Teaching quantum mechanics on an introductory level

Rainer Müllera) and Hartmut Wiesnerb)

Lehrstuhl für Didaktik der Physik, University of Munich, Schellingstrasse 4, D-80799 München, Germany
共Received 25 July 2001; accepted 9 November 2001兲
We present a new research-based course on quantum mechanics in which the conceptual issues of
quantum mechanics are taught at an introductory level. In the context of virtual laboratories, the
students discover from the very beginning how quantum phenomena deviate from our classical
everyday experience. The results of the evaluation of the course show that most of the students
acquired appropriate quantum mechanical conceptions, and that many of the common
misconceptions encountered in traditional instruction have been avoided. © 2002 American Association
of Physics Teachers.
关DOI: 10.1119/1.1435346兴

I. INTRODUCTION cuss the foundations of quantum mechanics at a moderate

mathematical level. An example of a university course ex-
Quantum mechanics has forever changed the physicists’ plicitly devoted to nonphysics majors is the Visual Quantum
picture of the world. At the beginning of the 20th century, the Mechanics project,5 where a hands-on approach to the appli-
advent of relativity and quantum mechanics marked not only cations of quantum mechanics is pursued.
the discovery of just another new theory, but an entirely new
framework for all of physics. Relativity changed our ideas of II. RESEARCH ON STUDENTS’ CONCEPTIONS
space and time, and quantum mechanics introduced indeter-
minism, probabilities, and nonlocality into the foundations of It is known that after traditional instruction, students are
physics. likely to show classical misconceptions and to confuse clas-
Quantum mechanics shapes our view of nature in a funda- sical and quantum notions. Given the counterintuitiveness of
mentally new way. We think that not only physicists should quantum mechanics, these misconceptions are not surprising.
have the privilege to understand how the world works. Edu- To avoid them and lead the students to a correct understand-
cated citizens should at least have the possibility to become ing of quantum mechanics, it is important to know the com-
acquainted with the strangeness and beauty of quantum phe- mon misconceptions that traditional instruction is likely to
nomena. However, most students who do not major in phys- promote.
ics never have a chance to learn about the conceptual issues There have been a number of investigations of students’
of quantum mechanics. For example, in the German Gymna- misconceptions and their difficulties in understanding quan-
sium 共whose upper level is comparable to the first two years tum mechanics. Much of the early work came from the
of college in the United States兲, atoms and quanta are stan- Frankfurt, Bremen, and Berlin groups in Germany 共for sur-
dard parts of the curriculum. However, the emphasis is on veys in English see Refs. 6 –9, where the courses developed
aspects such as the photoelectric effect or Bohr’s atomic by the German groups are also discussed兲. Further research
model, which do not really probe the classical conceptions has been carried out by Mashhadi,10 Styer,11 Johnston, Craw-
prevalent in the students’ minds. ford, and Fletcher,12 Bao, Redish, and Steinberg,13 and
Our new course on quantum mechanics deals with the Ireson.14
conceptual questions of quantum mechanics. It is addressed Here we want to report on the results of our own investi-
mainly to nonphysicists. Special emphasis is placed on quali- gations which up to now have been published only in the
tative reasoning. Physicists can rely on their knowledge of German literature. In our first research project,15,16 523 Gym-
the quantum mechanical formalism to overcome conceptual nasium students answered a questionnaire on their concep-
difficulties, but nonphysicists do not possess such a strong tions of quantum physics after instruction. In addition, 27
supporting basis. Therefore, conceptual clarity is even more students were interviewed orally. The interviews lasted about
important. The strange and counterintuitive phenomena of 1 h. The questions ranged from fact reproduction 共‘‘How
quantum mechanics cannot be incorporated in a coherent would you measure an atomic spectrum’’兲 to interpretational
cognitive picture without the aid of carefully chosen basic issues such as their view of determinism/indeterminism. In a
concepts that help to organize them. Our strategy is to let the second project,17,18 37 university students 共future physics
students discover some of the exciting and bizarre quantum teachers兲 were interviewed in a similar manner. It was found
phenomena that deviate from our classical everyday experi- that 52% of them had already heard about quantum physics
ence. At the same time, we want to provide a conceptual in school, 79% had attended a theoretical quantum mechan-
framework within which a solid understanding can be con- ics lecture. It is remarkable that both groups gave very simi-
structed. lar answers. This similarity indicates that the results can be
There are several books that follow a semiqualitative ap- considered to be typical. In the following, we give an over-
proach to quantum mechanics. First of all, there are the chap- view of the main misconceptions found in our investigations.
ters on the double-slit experiment in Feynman’s famous The first three questions 共from Ref. 16兲 show how students
lectures,1 which have had a great influence on most of the distinguish classical and quantum objects.
subsequent attempts to teach the conceptual aspects of quan- 共1兲 What are the essential properties of classical objects?
tum mechanics. From the more recent approaches we men- The student responses can be categorized into the following
tion the books by Rae,2 Albert,3 and Silverman4 which dis- items 共multiple replies possible兲: 共i兲 mass, weight 共85%兲, 共ii兲

