Interactive Animations as a Tool for Conceptualization - an Example from Semiconductor Devices

Per Lundgren¹ and Lars-Erik Jonsson²

1_{Solid State Electronics Laboratory, Department of Microtechnology and Nanoscience, Chalmers University of Technology, S-412 96 Göteborg, Sweden}

2_{Department of Education, Göteborg University, S-405 30 Göteborg, Sweden}

Abstract

Interactive animations intended to help students understand and visualize physical processes were designed, studied and assessed for use within the frames of a compulsory course in Semiconductor Devices at university level. We report on the difference between intended and perceived animation content from results of individual video recorded interaction sessions. We further describe a full scale incorporation of interactive animations into the curriculum, where the desired effects on student performance were not observed despite a substantial effort put into learning about animations as an educational tool.

1. Introduction

In a traditional semiconductor device course much of the content concerns the transport of charge carriers. Carrier transport is a topic which is neither trivial to teach nor to learn and therefore supporting didactic material is needed [1].

Although it seems obvious that illustrations could powerfully support the conceptualization it has been shown that this is not necessarily true [2]. It has also been shown that it is a delicate matter to successfully utilize animations to illustrate dynamic physical processes in higher education [3].

In a compulsory course on Semiconductor Devices given to second year students at the electrical and computer engineering programmes at Chalmers University of Technology we have observed that the concepts of drift and, in particular, diffusion are problematic in terms of how students perceive and describe these phenomena in relation to how they are used in the semiconductor device field of expertise to describe charge carrier motion in semiconductor devices. Lundgren [4] – utilizing a phenomenographic approach – observed in elective oral mid-term examinations that students displayed a range of problematic conceptions. Furthermore the students were often good at reproducing images they had encountered in class but their ability to relate these images to the proper understanding was often inadequate.

Within the frames of a larger effort at the School of Electrical and Computer Engineering to advance the use of information technology in teaching we launched a project in which a series of interactive animations were constructed and tested in a compulsory university course for electrical- and computer engineers. The overall purpose was to provide students with support and to gain insights about how interactive animations work as a tool for learning. The specific aim was to research how – and if – interactive animations could support the students’ conceptualization of carrier transport, particularly the concepts of drift and diffusion.

In this paper section 2 describes the animations developed and used in the study. Section 3 deals with the methodology employed at various stages of the project and the outcome of these stages are accounted for in section 4. Section 5 is a discussion and section 6 summarizes the content of the paper. Animations can be found at “http://www.medialab.chalmers.se/lide/halvbild/”.

2. Materials

2.1 Diffusion animations

The starting point of the entire project was the problem identified among students to handle the concept of diffusion, and our hypothesis was that an aid to visualize the diffusion process would help students in their efforts to understand the concept.

A computer engineering student was first contracted to find animation examples available on the www and to develop a first set of diffusion animations. This work resulted in a basic set of examples where different aspects, such as dimensionality and context of the animations were varied (see Table 1).

The following two properties were controllable for several of the animation examples:

* Moving objects could collide with each other or not

* Moving objects could at any time change direction randomly or not

The properties could be changed by clicking a button. In the example shown in Figure 1 the inner walls are removed (the inner box opens) when a button is clicked. The ball-theme was varied in four examples with added levels of complexity of the motion.

View Animations at

http://toolearn.portal.chalmers.se/

Figure 1. Screen shot from the animation example used when changing collision and movement randomization controls.

The ball-like objects moving around in a box-like container represent an abstract example. However, there were also more concrete examples where people move in a specific environment (Figure 2). The wider range of more vivid associations that can be stirred by a richer context is an interesting issue. Associations have the potential of being very beneficial if they can link previous intense experiences of diffusion to what is being shown. On the other hand they could be detrimental if they are associated with strong negative feelings.

The use of sound effects in two of the animations was partly included to underline the importance of the diffusion-enabling click to emphasize that "now something happens". The music makes a dramatic change upon clicking at the same time as the diffusion starts. The sound effects also contribute to make these examples richer in context.

