sticking coefficient at ultralow energy: quantum reflection
TRANSCRIPT
![Page 1: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/1.jpg)
Pergamon Prt,grr\< I” Surface Scwwe. Vol 57. NC, I, ,‘,I (,I-‘J3. IYYX
0 1YYX Elsevier Science Ltd All n&h& reserved. Printed ID Great Br~lil~n
0079-6X I WY8 R I Y 00
PII:SOO79-6816(98)00013-6
STICKING COEFFICIENT AT ULTRALOW ENERGY: QUANTUM REFLECTION
CARLO CARRARO* AND MILTON W. COLE**
‘Dt-partmznt of Chcmicnl Enginrering, University of California, Berkeley, California 94720 “*Department of Physics, Pennsylvania State University, University Park, Pennsylvania 16802
Abstract
Some sixty years ago, Lennard- Jones and Devonshire predicted that the low energy behavior of the sticking coefficient is s oc fi, which van- ishes as the energy approaches zero. This behavior is called guantvm reflection. The prediction differs from the classical limit s = 1, because the wave function of the impinging particle is reflected at large distance from the surface. The present article explains why and when the pre- diction is valid, when it is not valid, and the nature of the experimental confirmation to date. Other related predictions are discussed.
1. Introduction
2. Theory
Contents
A. Model atom-surface Hamiltonian
l3. Rigorous results in Born approximation. I. Short range potentials
C. Rigorous results in Born approximation. II. Long range potentials
D. Rigorous results in Born approximation. III. Resonances near zero energy
E. Nonperturbative many-body effects: exact threshold behavior of sticking
coefficient
3. Experiment
A. Review of early experiments
62
64
64
69
72
73
74
79
79
61
![Page 2: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/2.jpg)
62 C. Carraro and M. W. Cole
B. Hydrogen atom scattering from helium films
C. Hydrogen atom scattering from liquid helium
4. Summary and conclusions
Acknowledgements
Appendix
A. Elastic potential
B. Coupling to ripplons
C. Effect of substrate
References
Acronyms
DWBA Distorted wave Born approximation
WKB Wentzel Kramers Brillouin
80
83
86
87
87
87
89
90
90
1. Introduction
Understanding the nature of inelastic collisions between slow atoms and cold surfaces is
of vital importance in controlling the kinetics of thermal equilibrium, which is of paramount
interest in atomic and low-temperature physics. It is well-known that in the very low energy
regime, many physical phenomena exhibit behavior which is qualitatively different from
that seen at higher energy. Energy exchange at solid or liquid surfaces is no exception
to this rule. Recent progress in atom trapping and cooling has made the extreme low
temperature (microkelvin and even nanokelvin) regime experimentally accessible, and has
allowed the determination, for the first time, of the ultra-low energy behavior of the sticking
coefficient. Such behavior has been long anticipated; for reasons which will become clear,
it is called quantum reflection. Here, we review the theory of sticking at ultra-low energy,
and of the experimental evidence of quantum reflection. We shall see that the problem
involves both general physical principles, such as impedance mismatch, as well as quite
elusive and esoteric behavior, such as retardation of van der Waals forces.
![Page 3: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/3.jpg)
Quantum Reflection 63
The basic physics of ultra-low energy sticking is straightforward. An atom incident at
low energy has a wave length A that is long compared to the characteristic distance scale
of the gas-surface interaction V(F), which is an extreme example of impedance mismatch.
The general theory of waves yields the expectation that in this limit the wave function
is likely to be reflected long before the atom arrives at the attractive well. Hence, the
sticking coefficient s falls to zero; the specific prediction [l] is that s is proportional to the
amplitude of the impinging wave within close proximity to the surface, which leads to the
dependence
SCXdE. 0.1)
This is the statement of quantum reflection. Such quantum reflection behavior differs
dramatically from the the classical expectation that s ought to approach unity at low E
for the case of a very cold surface. The latter belief is intuitively obvious, because the
approach to the surface of an infinitely slow particle takes infinite time, during which the
particle exerts a nonzero force on the surface. Thus, there will ensue some excitatation of
the surface, resulting in energy loss; the final energy of the particle will then be less than
zero, meaning that the particle will be trapped at the surface, and hence, s = 1.