200 Am. J. Phys. 70 共3兲, March 2002 http://ojps.aip.org/ajp/ © 2002 American Association of Physics Teachers 200
size, volume, shape 共43%兲, 共iii兲 velocity, movement 共38%兲, the starting point of the discussion. Because most of the stu-
共iv兲 momentum 共27%兲, 共v兲 position 共15%兲, 共vi兲 density dents interviewed for this study had quantum physics courses
共15%兲, 共vii兲 energy 共12%兲. It is remarkable that nearly all in high school as well as in the university, it is legitimate to
students mentioned mass, but only few mentioned position. say that the Bohr model is a very dominant and stable con-
The 共dynamical兲 property velocity/momentum is considered ception.
more important than position or energy. Although it is not compatible with the quantum mechani-
共2兲 What are the essential properties of quantum objects? cal conception of the atom, the Bohr model may be inevi-
共i兲 Mass 共37%兲, 共ii兲 charge 共37%兲, 共iii兲 velocity/momentum table as an intermediate step.19 Possibly the lack of an easy
共37%兲, 共iv兲 energy 共26%兲, 共v兲 spin 共22%兲, 共vi兲 energy levels/ visualization of the quantum mechanical model forces stu-
quanta 共15%兲, 共vii兲 position not exactly determined 共11%兲, dents to stick to this model. If this hypothesis is true, the goal
共viii兲 no absolute mass 共11%兲, 共ix兲 de Broglie wavelength of our instruction should not be to erase the Bohr model in
共7%兲. For quantum objects, mass is not as dominant as for the students’ minds, but to convey the conscious use of
classical objects. Charge is mentioned often. Category 共vii兲 physical models and let them have insight into the models’
indicates the conception of ‘‘smeared’’ quantum objects, limitations.
which is discussed in more detail below. 共6兲 Permanent localization 共from Ref. 17兲. The students
共3兲 What is the main difference between a classical and a were asked, ‘‘Does an electron in an atom have a definite
quantum object? 共i兲 For many students 共30%兲, there is a position at each moment of time?’’ The answer categories
smooth transition between quantum and classical physics. As were 共i兲 the electron has a definite but unknown position
objects become smaller, quantum behavior shows up more 共21%兲. 共S1: ‘‘Yes, it has to be somewhere, but it isn’t acces-
clearly. 共ii兲 26% of the students argue in terms of dualism or sible through a measurement.’’ S2: ‘‘I would say in principle
the necessity of model descriptions. 共iii兲 Quantization, espe- it has a definite position, we just don’t know it. That’s how I
cially energy quantization 共19%兲. 共iv兲 Large velocities are imagine.’’兲 共ii兲 The electron has a position but no trajectory
possible for quantum objects 共15%兲. 共v兲 Quantum effects 共due to insufficient knowledge of initial conditions兲 共7%兲.
such as the Compton effect, spin, interference. 共vi兲 Quantum 共iii兲 Localization in a region with some probability 共25%兲.
objects do not possess a position property 共11%兲. 共vii兲 Other 共S8: ‘‘It’s like that, they have no definite position, to my
answers 共7%兲. mind, they are just located arbitrarily somewhere in a certain
The following questions give an overview of the common region.’’ S32: ‘‘You cannot localize it that precisely, you can
conceptions and misconceptions of quantum objects such as only give a probability of finding ...’’兲 共iv兲 No definite posi-
photons and atoms. To give a more vivid illustration, we tion because of the uncertainty relation 共18%兲. 共v兲 Other
have included typical student answers to most categories. 共11%兲, indifferent 共18%兲.
共4兲 What do you mean by ‘‘photon’’ 共from Ref. 16兲? 共i兲 共7兲 Heisenberg’s uncertainty relation 共from Ref. 17兲:
One third of the students described a photon as particle of Question: What is the meaning of ⌬x and ⌬p? 共multiple re-
light that has wave as well as particle properties; 17% mis- sponses allowed兲. 共i兲 Measurement uncertainties 共15%兲.
interpreted the wavy line that symbolizes a photon in many 共S15: ‘‘Suppose you know the error ⌬x. Then you can deter-
Compton effect diagrams as the trajectory of the photon. mine the minimum error you have done in the momentum
关Student P2: ‘‘Photon is denoted a light quantum, a particle, measurement.’’兲 共ii兲 Disturbance during measurement: posi-
and it moves in the form of a wave.’’ Interviewer: ‘‘The tion measurement influences the particle’s momentum
photon itself?’’ P2: ‘‘In the form of a wave forward 共draws 共21%兲. 共S13: ‘‘When I measure the position very precisely, I
wavy arrow兲.’’兴. 共ii兲 25% of the students remark that photons alter the momentum.’’兲 共iii兲 ‘‘Regions of localization’’
do not have a rest mass 共‘‘they only have a mass when they 共18%兲, for example, spatial region where the particle is con-
move at the velocity of light’’兲. 共iii兲 17% define a photon as fined; width of the wave function. 共iv兲 Interval within which
an energy quantum, and 共iv兲 8% state that a photon is emitted the exact value lies with some probability 共18%兲. 共S18: ‘‘It
in the transition of an atom from an excited state to the is, so to speak, the probabilities of the momenta at this place.
ground state. This is the most precise statement about the momentum. I
can only say the momentum lies in the interval between p
共5兲 How do you conceive electrons in an atom 共from Ref.
⫾⌬p.’’兲 共v兲 Standard deviation of a statistical distribution
17兲? 共i兲 Bohr’s atomic model or planetary model 共17%兲.
共13%兲. 共S21: ‘‘If I repeat an experiment several times and
共S31: ‘‘There are circles ... around the nucleus ... just orbits.
measure position and momentum, I don’t get always the
They are circles. And the electrons are on different orbits.
same, i.e., if I have identical initial conditions, I don’t get
They move on them and they can jump from one orbit to
always the same x and the same p, but it varies. If I graph it
another ... if they get more energy, they can jump to a higher
I get a standard deviation.’’兲
orbit.’’兲 共ii兲 Bohr’s model with cautionary remarks 共24%兲.
共S18: ‘‘The orbits ... I still have that picture when I think of
an atom. One is told that it’s not correct, but one is so used to III. OUTLINE OF THE COURSE
it and, after all, it is employed again and again.’’兲 共iii兲 Con-
The results presented in Sec. II form the empirical basis
crete ideas of ‘‘clouds’’/smeared charge 共14%兲. 共iv兲 ‘‘Orbit-
for the development of our course. We want to avoid classi-
als’’ with probability distribution 共38%兲. 共S29: ‘‘It’s the wave
cal misconceptions and help our students to construct a cor-
function that represents the particles, there is the theory of
rect quantum mechanical understanding. For these reasons
orbitals, the orbitals can be represented in space. Then you
we concentrate on those features of quantum mechanics that
know where the electrons are approximately and the whole
are radically different in comparison to classical mechanics.
thing works with the probability interpretation’’兲.
We therefore focus on the following aspects:
The two dominant conceptions are the two variants of
Bohr’s model 共together 41% 兲 and the picture of orbitals 䊉 Born’s probability interpretation is introduced early and
共38%兲. It is remarkable that even if quantum mechanical used throughout. In introductory courses, wave–particle
ideas are mentioned, Bohr’s model is almost always used as duality is often characterized as the main mystery of quan-