The concept of variation in teaching is described by Bowden [5]. In order for the learner to discriminate what is the meaning (relevant content in a given context), there should be opportunities to experience variation of critical aspects (such as collisions on/off). Following Bowden dimensionality was varied. Three classes of animations can be identified in our set of examples:

* plain one-dimensional,

* two-dimensional movement in a three-dimensional environment, and

* three-dimensional movement.

Figure 2. A discotheque/ballroom example of diffusion.

The plain one-dimensional animation was also on the highest level of abstraction with bars representing the number of particles as a function of one single room coordinate. The two-dimensional movement was represented in three different ways. Although the objects move in two-dimensions, diffusion can occur in either just one (Figure 2) or in both dimensions (Figure 1). The objects can furthermore seem to be distributed in three-dimensions, although their movement is really only two-dimensional. Five of the diffusion animations belonged to the two-dimensional class. Two animations were aimed at showing three-dimensional movement, and one of those also displayed three-dimensional diffusion.

2.2 Drift animations

Much of the thoughts on diffusion animations were carried on to the design of drift animations, which were implemented by Chalmers Medialab. The issue of drift is potentially more complex than diffusion, since we have more factors playing a role in the physical process, but at the same time the net effect is more intuitively obvious: if something pulls at something, it moves in that direction. Four examples were generated and subsequently employed in the course. A brief description is given in Table 2. The "Pipe" example is a direct extension of the diffusion animation with the same name.

3. Methods

This study can be divided into (at least) two major phases: the first phase concerns the development of an initial set of animations illustrating the diffusion concept, which were evaluated using a few carefully studied cases of users interacting with the animations under special circumstances described in paragraph 3.2. The second phase comprised setting up an animation event available on the www and integrating it into the curriculum. In this case both diffusion and drift were treated in the animations. The effect of this event was searched for in subsequent oral examinations.

3.1 Questions

Following our general purpose the following issues were in focus in the two phases:

1. What is the quality of the user's perception of the animations in a "neutral" or at least controlled non-teaching context? (phase one)

a. Operationalised: Collision control on/off

b. Operationalised: Random motion on/off

2. What are the effects of course-integrated animations on oral examination outcome regarding references to the specific material presented in the animations? (phase two)

3.2 Procedure

First phase: getting a grasp

Three students agreed to be video recorded while interacting with a set of diffusion animations in the office of the teacher. A pedagogue consultant (LEJ) was observing and taking notes during the sessions. The student was initially welcomed and introduced to the procedure of the session. The purpose of the student's involvement was explained to be for us to collect feedback on how animations produced at Chalmers work as learning tools. The role of the project within the frames of a larger strategic project was explained, as well as the role of the teacher/project leader and the role of the participating pedagogue. The students were not informed that the animations concern the concept of diffusion. Since the three test persons were all first year students at the electrical engineering programme, and since they were exposed to the project leader also as a tutor during their first semester, they had some previous knowledge on what kind of phenomena that were likely to be treated in the animations; during the tutorial sessions we had been discussing the nature of current transport in conductors. The computer program was running from the start, displaying an introductory web page. Three chairs were placed in front of a desk; two of them directly in front of the computer display (CRT) with access to the keyboard and mouse. The camera was rigged to record both the computer display (CRT), the teacher (Int) and the student (User) (Figure 3). In the room there is also a white board.

Figure 3. Schematic view of the set up for video recorded animation sessions.

The user was given technical instructions, e. g. that it was important that she speaks about whatever comes to her mind when interacting with the animations. Her reasoning and thoughts are the focus of the session. Then the camera was turned on and the session started with the interviewer activating the first animation. During the session the user was given questions regarding her interpretation of what was shown in the current animation. The interviewer made control changes, i. e. turning collision and movement randomization on/off, and the user's response was observed. After all the animations had been interacted with, the students were asked to explain the concept of diffusion with the aid of a white board and pencils.