Despite its simplicity, the prediction of quantum reflection had to await heroic efforts of
atomic physicists for its definitive test. The reason is that gas-surface interactions possess
very long range, which make it easy for an impinging particle to adjust its de Broglie wave
length smoothly as it travels towards the the surface. In this way, penetration into the
adsorption well can be achieved with low reflection. In fact, the smoothest of all surface
potentials, the Coulomb (or image) potential of charged particles, has been proven not
to quantum reflect at all. In general, quantum reflection is expected to be important
below an energy threshold which, for neutral particles interacting through the van der
Waals potential, is extremely low; so low, in fact, as to be reachable only with great
experimental effort. However, once the quantum reflection regime has been reached, its
onset provides interesting information concerning the very long-range decay of the surface
potential, including the retardation regime of van der Waals forces.
![Page 4: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/4.jpg)
64 C. Carraro and M. W. Cole
The need to attain exceedingly low temperatures, in order to observe quantum reflec-
tion, was first recognized by Goodman [2] and then again by Brenig [3]. Nonetheless, this
recognition does appear to have eluded some researchers and, from time to time, the finite
values of the sticking coefficient observed in some experiments at low energy (but not low
enough!) have been mistakenly construed as evidence that quantum reflection is somehow
violated (41. It has been claimed that exceptions to quantum reflection can occur when
the coupling between incident particle and surface is large [5]. However, those claims have
been shown to be in error [6,7].
The review is organized as follows. Section 2 deals with the theory of low energy
surface scattering, with particular emphasis on the threshold behavior of the sticking co-
efficient. In $3, we specialize to the physical system where quantum reflection has been
observed, namely, atomic hydrogen reflecting from liquid helium surfaces and helium films.
Concluding remarks and an outlook on future work are contained in $4.
2. Theory
A. Model Atom-Surface Hamiltonian
Low-energy scattering of an atom from a condensed medium is a difficult quantum
many-body problem. Throughout this review, we shall assume without proof that
i) because the target is a condensed medium (liquid or solid), its wave function decays
exponentially at infinity (at least in one direction);
ii) the Hamiltonian of the condensed medium has been diagonalized. The ground state
and the excitation spectrum are known;
iii) elastic scattering from the condensed medium is described by a one-body static po-
tential. From (i) and (iii) it follows that the asymptotic scattering states of the atom
are free-particle states. (The case where the one-body potential is of the Coulomb type
deserves special treatment. See below.)
iv) the atom can undergo inelastic collisions with the condensed medium. These processes
(which are responsible for sticking) are caused by interactions whose range is comparable
to, or shorter than, that of the static (elastic) potential.
![Page 5: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/5.jpg)
Quantum Reflection 65
That (i) is true is self-evident. Assumption (ii) is made for notational convenience;
most general conclusions about the threshold behavior of the sticking coefficient can be
drawn without a specific knowledge of the surface excitation spectrum [see ref. [8], which
also contains a proof of (iii)]. Note that (iii) does not imply, nor does it require, that the
one-body elastic potential can be defined uniquely. Finally, assumption (iv) is justified on
physical grounds, since the interactions that give rise to elastic and inelastic scattering at
low energies originate from the same physical mechanism (e.g., the electronic polarizability
of the medium).
A model Hamiltonian of the atom-surface system can be split up into a free part, which
we take to describe an atom in an external potential and a set of harmonic oscillators
and an interaction part, which couples the oscillators to the particle linearly
(2.2) a ’ /
The free Hamiltonian HO is obviously separable. The interaction part VI allows for
energy to be transferred between the atom and the condensed medium, through creation
or annihilation of elementary excitations.
This model Hamiltonian is relevant to many physical systems. In the case of atom
scattering from liquid helium, the excitation spectrum includes bulk excitations, such as
phonons and rotons, as well as quantized surface capillary waves, the ripplons. In the case
of scattering from solid surfaces, the modes include bulk and surface phonons [9]. Charged
particles scattering from solids can also couple to plasma modes. In all these examples,
the bulk excitation spectrum is labeled by a three-dimensional momentum vector @‘, while
surface modes have wave vectors restricted to a two dimensional plane.
Other geometries, besides the planar one, can be important. Most notable is the
spherical symmetry which would be appropriate, e.g., for the case of atom scattering from
a liquid cluster, electron scattering on a closed-shell atom, or nucleon scattering from a
closed-shell nucleus. In this review, we focus on scattering from planar, infinite surfaces.