201 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 201
 tum mechanics. In contrast, we point out that there is noth-
ing mysterious about wave–particle duality once a proper
understanding of the probability interpretation has been
achieved.
䊉 A major new feature of quantum mechanics is that classi-
cally well-defined dynamic properties such as position,
momentum or energy cannot always be attributed to quan-
tum objects.20,21 If an electron is not in a momentum eigen-
state, it does not possess the ‘‘momentum’’ property 共at
least according to the standard interpretation兲. Similarly, an
electron in an atomic orbital 共an energy eigenstate兲 does
not possess the ‘‘position’’ property. We consider this no-
table feature as a central element of quantum mechanics.
Its discussion therefore takes a prominent place in our
course.
䊉 The measurement process has perhaps led to more debate
than any other topic in quantum mechanics. In contrast to Fig. 1. Structure of the basic course.
classical mechanics, measurement can no longer be con-
sidered as a passive reading of pre-existing values. In
quantum mechanics, measurement is an active process. of the spiral, we consider electrons in the double-slit experi-
There is a difference between ‘‘to possess a property’’ and ment. The insight gained with photons is deepened at a
‘‘to measure a property.’’ The special role of the measure- higher level. The probability interpretation is formulated
ment process comes to light in the process of state reduc- qualitatively with wave functions, and the concept of state
tion and is illustrated, for example, by Schrödinger’s cat superposition is introduced. The basic course ends with a
paradox. discussion of more complex issues such as state reduction,
complementarity, Schrödinger’s cat, and decoherence.
Although we consider nonlocality, the Einstein–
Podolsky–Rosen 共EPR兲 paradox, and Bell’s inequality as im- IV. PHOTONS
portant as the subjects above, no attempt has been made to
include them in the course. The primary reason is time limi- We now discuss the contents of the course in more detail.
tation. A. Photoelectric effect
Conceptual clarity is a vital condition for the success of
the course. We therefore base our course on the ensemble The course starts with the photoelectric effect, its explana-
interpretation of quantum mechanics,21–23 according to tion in terms of photons, and the determination of Planck’s
which the predictions of quantum mechanics apply to en- constant. This topic is fairly standard and needs no further
sembles of identically prepared objects. In our view, this in- explanation.
terpretation provides a clear and comprehensible way of talk-
ing about quantum phenomena. Similarly, the idea of state B. Preparation of dynamical properties
preparation20–25 helps one to construct a conceptual frame-
As stated earlier, the notion of preparation plays a major
work that serves as a basis for a deeper discussion.
role in our course. Already in classical physics, the prepara-
The course consists of two parts with different goals. The
tion of initial conditions is important. For example, to test
emphasis of the first part, or basic course, is on purely quali-
the law of projectile motion, one needs to prepare definite
tative reasoning. Students explore the foundations of quan-
values of position and velocity for the projectile. In this
tum mechanics without the difficulties introduced by the for-
stage, we define preparation as the systematic production of
malism. Simulated laboratories provide an environment for
a dynamical property of a classical or a quantum object 共this
their experiences. They are confronted early with the strange
definition is extended later to state preparation兲.
behavior of quanta and the central aspects of interpretation
Preparation of properties can be illustrated nicely with the
are discussed. In the second part, the advanced course, an polarization of light. A horizontally oriented polarization fil-
introduction to the quantum mechanical formalism is given. ter is a device that produces light with the ‘‘horizontal polar-
It is intimately linked with the discussion of the quantum ization’’ property. A second polarization filter can serve as a
mechanical interpretation given in the basic course. test for this property. If it also is oriented horizontally, nearly
The division into two parts allows the course to be easily all of the light passes through it showing that the light pos-
adapted to various demands. For example, in a course for sessed the property horizontal polarization. In contrast, a test
liberal arts majors, we would stop after the basic course. with a vertically oriented polarization filter shows a negative
Engineering science students 共or, in the German school sys- result: no light passes through.
tem, the Leistungskurse兲 would obtain a first introduction
into the more formal elements of quantum mechanics in the
C. Wave and particle behavior in a Mach–Zehnder
advanced course. The two parts of the course will be dis-
interferometer
cussed in detail in the following sections.
The structure of the basic course is summarized in Fig. 1. The Mach–Zehnder interferometer sketched in Fig. 2 is
We proceed in a spiral fashion. First, we introduce photons the playground for the exploration of the quantum properties
and give a qualitative discussion of wave–particle duality, of light. We have developed a program that simulates a vir-
the probability interpretation, and the nontrivial notion of a tual laboratory in which all the experiments needed in the
dynamic property in quantum mechanics. In the second turn course can be performed by the students.26

202 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 202
 Fig. 2. Mach–Zehnder interferometer.
Fig. 4. No interference pattern emerges with orthogonal polarization filters.