Second phase: for real

The first phase provided substantial and valuable information relevant to the choice of implementation within the course. It was clear from the previous tests that the students would benefit from more supervision when interacting with the animations. The compatibility of the implementation software was also limited with respect to the computer resources available to the students at home or at the student computers at school. We accordingly chose to have the animations appear in the classroom with teachers present.

The animations were used in the curriculum during two consecutive seasons of the course. During the first year the animations were treated during the second half of a four-hour compulsory lab session in a studio where there were 24 computers and workspace for 48 students. The students worked in pairs. Two teachers were present during the session, which lasted for about an hour. The students were given an assignment at the beginning of the session and were informed that some of them would be selected to present their results in front of the class at the end of the session. The assignment was either to select one good and one bad example of both drift and diffusion and to motivate why these were good or bad, or to identity two criteria that need to be fulfilled in order to have diffusion, and to find an example that could show both drift and diffusion counteracting each other. Four animations of diffusion and four animations of drift were available for the students on the www (http://www.medialab.chalmers.se/lide/halvbild). There were also links available to other sites with animations on the www.

The set-up during the second year was somewhat different; the animation sessions were two-hour stand-alone events. Only one teacher was present. In some sessions the students were asked not only to discuss the animations but also to give a physical/mathematical model for the drift and diffusion processes.

The teachers assisted the students by moving around and interacting with the different student pairs each at a time. This interaction was based on the activity that the students were undertaking at the time the teacher reached their workspace. Sometimes the teacher's presence was called for by the students, often not.

The potential impact of this event on the students' resources for talking and reasoning about drift and diffusion, and also to use alternative images and metaphors was to be assessed in the oral examinations in the middle of the course, where approximately 70% of the students go to an elective meeting with a teacher for twenty minutes. Those who pass this event successfully can be rewarded up to two points (a total of eight out of eighteen are needed to pass). During the first year the teachers taking part in the oral examinations were not told in advance to look specifically for any references to the animations, and the oral examination was basically conducted in a traditional manner. When conducted the second year, the teachers were asked to put on record the outcome of specifically asking about the concept of diffusion. The oral examination has been shown to enable phenomenographic studies of student conceptions [4] and should as such constitute an appropriate tool for assessment in this kind of study.

The students also filled in questionnaires on lectures following the studio sessions.

4. Results

4.1 Grasping

The eight animations used in the video recorded interview sessions were all designed to be illustrative examples of diffusion. The three students who interacted with them clearly had different views of diffusion before the session. The first one used the concept of diffusion in the same way as we do, in accordance with what was shown in the animations. The second student had a relatively incomplete conceptualization, where there needed to be a cause of movement to give diffusion, rather than random movement and concentration gradient being sufficient criteria. The last student had not yet formed an understanding of diffusion. The first student could identify that diffusion was the common denominator in the examples. The third student saw a range of different examples that not necessarily illustrated the same phenomenon at all. The conceptualization of the second student was not changed during the animation session. It was clear that in this "non-teaching" context of interacting with the animations as a stand-alone event, the learning outcome was negligible with regards to diffusion conceptualization.

On a more detailed level, the animations proved to be all but obvious in terms of the user perceiving what was actually shown on the screen. Although it is easy to understand that it would be difficult to convey the abstract concept of diffusion by means of animations, it was somewhat surprising that what was thought to be more or less obvious parts of the animation's expression were actually hard to perceive or not perceived at all, as shown in the following examples.

Collision control on/off

These are translated excerpts from the three sessions where the collision control is either activated or turned off.

Int: (collision switched from on to off) Do you notice a difference now?

User 1: No. (Silence). They might be moving a bit faster. I don't know. I don't think I see any difference.

Int: No. Could you open that again?

User 1: Yes, they probably move a little bit faster.

Int: Yes. Is the motion changed in any other way, or...

User 1: Yes, they have time to move in longer trajectories, I think, before they change direction.

Int: Yes.

User 1: Without colliding with something.

It takes User 1 quite a while, and some active pushing from the interviewer before User 1 gives a statement indicating that User 1 at all sees the effect of turning the collision control off. Next is User 2:

Int: Do you notice any difference if I do like this (collision is switched from on to off)?