![Page 6: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/6.jpg)
66 C. Carraro and M. W. Cole
A schematic depiction of the sticking process appears in Fig. 1. It is worth mentioning
that many of the physical phenomena we discuss, especially the threshold behavior at low
energy, have a direct counterpart in s-wave scattering from finite systems, like clusters or
atomic nuclei.
Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..__..........................
Parallel wave vector
. . . .
. . . . . . . . . . . . .
.
. . . . . . ;;$
. . . . . . . s
Fig. 1. Schematic depiction of the sticking process. An incident atom, whose total energy and surface-parallel wave vector are indicated by the open circle, undergoes a transition to a bound state, indicated by the closed circle. The shaded region is the domain of scattering states, bounded from below by the free particle parabola. The bound state parabola assumes that the particle has an effective mass equal to the free mass. The angular frequency and wave vector shown must correspond to the dispersion relation of a mode of the surface. This relation is indicated by the curve consisting of x’s.
To make a concrete connection to the case of ultra-low energy hydrogen scattering from
liquid helium, which will be the main focus of !j3, we also assume that the primary inelastic
channel is the excitation of (quantized) surface modes. At this stage, this assumption is
made purely for notational convenience, and does not affect the conclusions derived about
![Page 7: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/7.jpg)
Quantum Reflection 67
the threshold behavior of the sticking coefficient, which hold in general for couplings of
any form, under assumptions (i)-(iv). The relative importance of scattering from bulk
vs. surface modes should be gauged on a case-by-case basis. For H atoms [lo] and He
atoms [ll] incident on liquid helium surfaces, it is established that surface excitations
are the dominant mode. Thus, neglecting direct coupling of the incident particle to bulk
modes, the harmonic oscillator operators a$, a< represent quantized fluctuations of the
surface profile v(J), where g, $ are now two-dimensional vectors. The surface normal
mode quantization is carried out explicitly in the Appendix for the case of ripplons. The
interaction Hamiltonian can be written by recognizing that the interaction originates from
the same fluctuations of the medium and atom as those, which give rise to the static
interaction:
(2.3)
where 7~ is discussed in Appendix B and
v,‘m = J d2R e’+%(~~),
with
(2.4)
v(r) = &$w . .?=r
(2.5)
Here, the static surface potential has been taken to depend on z only, which is appropriate
for an uncorrugated surface such as the surface of a liquid. Note that (4) to (6) are
consistent with assumption (iv).
Several important parameters can be extracted from the Hamiltonian. Their values
provide a rationalization for the trends found in real calculations. Important energy scales
are the kinetic energy of the incident atom, the binding energy of the lowest bound state
of the static potential, and the average energy of surface exitations. Important length
scales are the mean square surface displacement, the range of a surface bound state, and
(when they can be defined) the effective range and scattering length of the potential. An
important dimensionless coupling constant can be defined as the ratio of the surface root
mean square displacement to the range of the bound state wave function, ao:
A=imiy
a0 (2.6)
![Page 8: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/8.jpg)
68 C. Carraro and M. W. Cole
Qualitatively, X is a measure of the overlap of surface fluctuations and the final state of a
surface bound state. At very low incident energy, the square of this ratio is essentially equal
to the ratio of the mean energy transferred in a single collision to the average energy of a
surface mode. Hence, a small value of X (weak coupling regime) indicates that dominant
scattering events are those in which at most one surface mode is excited. Furthermore,
in the limit, where the incident atom energy is much smaller than the binding energy,
inelastic reflection will be negligible: almost all inelastic processes will lead to sticking.
(As is intuitively manifested in Fig. 1, insofar as low incident energy means few available
inelastic scattering states.)
One might also interpret the ratio X2 as the deviation of the Debye-Waller factor from
unity or, equivalently, as a measure of the elastic reflection probability:
&1=1-X2.