The path lengths in the two arms are slightly different so

that a circular interference pattern appears on the screen The next experiment shows that the situation is even more
when laser light is sent through the interferometer. The stu- weird. An obvious objection is that a photon could split into
dents know that interference is a characteristic of wave be- two parts at the beam splitter, go through the interferometer
havior. They are next confronted with an experiment where separately, and then recombine somehow at the end. That
single photons are sent in. The interference pattern gradually there is no such splitting can be shown if the polarization
builds up from the particle-like detection events of single filters are replaced by photon detectors. If single photons are
photons. For the first time, the students see wave and particle incident on the first beam splitter, it turns out that the detec-
behavior in the same experiment. The result shows that nei- tors never click simultaneously. A photon is always detected
ther wave nor particle model suffices alone to explain the as a single entity; parts of a photon are never found. It is
experimental results. A satisfying model must incorporate as- remarkable that this experiment has been carried out in the
pects of both. laboratory.27 Although its outcome sounds fairly obvious
The photon is detected as a localized object on the screen. from a photon point of view, it is one of the few experiments
It is natural to ask whether it is similarly localized within the that clearly contradict semiclassical theory 共quantized atoms
interferometer. Or, formulated in the language of dynamical plus classical light兲.
properties, does a photon in the interferometer possess the Together, these experiments show that a single photon is
position property? as strange an object as one can imagine. The results cannot
To answer this question, we place a polarization filter in be explained by any classical model. They hint to the neces-
each of the interferometer arms. If both are oriented horizon- sity of exploring the quantum mechanical measurement pro-
tally, the interference pattern gradually builds up as in the cess in more detail.
previous experiment 共see Fig. 3兲. The same is true if both are
oriented vertically.
However, if the filters are oriented in orthogonal direc- D. The probability interpretation of quantum mechanics
tions, a different situation emerges 共see Figs. 3 and 4兲. As
before, each photon leaves a localized trace at the screen. But As we have mentioned, a basic observation in an interfer-
from these traces, no interference pattern emerges. A struc- ence experiment with single photons is that the pattern on the
tureless distribution develops instead. Photons are found in screen builds up from the ‘‘hits’’ of single photons. It is
places where there were interference minima in the previous legitimate to ask whether these positions are predetermined
experiments. as in classical physics and can be predicted from the initial
What does this result mean for our conception of the pho- conditions. In this stage of the course, the students learn that
ton? Imagine that a photon was a localized object traveling one cannot predict the position of a single hit, but that it is
on exactly one of the two arms through the interferometer. If nevertheless possible to make accurate predictions for the
this idea were correct, the photon could interact with just one statistical distribution of many hits. This observation is gen-
of the polarization filters. But to determine whether it is en- eralized to the following important statement: Quantum me-
titled to land on the position of an interference minimum, it chanics makes statistical predictions about the results of re-
needs information about the orientation of both filters. The peated measurements on an ensemble of identically prepared
photon has to ‘‘know’’ whether they are parallel or orthogo- quantum objects. This preliminary version of the probability
nal. interpretation is later, in the context of electrons, formulated
If we exclude action-at-a-distance arguments, we have to more precisely in terms of the wave function.
give up the picture of a photon as a localized object travers-
ing the interferometer. In the interferometer, the photon does
not possess the position property. V. ELECTRONS
The second part of the basic course is devoted to electrons.
Here, the insights gained with photons are broadened and
deepened at a higher conceptual level.

A. Wave behavior of electrons

The demonstration of electron wave behavior with an
electron diffraction tube is a standard experiment that does
not need to be described here. The students obtain the de
Fig. 3. Parallel polarization filters do not prevent the emergence of the Broglie relation from the analogous relation for photons and
interference pattern. confirm it experimentally with the diffraction tube.

203 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 203
 Fig. 5. Simulation program for the double-slit experiment.

B. Double-slit experiment and probability interpretation

Since Feynman’s lectures,1 the double-slit experiment is a
basic ingredient in many quantum mechanics courses. We
have written a computer program to interactively simulate
the double-slit experiment 共see Fig. 5兲.28 Electrons 共or other
objects兲 are emitted by the source on the right, pass through
the double slit, and are detected on a screen 共the light bulb
shown in Fig. 5 is needed later in the discussion of the mea-
surement process兲. Many of the experimental parameters can
be varied by the students.
In our course, we first use the double-slit experiment to
resolve the problem of wave–particle duality in terms of the
probability interpretation. We consider single electrons pass-
ing through the apparatus. The students see that in complete
analogy with the photon case, the interference pattern gradu-
ally emerges from single electron detection events. Again,
only statistical predictions are possible.
Next we go one step further. We introduce the wave func-
tion to describe the state of electrons in a completely quali-
tative way in analogy with water or sound waves. For ex-
ample, the wave function behind the double slit can be
visualized as a superposition of two cylindrical waves
Fig. 6. 共a兲 Double-slit interference; 共b兲 superposition of two single-slit pat-
emerging from the slits: ␺ ⫽( ␺ A ⫹ ␺ B 兲/冑2, where A and B terns.
label the slits.
An essential point for the interpretation of quantum me-
chanics is Born’s interpretation of 兩 ␺ (x) 兩 2 as the probability
density of finding an electron at the position x in a measure-
ment. With this interpretation, the duality of wave and par- periment. This conflict is very counterintuitive, and we can-
ticle, which is often so much emphasized in popular ac- not expect the students to accept such a far-reaching
counts, is no longer a mystery: The wave function spreads in conclusion at once. Therefore we present analogous reason-
space much like a classical wave and shows typical wave ing for electrons as well. The argument has been published in
phenomena such as superposition and interference. Math- similar form many times in the literature so we will give only
ematically, interference arises from the cross terms in the a brief sketch. Suppose the electron goes as a localized entity
square of ( ␺ A ⫹ ␺ B 兲. However, when an electron is detected, through one definite slit. Then it cannot ‘‘know’’ whether the
that is, when a position measurement is made, it is always other slit is open or closed. Its final position on the screen
found localized at a certain position. The statistical distribu- should not be influenced by this fact. Therefore, the same
tion of detection events can be calculated if the wave func-
pattern on the screen should appear if 共a兲 both slits are
tion is known. With the Born interpretation, wave and par-
opened together for a time t, or 共b兲 first one slit is opened for
ticle behavior which seemed to be incompatible are captured
in a single picture. the time t, then the other one. However, experiment shows
that it does matter whether 共a兲 or 共b兲 is realized. Double-slit
interference is observed in case 共a兲, whereas a superposition
C. Position property of electrons
of two single slit patterns is observed in case 共b兲 共see Fig. 6兲.
In the interferometer experiment discussed in Sec. IV we This result demonstrates that the initial assumption was
discussed how the conception of a photon as a localized ob- wrong. In this experiment, electrons do not behave like lo-
ject with a well-defined position leads to conflicts with ex- calized objects. They do not possess the position property.