User 2: (Silence). Yes, they bounce a lot against each other...yes, they take the...well now they are affected by each other too, that they bounce against each other, or they might have done that before too, but now they seem to have different speed too. Before it looked like most of them moved similarly, I mean with respect to speed.

Int: Yes.

User 2: But now it seems to be more irregular too. Sometimes they just turn around. They don't stick to a straight trajectory.

Int: No.

User 2: That is what I can see anyway. (Silence). They kind of turn suddenly without having bounced against anything.

User 2 notices the random change of motion, which was present before as well as after switching the control, and it is interesting that User 2 still reports seeing collisions when these were just switched off. User 3 is last:

Int: Do you notice any difference if I do this (collision switched from on to off)?

User 3: Ordered pattern. It organizes itself, or am I imagining? Yes, it looks more ordered.

Int: Yes.

User 3: It looks like loosely connected circles, which move that...now I don't know if I am imagining this.

Int: No.

User 3: But it looks as if you could freeze this picture and you could find a circle.

Int: Yes, yes.

User 3: So you see here, right, but now it has changed. Sometimes it looks as if its rather well structured.

Int: Yes. Did it appear to be more well structured when I...

User 3: Well, it could be imagining too, since you asked.

Int: Well, otherwise you would not have noticed anything, right?

User 3: I couldn't say, but, yes now it looks more well structured.

User 3 does not notice the change in the collision control. From these three samples it appears to be very difficult for the user to appreciate the collisions between objects on the screen.

Random motion on/off

The next three examples concern turning the random change in motion off:

Int: (Random motion switched off). Do you notice any difference?

User 1: Well, I think that they move a bit in...or they move with different speed do they also. I don't know if they did that before.

Int: No.

User 1: They move...some move very quickly and some move rather slowly.

Int: Do you notice a difference if you open?

User 1: Yes, they spread out much more quickly from that enclosure and they end up in small bunches.

We move on to user number 2:

Int: If I do like this (turn random motion off); does it change?

User 2: (Silence). Now they move...I don't know if they did this before, but now they seem to move all the way out to this three dimensional wall.

Int: Mmm.

User 2: I don't think they did that before.

Int: Not?

User 2: Then they just moved towards the first frame of the border, now they seem to go all the way out. I think, I am not quite sure of it, I think so. Then it's the same when they...but now they seem to move straight ahead until they are affected by something - another ball or a wall can touch them.

So user 2 declares to be observing the collisions, which were actually switched off, but also seems to appreciate the fact that the randomization of motion is no longer present. The final example in this set:

Int: If I do like this (turn random motion off)?

User 3: Well. (Silence). I wonder whether anything changed or if I simply wasn't ready. (Silence). Well.

Int: Did you notice a difference?

User 3: No, that was difficult.

Judging from these few but telling examples, we cannot assume that the user will perceive the appearance on the screen, or the behavior of the objects in the animation in the intended way. It is important to note that the users had no indications or clues as to what the changes of the controls would imply for the animation; they were not guided to look for any particular changes. We can conclude that what is intended to be shown in the animation is not automatically perceived by the user. Even those properties of the animation that seemed rather obvious to the interviewer were not at all obvious to the user.

A parameter that was not significantly varied for the diffusion examples was the level of interaction. The controls that enabled changing the properties of the animated objects, such as randomization of motion, were in the hands of the interviewer. The user had one main button to click, and this click activated the diffusion process. It was unexpected that the users were almost reluctant to use this option. Maybe this was an effect of the experimental set-up, where the presence of the interviewer hampered the user's power of initiative. The user was anyhow fully occupied with reflecting on and analyzing the scene in view as just a viewer, not as an interacting agent. We would have thought that the focus of attention would naturally fall on the effects of "making the click". It did not in these cases, and again we conclude that what we take as an obvious main feature of the interactive animation is not necessarily perceived as such by the user.

As an example of the impact of the associations invoked by the context, we can take the following quote from one of the users when looking at the example:

"Oh, gosh! It just...yuck...reminds me of some kind of disco. Its really crowded and gets hot."