$?rom this estimate of l&l, and from the unitarity condition
(2.7)
(2.8)
one obtains an approximate upper limit for the sticking coefficient
s I x2. (2.9)
In the case of weak coupling (X < l), this upper limit is qualitatively different from
the naive classical prediction that s -+ 1 at low incident energy. As pointed out by
Brenig [7], this is a consequence of properly treating the substrate quantum mechanically,
thereby accounting for the fact that energy exchange between particle and surface occurs
in quantized bits. Equation (2.7) does not treat the incident atom quantum mechanically,
however. Therefore, it fails at sufficiently low incident energy. Its range of validity can
be estimated in the WKB approximation, where the wave vector of the elastic potential is
given by
k(z) = j/k2 + 2mV(z)/P. (2.10)
![Page 9: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/9.jpg)
Quantum Reflection
For the WKB approximation to be valid, one must have
69
(2.11)
which means that the potential must vary slowly over the (WKB) wave length of the in-
cident particle. This condition fails to be met below a sufficiently small incident energy,
EC, except for potentials that decay more slowly than l/z2 (see s2.C). Thus, (2.7) be-
comes qu&atively inaccurate below EC, where a full quantum mechanical treatment of
the sticking process is needed. The value of EC below which semiclassical theory fails can
be surprisingly small for physically relevant potentials [2,3]. For example, for the retarded
van der Waals potential between a Cs atom and the surface of liquid helium, it is estimated
that EC NN lo-‘OK [12]. H ence, the WKB approximation does an excellent job describ
ing the sticking process (and the absence of quantum reflection) in most experimental
situations.
B. Rigorous Results in Born Approximation. I. Short-Range Potentials
We begin by studying the sticking coefficient in Born approximation, whose pertur-
bative nature is accurate in some practical situations, where the atom-surface interaction
is weak (A2 << 1).
Consider an incident atom beam impinging on a surface, which bounds a medium
occupying the half space I < 0. We assume for the time being that the surface is at zero
temperature, and that the incident beam is monoenergetic. Let Ei = h2(ii1 + k,)2/2m =
El1 + E, be the incident beam energy. The sticking coefficient is defined as the sum of
transition probabilities from a continuum state at incident energy Ei, normalized to unit
incident flux, into each of N bound states, In >. In Born approximation, this reads as
![Page 10: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/10.jpg)
70 C. Carraro and M. W. Cole
where the distorted wave < zlk, > solves the Schrijdinger equation
2 2
-$-&+V(z)-Ez >
< zlk, >= 0,
and In > is a bound state.
Effective range theory can be used to prove that
s(E) 0: fi
(2.13)
(2.14)
at low energy, if both of the following assumptions are valid:
i) the potential is of (sufficiently) short range, so that one can find a finite value of z, say
~0, such that kzo < 1 and for all I > zc, I*V(z)) 5 k2; and
ii) the elastic potential has a finite scattering length or, equivalently, the low-energy phase
shift vanishes linearly with k
6(k) oc k, k --) 0.
The proof is as follows. The distorted wave obeys the Schrodinger equation
d+‘(z) + [u(z) + k2]$(z) = 0
u(z) = - $%(z).
Now write
u(z) = uo(.z) +w(z)
with
so that
Q,(Z) = u(z) - k2, z < zo
uo(z) = 0, z > zo
Us 5 k2 for all z.
(2.15)
(2.16)
(2.17)
(2.18)
(2.19)
(2.20)
Consider the zeroth-order wave function @c(z), which obeys the Schrodinger equation
&‘(z) + @o(z) + k2)40(4 = 0. (2.21)
![Page 11: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/11.jpg)
The solution is
Quantum Reflection 71
$0(z) = sin(kz + 60(k)), I > ~0 (2.22)
$O(%) = aC(%) + bS(%), 22 < zo where C(Z) and S(Z) are linearly independent solutions of the Schriidinger equation for the
original potential u(.z) at zero energy, and therefore, they do not depend on k. Now, the
matching conditions at E = .rc imply that
a,bcx k. (2.23)
Defining $(z) = tic(z) +$1(-z), then @I(Z) = O(k2) by virtue of (2.20). It follows that the
amplitude of the wave function is of order k for z < ~0. But, since .rc is also the range
of the static potential and of the inelastic coupling, the matrix elements < nlVilkz > too
must be proportional to k, so that s(E) K fi, which is indeed the statement of quantum
reflection in the Born approximation.
The vanishing of the matrix element as k -+ 0 was just proved under the assump-
tions of sufficiently short-range potential and finite scattering length. In the next two
sections, we show that quantum reflection can indeed be violated, if either of these two
assumptions is abandoned. Provided they hold true, the solution to the scattering problem
essentially requires that we match a wave of finite amplitude and infinitesimal wave vector
(the asymptotic scattering state) with one of finite wave vector (the scattering state in the
potential well), which therefore will necessarily have infinitesimal amplitude.