204 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 204
D. Position measurement and the measurement postulate The interaction with its natural environment 共for example,
photon or air molecule scattering兲 renders the cat effectively
The next major discussion is on the quantum mechanical classical. That photon scattering may inhibit interference has
measurement process. We start directly from the previous already been observed in the double slit experiment with the
experiment and ask the seemingly innocent question: What is light bulb. This experience can be used to make the decoher-
the result of a position measurement if the electrons actually ence mechanism plausible.
do not possess the property position? The experiment we use
to answer this question has been conceived by Feynman,
although we interpret it slightly differently in our course. F. Heisenberg’s uncertainty relation
The light bulb in the double slit simulation program sym-
bolizes a position measurement device. It emits light that is It has been mentioned that there are many misconceptions
scattered by the electrons that pass the slit. Eventually, the of the Heisenberg uncertainty relation. In our view, its clear-
scattered light ends in our eye or a detection apparatus, and est formulation is as a statement about the preparation of
we can infer from the direction of incidence where it was quantum objects. It restricts the possibility of simultaneously
scattered. preparing certain pairs of observables on an ensemble of
If we turn on the light bulb, we see little light flashes quantum objects.23 We start from an arbitrarily prepared en-
behind one of the slits 共visible in Fig. 5兲. They give the result semble of electrons described by the wave function ␺. We
of the position measurement. The electron is always found now take a subensemble of this ensemble and perform many
behind exactly one of the slits. For a single electron, we position measurements on its members. We obtain a statisti-
never see the light flashes at two or more positions simulta- cal distribution of measurement results and call the standard
neously 共see the analogous result for photons in Sec. IV兲. deviation of this distribution ⌬x. Now we take another sub-
This result seems to contradict our previous finding that an ensemble and perform a large number of momentum mea-
electron does not possess the position property. But this ap- surements to obtain ⌬p. The uncertainty relation states that it
parent contradiction once more underlines the special status is not possible to prepare a state ␺ such that the product of
of a measurement in quantum mechanics. It is resolved by ⌬x and ⌬p is smaller than ប/2. This way of reasoning shows
the fundamental measurement postulate: In each measure- that the uncertainty relation is not a statement about simul-
ment, a definite value for the measured observable is found. taneous measurements or the mutual disturbance of two mea-
Later, in the advanced part of the course, we make a connec- surements. Quantitatively, the uncertainty relation can be il-
tion to the idea of an eigenvalue: The possible results of a lustrated by the well-known example of electrons incident on
measurement are the eigenvalues of the measured observ- a single slit.
able. The reason why eigenvalues are interesting in quantum
mechanics is that they are the possible results of measure-
ments. VI. AN INTRODUCTION TO THE QUANTUM
On the other hand, this experiment shows that there is a MECHANICAL FORMALISM
difference between ‘‘to possess a property’’ and ‘‘to measure
a property.’’ If we find an electron behind one of the slits, we In the above we presented the first part of our course,
cannot assume that it did possess the position property al- which is addressed to students who are not likely to choose
ready before the measurement. A definite position value is physics at the university. The advanced course is aimed at
only realized in the measurement. students with a special interest in physics 共Leistungskurse兲.
In this part of the course we give an introduction to the
formalism of quantum mechanics and its applications at a
E. Advanced topics in the interpretation of quantum mathematically very basic level. The discussion is based on
mechanics the qualitative understanding gained in the basic course and
With the double slit experiment, some other important as- takes up the ideas introduced there 共such as preparation or
pects of quantum mechanics can be easily demonstrated, quantum mechanical properties兲.
such as state reduction and complementarity. State reduction
is the phenomenological way of describing the influence of a A. Wave functions and operators
measurement on the subsequent state of the measured object.
The wave function has already been introduced qualita-
In our example, state reduction takes place after the detection
tively to describe the state of electrons. We now discuss
of the electron behind one of the slits. As a result of state
reduction, instead of the usual double-slit pattern, a pattern mathematically the wave function of free electrons ⌿ E kin . It
such as the one shown in Fig. 6共b兲 builds up behind the describes an ensemble of electrons prepared to have a fixed
screen. value of the property ‘‘kinetic energy’’ 共or, likewise, momen-
Complementarity has been a central element in Bohr’s tum兲.
philosophy of quantum mechanics. A simple form of it is the Next we introduce the concept of an operator. We look for
complementarity between interference pattern and path infor- a mathematical operation that extracts the value of the ki-
mation. It can be demonstrated by a series of experiments. In netic energy from the wave function of free electrons. By
each one the intensity of the detection light is decreased so trial and error, the students find that the operation
that more and more electrons escape undetected. Accord-
ingly, the visibility of the interference pattern increases in ប2 d2
⫺
each experiment. 2m dx 2
The discussion of superposition states provides a basis for
the discussion of Schrödinger’s cat paradox where a super- applied to ⌿ E kin , leads to the desired result. It leaves the
position of macroscopically different states is considered. wave function unchanged and pulls a factor E kin out of it.
The paradox is resolved by the mechanism of decoherence. They have found the operator of kinetic energy.

205 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 205
B. Eigenvalue equation Table I. Some of the statements the students had to rate in the questionnaire.