The spontaneous dislike brings to focus the strong negative associations of the user, who appears to relate to what is remembered from experience rather than what we want to show.

4.2 Realizing

An elective oral examination at mid-term (a few weeks after the animation exercise) was used to detect the effect of including animations in the curriculum. After the oral examinations of the first year attempt the teachers were asked whether they had perceived any difference in the ways students expressed themselves or treated the concepts of drift and diffusion during their discussions. No such perception was reported. There were thus no obvious traces of the learning experience invoked by the interactive animation session earlier in the course, at least not when searched for in the way the students reasoned around the concepts dealt with in the animations. The teachers were not prepared to particularly look for any changes in student response, which might have made them inclined to interpret or interfere with the oral examination in an undesired way. The oral examinations were executed as normal, and no difference compared to normal was detected.

When using the oral examinations for animation evaluation the second year, the examining teachers were instructed to explicitly ask half of the students taking the examination to explain diffusion. The teachers were then to fill in a form describing the student's answer. The options for descriptions that the teachers could choose to mark on their form are given in Table 3. In total 136 forms were collected, and only in three cases did the students refer to the animations of the exercise when explaining diffusion, which is a reflection on the direct impact of the animations on the students ability to articulate her conception on diffusion. This impact appears to be very weak. Although the overall result is satisfactory, with 62% giving reference to both random movement and concentrations gradients in their explanations and only 24% mentioning action of force in conjunction with diffusion, it is not obvious how much the animations have contributed to this result.

When looking at the answers an apparent correlation can be found between the description of the students' answers and whether they were given a physical modeling assignment during their animation interaction or not. Table 4 shows the distribution of different categories of answers for the two different groups of students who either did or did not receive a modeling assignment during the exercise. It looks like the students who were not asked to do modeling performs slightly better; a larger fraction gives the desired explanation with concentration gradients and random motion in this group than for the modeling group, and fewer give the incorrect association to force. There does not seem to be a reason to suspect that the modeling inclusions results in significant improvement of the outcome of the animation exercise. What is more interesting is that the oral form comments that correlate most strongly to the student's final course score is whether she has given a simile of her own (comment 6), or whether she uses the white board when explaining (comment 11).

In the first year the session when students interacted with the animations occurred in the second half of a nominally four hour lab session. It is likely that the students generally felt that the animation interaction event was overshadowed by the first half of the exercise, where they do computer simulations of current measurements on semiconductor materials. They are also somewhat fatigued by this exercise. In a feedback questionnaire the students answered on open questions regarding the lab that the animation part could have been treated in less time than was allotted. This feedback can be taken as an indication that the relative importance of the animations was perceived to be small compared to the simulation part. The questionnaire was handed out during two separate lectures (with different participants) and one single questionnaire copy was circulated at each lecture. On both occasions the students themselves suggested a comment on the time for the exercise being too long. It appears that we have failed to provide a situation or learning context that engage the students sufficiently in their interaction with the animations. Half of the students agreed to the statement that the lab session in its entirety was stimulating, and a spontaneous comment that it was "good and fun" also appeared. The questionnaire results are summarized in Table 5. Similar results are obtained for the second year (48% of 250 agree that the exercise is stimulating).

5. Discussion

Judging from the outcome of these different phases of using the interactive animations it is quite clear that the effect of these on student learning is not trivial to detect. Whether this is due to invalid or incomplete assessment, improper integration of the animations in the course, bad animations or the inadequacy of the interactive animation medium for this kind of learning objectives is hard to say, but we can at least state that each of the first three issues above were given significant attention during the project which constituted an extraordinary effort of course development (a total effort of at least 500 man hours). If interactive animations per se would constitute a robust, simple and straightforward way of affecting student understanding as we measure it, then we should have seen something. Hence we conclude that using interactive animations to aid conceptualization in higher education is a non-trivial pursuit, and that it remains to be shown what designs and what kind of integration into the curriculum will give the desired effects on learning.