It is important to realize that the incident particle is reflected outside of the matching
distance ~0, and not at the surface. In other words, it is the attractive tail of the surface
potential, not the repulsive wall, that causes quantum reflection. Quantum reflection is
present even in the absence of a repulsive wall. Interestingly, a purely attractive surface
potential can be realized in practice; in such case, there is a continuum of bound states
into which the sticking atom may fall. One such potential is the potential felt by a he-
lium atom scattering at low energy from the surface of liquid helium. The helium atom
will be quantum reflected in the limit of zero incident energy. Note that the latter situ-
ation, which has been studied experimentally [13], exemplifies the fact that the quantum
reflectionphenomenon pertains even to the case of a continuous bound state spectrum.
![Page 12: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/12.jpg)
72 C. Carraro and M. W. Cole
C. Rigorous Results in Born Approximation. II. Long-Range Potentials
Long-range potentials, falling off as negative powers of the distance, are very inter-
esting from a theoretical viewpoint because the proof of quantum reflection given in the
previous section may not apply here, since assumption (i) of 52.B may fail. Therefore, we
consider now elastic potentials of the type
V(z) = -cJc, a > a, (2.24)
where a is a small distance cutoff at the surface. Examples of long-range potentials are
the van der Waals dispersion potential (n = 3, crossing over to n = 4 in the retarded
regime) and the Coulomb potential (n = 1). These long-range potentials lack a character-
istic length scale, and one might wonder whether their smooth approach to infinity allows
smooth matching of the asymptotic scattering wave function with the wave function inside
the potential well, without quantum reflection. More precisely, we ask whether the scat-
tering state can adjust its wave length smoothly from the large asymptotic value of k-l
to the value of l/J- near the surface. If this is possible, then the WKB
approximation will hold true,
Wz) I I --yjg- e k2(4, 2 > a. (2.25)
and the sticking coefficient will be nonzero even at very low energy.
Following Ref. [3], we adopt dimensionless units where lengths are measured in units
of the lower cutoff a, and energies in units of h2/(2ma2). Inequality (2.25) thus reads
For this to be satisfied for all a > 1, we ask that it be satisfied when the 1.h.s. attains its
maximum, i.e., for a k-dependent value
z*(k) = k-2/” [ $;;);I ? (2.27)
For n < 2, inequality (2.26) is always satisfied. Smooth matching is possible, and there
is no quantum reflection, irrespective of the strength of the potential, as is demonstrated by
ezact calculations for the one-dimensional Coulomb potential, n = 1, described in Ref. [7].
![Page 13: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/13.jpg)
Quantum Reflection 73
The case n = 2 is marginal [14]. For c2 > 1, inequality (2.26) is satisfied for all k
values and all z > 1. Thus, if the coupling constant is strong enough, there is no quantum
reflection for n = 2, as for the Coulomb potential.
For n > 2, inequality (2.26) ceases to be satisfied for I = .a* when the incident wave
vector is smaller than a characteristic value
k c
= 21/(“--2) &iq(, + 1)(“+1)/(“-2) 33”/2(n-2),&2(n-2)&h-2) .
(2.28)
Thus, as the incident energy is lowered towards zero, quantum reflection will always set
in, although this energy threshold may be extremely small. (This conclusion is valid even
though the effective range expansion fails for these long range potentials: both scattering
length and effective range have logarithmic singularities at low energy.) The sticking
coefficient is thus predicted to vanish as a for these potentials. This prediction has been
confirmed by experiment [15].