Now we ask the opposite question. Given a wave function An atom has a similar structure as the solar system 共planets that orbit the
⌿, how can we decide whether it describes an ensemble with sun兲
the property kinetic energy? The answer is to apply again the Electrons in an atom are like a smeared charge cloud that surrounds the
operator of kinetic energy. If the wave function is repro- nucleus
Electrons move around the nucleus in definite orbits with a high velocity
duced,
Nobody accurately knows the position of an electron in orbit around the
ប2 d2 nucleus because it is very small and moves very fast
⫺ ␺ ⫽E kin␺ , An electron that goes from the source to the screen in the double-slit
2m dx 2 experiment takes a definite path, even if it cannot be determined
If we knew the initial conditions precisely enough we could predict where
the corresponding ensemble possesses the kinetic energy the next electron is found on the screen
property, and the constant E kin gives the value of the kinetic The wave function determines the distribution of electrons on the screen
energy. The wave function is an eigenfunction of kinetic en- In principle, quantum objects can simultaneously possess position and
ergy and E kin is the eigenvalue. If the eigenvalue equation is momentum
not fulfilled, the ensemble described by ⌿ does not possess The uncertainty relation sets a limit on how good the momentum of an
the kinetic energy property. If a series of measurements on electron can be determined
If a precise position measurement is carried out on an electron, it is only
the corresponding electrons is made, the measured values
possible to make an imprecise momentum measurement afterwards
will have a distribution.

C. Schrödinger equation
The design of the study was as follows: The students were
With the concept of an eigenvalue equation at hand, it is
instructed by their regular teachers. Instead of a textbook
only a small step to the stationary Schrödinger equation,
they were provided with a text 共approximately 100 pages兲
which is the eigenvalue equation of total energy. Solving the
containing the contents of the course. The simulation pro-
Schrödinger equation means to search for states with the ‘‘to-
grams developed for the course were used in the classroom
tal energy’’ property. These are called stationary states.
and, in part, also at home.
The Schrödinger equation is the basic equation of quan-
We used several instruments to measure the success of the
tum mechanics, and we can solve it for some cases. The
course.
simplest example is the infinite potential well where the
共1兲 Questionnaire on students’ conceptions. The question-
quantization of energy appears for the first time. The three-
naire consisted of two parts. In the first part the students had
dimensional potential well is used to illustrate probability
to rate statements on a five-point scale from 1 共‘‘strongly
distributions in three-dimensional space and to introduce the
agree’’兲 to 5 共‘‘strongly disagree’’兲. They had to judge 44
idea of orbitals.
items from four different subfields of the course: Conception
of the atom, determinism/indeterminism, quantum mechani-
D. Atoms cal properties, and the uncertainty relation. Some typical
The final part of the course is devoted to atoms. The stu- questions are given in Table I. The questions were in part
dents observe line spectra of atoms and see the quantization adapted from Ref. 14. The second part of the questionnaire
of energy in the Franck–Hertz experiment. As mentioned, contained questions with the possibility of answers. Here,
the Bohr model is rooted deeply in the students’ minds. We students were asked to explain the uncertainty relation or to
therefore discuss the Bohr model critically and emphasize draw their visual image of an atom 共including a commentary
that it conflicts with some of the basic results of quantum whether there are features that cannot be drawn兲.
mechanics covered so far 共for example, Heisenberg relation 共2兲 Student interviews. In two of the classes 共N⫽22兲, semi-
and the impossibility of trajectories兲. As an alternative, we structured interviews were conducted. Each lasted approxi-
discuss the orbital model of the atom. mately 1 h. They were taped, transcribed, and analyzed. Top-
The mathematics necessary to solve the Schrödinger equa- ics again included the conception of the atom and the
tion for the hydrogen atom is beyond the scope of our stu- uncertainty relation. In addition, we investigated in more de-
dents. We therefore use an approach proposed in Ref. 29 and tail the understanding of more complex issues such as deter-
model the Coulomb potential by an infinite potential well minism, probabilistic laws, state superposition, and the in-
with appropriately chosen parameters. We obtain a self- sights gained in the double slit experiment.
consistent equation for the possible energy values and find 共3兲 Questionnaire on physics interest. We also used the
the ⫺1/n 2 behavior of the energy eigenvalues. This impor- questionnaire to compare the students’ interest in the physics
tant result completes our course. of waves and in quantum mechanics.
To express the success of the course by a single number,
VII. EVALUATION OF THE COURSE we calculated a statistical index C from 29 items of the ques-
tionnaire on students conceptions. An index value of
An important part in the development of a new course is C⫽⫹100 corresponds to fully quantum mechanical concep-
its evaluation. In a pilot study, preliminary versions of the tions, C⫽⫺100 means conceptions that contradict strongly
courses were taught in five classes by P. Engelhardt and the to the quantum mechanical ones, and C⫽0 corresponds to an
authors. In the actual evaluation, we tested our course in five indifferent attitude 共for example, rating each statement with a
other Gymnasium school classes with about 60 students. Two 3兲.
of the classes were nonspecialized physics classes 共Grund- Figure 7共a兲 shows the distribution of the index value C for
kurse兲 with 3 hours a week, and three classes were for stu- the students that were taught our course. All of the students
dents with a special interest in physics 共Leistungskurse兲 with have a positive value of C which means that quantum me-
5 hours per week. chanical conceptions dominate. The average is ⫹55.8 with a

206 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 206
 Fig. 8. Response to the statement: ‘‘An atom has a structure similar to the
solar system 共planets that orbit the sun兲;’’ 共a兲 experimental group, 共b兲 control
group 共1⫽‘‘strongly agree,’’ 5⫽‘‘strongly disagree’’兲.