It is obvious when working with the animations that the reaction among the students and teachers is very positive. Students become involved, show an interest and make positive comments which goes also for the teachers. The animations make a contrast to other learning situations and this is a positive factor on its own. Staff colleagues see a new tool to use in their teaching and this provokes new thoughts that influence the teaching in more ways than just adding animations.

6. Summary

We have developed, tested and run interactive animations intended to help students conceptualize on drift and diffusion. Closely studied sessions where single users were video recorded when using animations accompanied by an interviewer disclosed several issues regarding how the animations were perceived, where it appeared that animation features were much less obvious than expected. Although well received in terms of student attitudes, we were unable to clearly perceive any effect on student understanding of drift and diffusion from integrating the use of animations in the curriculum.

References

[1] S. Karmalkar, IEEE Transactions on Education, Vol. 42, p. 323, 1999.

[2] M. Hannus and J. Hyönä, Contemporary Educational Psychology, Vol. 24, p. 95, 1999.

[3] U. Harms, H. Krahn, and G. Kurz, European Journal of Engineering Education, Vol. 23, p. 503, 1998.

[4] P. Lundgren, International Journal of Engineering Education, Vol. 14, p. 294, 1998.

[5] J. Bowden and F. Marton, The University of Learning, Kogan Page Limited, London, 1998.

Biographical sketches of authors

Per Lundgren is currently vice-dean of undergraduate education at the Department of Microtechnology and Nanoscience at Chalmers University of Technology in Göteborg. He received his Ph. D. in solid state electronics at Chalmers in 1996 and has been involved in undergraduate courses on semiconductor physics and semiconductor devices since 1991.

Lars-Erik Jonsson is Licentiate of Philosophy and works both at the Department of Education, Göteborg University and at the IT-University in Göteborg. In his research he explores the appropriation of information technology in educational practice from a socio-cultural perspective. He has more than 25 years of teaching experience.

Tables

Animation	Context	Dim.	Features
Example 1	Blue balls in sky	2D
Example 2	Blue balls in container on table	2D	3D distribution of balls, but 2D-movement
Example 3	Blue balls in container on table	3D	2D-diffusion
Example 4	Disco	2D	Rich context, music
Example 5	Pipe	2D	1D-diffusion, shows current flow
Example 6	Bar diagram	1D	Abstract context
Example 7	Star Wars	3D	Rich context, music
Example 8	Red balls	2D	Poor context

Table 1. Listing of the diffusion animation examples.

Animation	Context	Dim.	Features
Example 1	Blue balls in container	2D
Example 2	Red and blue balloons	3D	Non-interactive
Example 3	Ants in rain	2D	Non-interactive
Example 4	Pipe	2D	1D-transport, shows current flow

Table 2. Listing of the drift animation examples.

Comment #	Comment
1	Explains in own words
2	Recites from memory
3	Explanation comprises concentration gradients
4	Explanation comprises random motion
5	Explanation comprises urge, wish or force or similar concepts
6	Gives a simile of ones own
7	Refers to the animations of the exercise
8	Refers to an analogy from a book
9	Gives a correct mathematical expression
10	Gives an incorrect mathematical expression
11	Draws on the white board

Table 3. Teachers options for describing diffusion answer.

Answer containing comment	#	Fraction of non-modeling	Fraction of modeling	Fraction below 8 points	Average points	Fraction above 10 points
Any	136			22%	8,7	24%
3+4	84	67%	56%	21%	9,0	29%
5	32	20%	27%	25%	8,7	22%
6	16	11%	14%	6%	9,8	31%
1+3+4+6	9	7%	7%	0%	10,7	44%
11	7	7%	3%	14%	10,0	43%

Table 4. Second year results from the oral examinations.

The last three columns give the fraction of students with a total course score of less than 8 points (=failure),

the average points, and the fraction of students with marks above the pass standard respectively.

Statement	% that agree (total 150 answers)
The exercise is stimulating and stirs interest	49
"Good and fun lab"	13
"Could have been much shorter"	33

Table 5. Results from questionnaire first year.

Lundgren, Intercative Animations as a Tool...