D. Rigorous Results in Born Approximation. III. Resonances near Zero Energy
In this section, we examine the possibility that the elastic potential has infinite scat-
tering length, despite its short range. Consider for simplicity a hard-wall-plus-square-well
potential: V(Z) = 00, r<O
V(Z) = -90, 0 < z < a (2.29)
w(z) = 0, z > a,
where ‘uc is a positive constant. The scattering state at energy E = A2k2/2m outside the
well has the form
$J> = sin(kz + 4(k)), (2.30)
while inside the well,
$< = A(k) sin(Ka), (2.31)
whereK = dV k + 2mvo/ti and A has been taken to be real. Matching at a = a determines
the amplitude inside the well
A(k)2 = k2
K2 cos2 (Ka) + k2 sin2(Ka). (2.32)
![Page 14: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/14.jpg)
74 C. Carraro and M. W. Cole
It normally implies A(k) 0: k for k -+ 0 (quantum reflection). However, an exception to
this rule occurs when
cos(Ka) = 0. (2.33)
If this is the case, jA(k = 0) 1 = 1 and there is a shape resonance at zero energy. Under this
circumstance, the wave function inside the well reaches the edge of the well with exactly
zero slope, which allows it to match the infinite wave length of the zero energy scattering
state without loss of amplitude, i.e., without quantum reflection. However, as we shall
see in the next section, many-body effects restore quantum reflection, and the sticking
coefficient does vanish at zero energy, although its behavior near zero energy will be highly
nonmonotonic, a characteristic mark of resonance scattering. Note that an infinitesimally
deeper potential would support an additional bound state.
The circumstance of a shape resonance near zero energy is quite fortuitous, since it
requires the depth and range of the potential to satisfy the condition
a (2.34)
with n an integer. Thus, the mechanism of resonant scattering has been usually overlooked
in discussions of quantum reflection, until Hijmans et al. [16] and Carraro and Cole [17]
invoked it to explain the experimental measurements of the sticking coefficient of atomic
hydrogen on liquid helium films [ES], to be discussed in 53.
E. Nonperturbative Many-Body Effects: Exact Threshold Behavior of the
Sticking Coefficient
The previous sections have dealt with exact results for the low energy behavior of
the distorted wave function, from which one can compute the sticking coefficient in the
distorted wave Born approximation (DWBA). In this section, we ask whether similar
rigorous statements can be made about the true sticking coefficient and not, just its first
order perturbative expression. As we shall see shortly, this question is answered in the
affirmative.
![Page 15: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/15.jpg)
Quantum Reflection 75
The DWBA to the sticking coefficient suffers from two principal shortcomings. First,
it is the first order term of a perturbative expansion in the parameter X2 defined in (2.6),
and thus it assumes that the coupling term in the Hamiltonian is small. Furthermore,
it allows only a single surface excitation to partake in the trapping process. One can
imagine situations where these assumptions break down, and hence, one has to resort
to nonperturbative methods to compute the sticking probability. Secondly, the DWBA
violates unitarity; the sticking coefficient computed in DWBA could exceed unity, which
is of course an absurd result.
Due in part to these shortcomings of the DWBA, it was conjectured in the litera-
ture [5] that the prediction of universal quantum reflection in ultra-low energy surface
scattering may fail. It was argued that strong coupling to the surface excitations might
destroy quantum coherence of the incident wave packet. Numerical “evidence” in support
of this conjecture was claimed in simulations of surface scattering of a charged particle
in the strong-coupling regime [19]. Unfortunately, those claims overlooked the fact that
the Coulomb potential, strong or otherwise, does not quantum reflect, even in the Born
approximation [2,7].
Although the many-body scattering problem cannot be solved exactly in general, we
now show that the threshold value of the sticking coefficient is known exactly:
$no s(E) = 0, -+
(except in the case of very long ranged potentials studied in 92.C). Indeed, one can rely
on the observation that in the limit of zero incident energy (and of a surface at T = 0),
there is no inelastic rejlection channel available to the incident particle. Contrary to some
previous studies, which claimed that sticking exhausts unitarity in this case [20], we will
prove the opposite conclusion, that elastic reflection, and not sticking, exhausts unitarity
under this circumstance. Furthermore, it will be clear that many-body effects actually
conspire to make the sticking coefficient vanish at zero energy even in the presence of a
low-lying resonance, in which case the Born approximation would erroneously predict a
violation of quantum reflection.
![Page 16: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/16.jpg)
76 C. Carraro and M. W. Cole
To prove the limit (2.35), we use the exact (closed form) solution of the many body
scattering problem in a restricted Hilbert space where inelastic reflection events are not
allowed. Such solution was worked out long ago by Wigner and Eisenbud [21]. A slight
generalization of their method is given in Ref. [22]. Closed-form solutions of the many-
body problem can be found by applying different restrictions to the Hilbert space. One
celebrated approach, that of Tamm [23] and Dancoff [24], actually allows for inelastic
scattering in the continuum, provided a restriction is placed on the number of surface exci-
tations (one quantum per mode). Clougherty and Kohn [14] have carried out a calculation
of the sticking coefficient ii. la Tamm-Dancoff. The results confirms that the threshold
behavior is given correctly by (2.35) also in the case when inelastic reflection events are
allowed. Because in practice these events are unimportant, we neglect them altogether in
the following analysis.