culated similar indices for the four subfields mentioned

above. In all of them, the experimental group was superior.
Figures 8 –10 show the responses of both groups to some
individual statements in the questionnaire. In Fig. 8, the fol-
lowing statement had to be rated: ‘‘An atom has a similar
structure to the solar system 共planets that orbit the sun兲.’’ The
distribution of the experimental group 关Fig. 8共a兲兴 is peaked
toward complete rejection 共average 4.38兲 in accordance with
Fig. 7. 共a兲 Distribution of the conception index C in the experimental group; the quantum mechanical model. The control group is much
共b兲 distribution of C in the control group. more indifferent and not so critical of the planetary model
共average 3.52兲.
In Fig. 9 the question of determinism/indeterminism is ad-
dressed. The following statement was given in the context of
the double slit experiment: ‘‘If we knew the initial conditions
standard deviation of 19.5. We consider this result as an in- precisely enough, we could predict where the next electron is
dication that the students successfully learned quantum me- found on the screen.’’ The experimental group 关Fig. 9共a兲兴
chanical conceptions. clearly denies the determinism of classical mechanics in the
To compare the results with a group of traditionally in- quantum domain 共average 4.75兲. The control group is much
structed students, we gave the questionnaire on student con- less certain 共average 3.88兲.
ceptions to a group of 35 first-year university students who Figure 10 shows the response to a statement on the uncer-
had been taught quantum physics during their time in Gym- tainty relation. The statement was, ‘‘In principle, quantum
nasium. We took care to include in the definition of C only objects can possess simultaneously position and momen-
those questions that we considered to be fair to the control tum.’’ Again, the experimental group definitely rejected the
group. The distribution of the index C in this control group is statement 共average 4.89兲, whereas there are mixed opinions
shown in Fig. 7共b兲. The average is ⫹35.2 with a standard in the control group 共average 2.92兲. In all of the examples
deviation of 23.7. A comparison of the two diagrams shows mentioned, the difference between the groups is highly sig-
that the experimental group has developed more pronounced nificant.
quantum mechanical conceptions than the control group. The The response to a question with an open reply possibility
difference is highly significant 共significance level 0.1%兲. To is shown in Fig. 11. The students were asked to draw an
appreciate this result one has to take into account that the atom according to their conceptions. We categorized the re-
control group is positively selected because these students sults according to the scheme of Ref. 30: 共i兲 Bohr 共nucleus
decided to major in physics at the university. We have cal- with electrons on orbits兲, 13% in the experimental group

207 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 207
 Fig. 11. Student drawings of an atom 共classification as in Ref. 28兲; 共a兲
Fig. 9. Response to the statement: ‘‘If we knew the initial conditions pre-
experimental group, 共b兲 control group.
cisely enough, we could predict where the next electron is found on the
screen.’’ 共a兲 Experimental group, 共b兲 control group.

versus 32% in the control group; 共ii兲 cloud 共distributed cloud

around the nucleus兲: 61% vs 29%; 共iii兲 dumb-bell 共reminis-
cent of a p- or d-orbital兲: 16% vs 27%, and 共iv兲 ‘‘no image
possible’’: 5% vs 11%. In accordance with the results dis-
cussed earlier, the Bohr model played a minor role in the
experimental group. Nearly all of the students drew a
‘‘cloud’’ image. In the control group the classical Bohr
model, the quantum mechanical cloud, and the dumb-bell
image were approximately evenly distributed.
Another interesting question is whether the students have
developed the competence to argue freely and unaided within
the new conceptual framework. This ability was tested in the
interviews. Students were asked to comment in their own
words on several questions and statements related to the con-
tents of the course. The results are instructive in particular
for the physically more complex topics. For example, an
important aim of the course was to realize that electrons do
not necessarily possess the position property. To what extent
were the students able to justify their statements instead of
merely memorizing the correct answer?
One of the interview questions was, ‘‘Someone claims that
an electron in the double-slit experiment goes either through
the left or through the right slit. How can you disprove this?’’
To evaluate the student answers quantitatively, we graded the
replies from 1 共physically correct, clear and careful reason-
ing兲 to 5 共insufficient or totally confused answer兲. The results
are shown in Fig. 12. Most of the students 共55%兲 were able
to argue adequately 共mark 1 or 2兲 whereas 32% apparently
did not develop a deeper understanding 共mark 4 or 5兲. A
typical example of an 共oral兲 student reply marked with 1 is
the following: ‘‘Well, it’s actually this point in physics that is
Fig. 10. Response to the statement: ‘‘In principle, quantum objects can most fascinating to me, in quantum physics. Because if it
simultaneously possess position and momentum.’’ 共a兲 Experimental group, would go through the left or the right slit, then the electron
共b兲 control group. would have to end up in the same region if both slits were

