses -- mask to wafer. We will consider this group at the different stages of life cycle of the device in order to invent possible failures and harmful effects which are connected with this group.

Stage of manufacture:

The press-strip can be deformed;

The mask can be deformed;

The wafer might have dirty surface;

The surface of wafer can be wavy;

The glass carrier can be cracked;

And so on……..

Stage of storage:

The press-strip can be deformed;

The mask can be deformed;

The wafer might have dirty surface;

The glass carrier can be cracked;

And so on……...

Stage of work:

The press-strip can be deformed;

The mask can be deformed;

The glass carrier can be cracked;

And so on……..

Nearly all of these harmful effects either don't exist or can be explained on the basis of common sense except for the mask deformation on "work stage". Moreover, the effect of thermal expansion of the mask in the sputtering machine was taken into account. That's why the mask was made of Kovar™ instead of stainless steel.

Thus, one of our problems is mask deformation (which will cause harmful effect - metal coating under the mask) inside of sputtering machine. Our problem belongs to the following generic type of diagnostic problems:

Finding the ways to perform the action that causes the harmful effect connected with the change of parameters or properties.

We can transform our problem into the inventive situation of the first type: Perform some function in order to get the harmful result that further is considered as useful.

We have to find a technical system that will perform the function: to deform the mask.

In this case:

The function - to deform the mask;

The object of the function - mask;

The known method of the function's performance - heating the mask (the mask is pressed by press-strips to the wafer, thus thermal expansion can deform it);

UDE - the thermal expansion is too small and it isn't capable to deform the mask.

It is easy to find the following solution: "To use mask coating by other metal in order to cause its deformation" using the DPS algorithm represented in the previous section.

Let us check this solution:

The mask is coated with Cr and then with Cu, so we have the thermal expansion effect of the "bi-metal sandwich" from materials with the different coefficient of thermal expansion

The effect will be realized if:

1. The layer of metal coating on the mask will be thick

or

2. The mask will be thin.

These conditions are present in our case, hence our solution is correct.

7. Conclusion

The correct classification of the diagnostic problems and the proposed algorithm enable us to solve the most important problems that occur during development process of new products and technologies, increase yield and achieve very high reliability of these products and technologies.

REFERENCES

1. G.S. Altshuller "To Find an Idea" (Novosibirsk, Nauka, 1991, 226 p.)

2. S. D. Savransky, "TRIZ" (To be published, 1999, 454 p.)

3. Solution of Research Problems, Eds.: B.L. Zlotin and A. Zusman (Kishinev, 1991).

4. J. Terninko, A. Zusman, B. Zlotin "Systematic Innovation: An Introduction to TRIZ" (Saint Lucie Press, 1998, 150 p.)

5a. G. Frenklach. The Research (Diagnostic) Problems' Classification. The March 1998 issue of the triz-journal. (www.triz-journal.com)

5b. G. Frenclach. The "Diversionary" Method. The April 1998 issue of the triz-journal. (www.triz-journal.com)

6. S. Kaplan, Anticipatory Failure Determination (AFD): The Application of TRIZ to Risk Analysis // Finding Failures before They Find Us: An Introduction to The Theory of Scenario Structuring and The Method of Anticipatory Failure Determination. In "Transactions of the Ninth Symposium on QFD" 1997, p. 195-219.

7. G. Yezersky, Failure Prevention Analysis (FPA): An Application of Systemology-TRIZ to FMEA and Failure Prevention. In "Tutorials of the 1998 Tenth Symposium on Quality Function Deployment ", Novi Michigan.

8. D.H. Stamatis, "Failure Mode and Effect Analysis: FMEA from Theory to Execution" Quality Press, 1995

9. M. Ammerman, "The Root Cause Analysis Handbook : A Simplified Approach to Identifying, Correcting, and Reporting Workplace Errors" Quality Resources, 1997.

10. Guo Quan Huang, "Design for Quality - Principles and Techniques" John Wiley & Sons, 1998.

11. "Team Oriented Problem Solving: Skills Workbook" Fairlane Traning and Development Center (FORD) Second Edition. 1991.

12. P. Mears, "Quality Improvement Tools and Techniques" McGraw-Hill, 1995.

13. S. D. Savransky, Few words about Altshuller's contradiction matrix. The August 1997 issue of the triz-journal. (www.triz-journal.com). See also WWW site http://www.trizexperts.net

About the Authors

Gregory Frenklach became acquainted with TRIZ in 1986-1987. He is the author of 6 patents and about 30 works in TRIZ, value engineering and TRIZ pedagogics. Gregory Frenklach was one of the founders TRIZ-company in Gomel (Belarus) which provided TRIZ-based consulting, value engineering and TRIZ courses for engineers, teachers and students. Now he is a member of TRIZ Experts. Gregory Frenklach was one of the TRIZniks who brought modern TRIZ to Israel and have supported increasing popularity of this methodology in the country. During his work at the Jerusalem College of Technology (Israel) Gregory Frenklach had developed the computer system "Inventor Assistant" for the inventive problem solving support. In addition, Gregory touhgt TRIZ-courses for the college's students. Currently Gregory Frenklach works at AVX Israel as a mechanical engineer of the Maintenance & Equipment Department.

Semyon D. Savransky, a member of the International TRIZ Association, became acquainted with TRIZ in 1980-1981 (due to the extensive and critical study of  the Altshuller's methodology). His encyclopedical knowledge of natural science and electronics  was soon recognized in the TRIZ community.  He has applied TRIZ for  several R & D projects in various high-tech industries and for  pure scientific research. Semyon helds 8 patents, he is the author of about 150 scientific papers  about the development of TRIZ, in physics and materials science. He received  a Ph.D. in Leningrad (1989), and his academic background is split  between Novgorod State University (Russia), University Pais Vasco (Spain) and New York City University (USA).  Semyon is currently the head of the Division of the West Coast Quartz Corporation (California). He is the founder of the Research Center in Novgorod State University - NGPI (Russia) and  TRIZ Experts International Company. Semyon is the Member of the International TRIZ Association.