. . . +-- ..*... +
+-f!z?4+,=4 Fig. 2. Diagrammatic representation of (2.36). The broken line propagates
the particle in the bound state, while the solid line propagates it in the continuum. The wavy line represents the ripplon propagator.
In regards to the problem of low energy atom-surface scattering, the solution of the
many-body problem amounts to resumming the diagrammatic expansion of Fig. 2 for the
Green’s function of the atom. Knowledge of the atom’s Green’s function then allows us
to reconstruct the atomic wave scattering amplitude, which obeys a Lippmann-Schwinger
equation:
pc, >*= (k, > +A Jqc dq’q’D(q’, (Ez - Eo - Eql)/h) 0
J
O” dk’ Ik: >< k:lv,:lo > ww
x + 0 = E, -E;+iO < OlV,:lk, >*,
![Page 17: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/17.jpg)
Quantum Reflection 77
where D(q,w) is the propagator of surface excitations. Here, we assume that the static
potential supports only one bound state, ]$~c >. Such a physically important situation
pertains to the case of atomic hydrogen sticking on the surface of liquid helium, which will
be studied in detail in $3. However, the results can be easily generalized to the case of
many bound states.
Equation (2.36) is suitable for numerical inversion. It is often assumed that the force
is either independent of q [3], or that it is a smooth function of q in the neighborhood of
a wave vector qo, defined as the root of Es = FL+,, + Eq,, [17]. Then a simple analytical
solution is readily available for the transition amplitude into the bound state:
and hence, for the sticking coefficient:
4%) = -$$ < olv;Jk, >* I~~“. (2.38)
(2.37)
where E, - E. - Eq
h ’ (2.39)
x” = Im(l,,) (2.40)
is the imaginary part of the generalized susceptibility of the substrate, and
Ik = J O” dk’ 1 < OlVJk > I2
0 7 k2 - (k’)2 + iv ’ (2.41)
The physical meaning (2.36) is that the scattering state of the impinging atom is
not only distorted by the presence of the static interaction Vi(z), but is also allowed to
make virtual transitions into the bound state by interacting with a virtual cloud of surface
excitations.
The sticking coefhcient given by (2.38) fulfills the unitarity condition s(E) 5 1. Note
that the discussion of the threshold behavior, presented in 52.B in the context of the Born
Approximation, continues to hold true when the perturbation series is resummed to all
orders, which is a direct consequence of the absence of inelastic reflection at zero incident
![Page 18: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/18.jpg)
78 C. Carraro and M. W. Cole
energy. In practice, the energy of the incident beam can never be made exactly to vanish.
However, the relative importance of inelastic reflection and sticking is easily estimated. In
the case of a hydrogen atom beam sticking on the surface of liquid helium, the estimate
relies simply on the available ripplon phase space and on the fact that V,I(z) ‘v W(z)/&
in a low energy process. With EO denoting the binding energy, the inelastic reflection
coefficient is suppressed by a power of (E/Eo)~ with respect to the sticking coefficient, an
extremely small number under usual experimental conditions.
Consider now the implications of (2.38) for the case when the potential has a resonance
near zero energy. The transition amplitude squared is given by
where y is a k-independent constant. Thus,
Ik=-r 2 &+%&) (
(2.42)
(2.43)
and the sticking coefficient is given by
4rn-yfk s(E) = - [h2(b + myx’/A2)2 + (m2-y2/A2)(xtt)2] - 2mx”yk + h2k2 *
(2.44)
Note that the effect of surface polarization, embodied in x’, results simply in the shift of
the position of the resonance from b = 0 to b = -myx’/li2. However, we see that for
k --) 0, the sticking coefficient now vanishes linearly with k as
in contrast to the naive Born approximation (cf., 32.C).
Note that the effect of dissipation, embodied in the imaginary part of the substrate
susceptibility, x”, is to prevent the resonance width from vanishing near zero energy [25],
which is a manifestation of the distruction of quantum coherence caused by inelastic scat-
tering. Here, quantum coherence is responsible for the presence of a resonance, that is,
for the build up of quantum mechanical amplitude inside the well, which would help to
bypass quantum reflection. Thus, dissipative effects that broaden the resonance promote
quantum reflection, rather than prevent it, contrary to what is sometimes stated in the
literature [5].