208 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 208
 1
R. P. Feynman, R. B. Leighton, and M. Sands, The Feynman Lectures on
Physics, Vol. III 共Addison–Wesley, Reading, 1966兲.
2
A. Rae, Quantum Physics: Illusion or Reality 共Cambridge U.P., Cam-
bridge, 1986兲.
3
M. P. Silverman, More Than One Mystery. Explorations in Quantum In-
terference 共Springer, New York, 1995兲.
4
D. Z. Albert, Quantum Mechanics and Experience 共Harvard U.P., Cam-
bridge, 1992兲.
5
Available online at http://www.phys.ksu.edu/perg/vqm.
6
H. Niedderer, T. Bethge, and H. Cassens, ‘‘A simplified quantum model: A
teaching approach and evaluation of understanding,’’ in Relating Macro-
scopic Phenomena to Microscopic Particles: A Central Problem in Sec-
ondary Science Education, edited by P. L. Lijnse et al. 共CD-␤ Press,
Utrecht, 1990兲.
7
Fig. 12. Why do electrons not necessarily possess a position property? Dis- J. Petri and H. Niedderer, ‘‘A learning pathway in high-school level atomic
tribution of marks on the free replies in the interviews. physics,’’ Int. J. Sci. Educ. 20, 1075–1088 共1998兲.
8
H. Fischler and M. Lichtfeldt, ‘‘Learning quantum mechanics,’’ in Re-
search in Physics Learning: Theoretical Issues and Empirical Studies,
edited by R. Duit, F. Goldberg, and H. Niedderer 共IPN, Kiel, 1991兲.
open or if only one slit was open. But this is not true. That is, 9
H. Fischler and M. Lichtfeldt, ‘‘Modern physics and students’ concep-
if I open only the left slit, then I get a single-slit distribution tions,’’ Int. J. Sci. Educ. 14, 181–190 共1992兲.
behind the left slit. If I open them both, I get the interference 10
A. Mashhadi, ‘‘Students’ conceptions of quantum physics,’’ in Research in
pattern; and on the location of the maximum of the single slit Science Education in Europe: Current Issues and Themes, edited by G.
pattern there is a minimum of the interference pattern. This Welford, J. Osborne, and P. Scott 共Falmere, London, 1996兲, pp. 254 –265.
11
means: it cannot go through just one of the slits, there must D. F. Styer, ‘‘Common misconceptions regarding quantum mechanics,’’
Am. J. Phys. 64, 31–34 共1996兲.
be something else 共S16兲.’’ 12
I. D. Johnston, K. Crawford, and P. R. Fletcher, ‘‘Student difficulties in
To summarize the results of the evaluation, we can say learning quantum mechanics,’’ Int. J. Sci. Educ. 20, 427– 446 共1998兲.
that the correct quantum mechanical conceptions were suc- 13
L. Bao, E. F. Redish, and R. N. Steinberg, ‘‘Student misunderstandings of
cessfully imparted to most of the students of the experimen- the quantum wavefunction’’ 共unpublished兲.
14
tal group. This understanding was achieved in spite of the G. Ireson, ‘‘The quantum understanding of pre-university physics stu-
large conceptual difficulties inherent in the subject. In addi- dents,’’ Phys. Educ. 35, 15–21 共2000兲.
15
tion, many of the common misconceptions encountered in H. Wiesner, Beiträge zur Didaktik der Unterrichts über Quantenphysik in
der Oberstufe 共Westarp, Essen, 1993兲.
traditional instruction, e.g., in the uncertainty relation, the 16
H. Wiesner, ‘‘Verständnisse von Leistungskursschülern über Quanten-
determinism/indeterminism problem, or the atom conception, physik. Ergebnisse mündlicher Befragungen,’’ Phys. in der Schule 34,
have been avoided. 95–99 共1996兲; 34, 136 –140 共1996兲.
17
R. Müller and H. Wiesner, ‘‘Vorstellungen von Lehramtsstudenten zu be-
VIII. TEACHER’S TRAINING ON THE WEB grifflichen Problemen der Quantenmechanik,’’ Didaktik der Physik, Bei-
träge zur Vol. 62 共Physikertagung, Regensburg, 1998兲, pp. S. 458 – 463.
18
Finally, we briefly mention our Web-project milq 共Munich R. Müller and H. Wiesner, ‘‘Students’ conceptions on quantum physics,’’
presented at the annual meeting of the National Association for Research
Internet Project for teacher’s training on quantum in Science Teaching, 1999; available online at http://www.phys.ksu.edu/
mechanics兲.31 In this project we want to use the Internet as a perg/papers/narst/QM_papers.pdf.
new medium for teacher’s training. Traditional courses have 19
H. Niedderer, ‘‘Physiklernen als kognitive Entwicklung,’’ Didaktik der
a capacity of 20-40 persons. With the Internet, there is no Physik, Beiträge zur 共Physikertagung Ludwigsburg, 1999兲, pp. 49– 66.
upper limit to the number of participants so that the potential 20
L. Eisenbud, The Conceptual Foundations of Quantum Mechanics 共van
impact is much larger. In addition, we can investigate the Nostrand Reinhold, New York, 1971兲.
21
L. Ballentine, Quantum Mechanics, A Modern Development 共World Sci-
changes and limitations offered by new possibilities such as
entific, Singapore, 1998兲.
the use of multimedia, simulations, and hypertext. Up to 22
D. I. Blokhintsev, Quantum Mechanics 共Reidel, Dordrecht, 1964兲.
now, milq has been available in German only. However, in 23
L. Ballentine, ‘‘The statistical interpretation of quantum mechanics,’’ Rev.
collaboration with the Visual Quantum Mechanics Project5 at Mod. Phys. 42, 358 –380 共1970兲.
24
Kansas State University, an English version is planned. W. E. Lamb, ‘‘An operational interpretation of nonrelativistic quantum
There are two main content areas in milq. We present the mechanics,’’ Phys. Today 22, 23–28 共1969兲.
ideas of the quantum mechanics course described in this ar-
25
G. Ludwig, Foundations of Quantum Mechanics 共Springer, Berlin, 1983兲.
26
It can be downloaded from www.physik.uni-muenchen.de/didaktik/
ticle, together with background information and teaching Computer/interfer/interfere.html 共the German and English languages are
material. Interested teachers can acquaint themselves with supported兲.
the new approach of teaching quantum mechanics. We also 27
P. Grangier, G. Roger, and A. Aspect, ‘‘Experimental evidence for a pho-
provide additional information relevant 共not only兲 to quan- ton anticorrelation effect on a beamsplitter,’’ Europhys. Lett. 1, 173–179
tum mechanics classes in school. This information includes 共1986兲.
28
current results of research in quantum mechanics 共e.g., quan- It can be downloaded from www.physik.uni-muenchen.de/didaktik/
Computer/Doppelspalt/dslit.html.
tum information兲 and also material on the ever-recurring 29
W. Theis, ‘‘Begründung diskreter Energiewerte für gebundene Zustände
questions of quantum mechanics such as the discussion of ohne Lösung der Eigenwertgleichung,’’ Phys. Didaktik 22, 198 –204
the EPR paradox and Bell’s inequalitites. 共1994兲.
30
M. Lichtfeldt, Schülervorstellungen in der Quantenphysik und ihre Mögli-
a兲
Electronic mail: [email protected] chen Veränderungen durch Unterricht 共Westarp, Essen, 1992兲.
b兲 31
Electronic mail: [email protected] The milq pages can be found at www.cip.physik.uni-muenchen.de/⬃milq.

209 Am. J. Phys., Vol. 70, No. 3, March 2002 R. Müller and H. Wiesner 209