![Page 19: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/19.jpg)
Quantum Reflection
3. Experiment
A. Review of Early Experiments
79
Definitive experimental tests of the theory of sticking at ultralow energies have been
carried out using hydrogen atom beams as a probe and liquid 4He (LHe) as the substrate.
Conceptually, the H/LHe system is very well suited to test the theory. This is because (a)
there is essentially no problem of surface imperfection due to the high purity of the liquid,
(b) both the surface and bulk modes are rather well known, (c) the pertinent interatomic
interactions are relatively well known, (d) the extreme quantum regime can be most easily
attained for the lightest of all atomic probes, hydrogen. However, one should also not lose
sight of the fact that the problem takes on a practical importance in the application to
spin-aligned H atom confinement experiments. Such experiments have been performed for
some time in the pursuit of Bose-Einstein condensation [26,27]. In those experiments, LHe
films are used because H does not dissolve in LHe, and the He is inert so surface adsorption
and spin flipping are limited, resulting in very slow recombination of H to Hz. Because
experiments often involve thermal hydrogen beams, it is useful to observe [28] that the low
energy behavior
s(E) = sl&‘* (3.1)
implies a corresponding low temperature law
J;; s(T) = -pT’/*. (3.2)
The sticking coefficient also enters (through detailed balance) the problem of desorption
from liquid He.
Among the dynamical properties of the LHe/gaseous H interface, the sticking coeffi-
cient plays a central role, since sticking is the dominant inelastic channel at low energies,
as we saw in $2. Hence, transport properties can be expressed in terms of the sticking
probability. For example, the Kapitza conductance between gaseous atomic hydrogen and
the liquid helium surface, defined as the ratio of the heat flux to the temperature difference
between gas (Tg) and surface (Ts), is given by [29]
GK(T',T,) = %~~o(Tgb(Tg,Z), (3.3)
![Page 20: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/20.jpg)
80 C. Carraro and M. W. Cole
where vo(Z”,) is the surface collision rate and the accommodation coefficient is given by
(3.4)
where r(T, T,) = Taln[s(T,T,)]/dlnT, and the explicit dependence on the surface tem-
perature has been noted throughout. Equation (3.4) is valid when the sticking channel
dominates [30].
Experimentally, the first determination of the sticking coefficient of atomic H on both
4He and 3He surfaces was obtained by Jochemsen et al. [31]. They found s N 0.046 at
T N 200 mK for L4He and s z 0.016 at T II 100 mK for L3He. The accommodation
coefficient was observed by Helffrich et al. [32] to decrease with temperature in the range
0.18-0.4 K. By the end of the 1980’s, the record low temperature determination of s was
firmly in the hands of the Amsterdam group [33]. Their experiment determined the flow
rate of spin-polarized H in a capillary tube, whose walls were wetted by liquid helium.
They observed an enhanced flow rate, compared to ordinary Knudsen flow, which they
related to a vanishing sticking probability. When plotted as a function of incident energy
(see Fig. 4 below), the Amsterdam data does indeed suggest a vanishing s(E) as E -+ 0;
however, the sticking coefficient seems to decrease proportionally to E rather than E’i2.
This is an indication that the threshold for quantum reflection is not yet reached even
at temperatures as low as 100 mK. Experiments in the microkelvin regime had to await
progress in evaporative cooling [34,35]. Using this technique, the MIT group has been
able to probe the sticking coefficient at these unprecedented low energies in a series of
experiments, where s(E) was first measured for H on He films [18], and subsequently on
the bulk liquid [15].
B. Hydrogen Atom Scattering from Helium Films
The experimental cell used in the first measurements of sticking in the submillikelvin
regime consisted of a cylindrical tube 4.4 cm in diameter and 65 cm tall. The walls of the
tube were covered with a saturated 4He film. The film was kept at a temperature of 50
mK during the experiments. In the upper part of the cell, H atoms, produced in an RF
![Page 21: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/21.jpg)
Quantum Reflection 81
discharge source, were confined by static magnetic fields. A quadrupole field effected radial
confinement, while two solenoids confined the atoms longitudinally. The energy barrier EC
at the bottom of the trap was adjusted either to evaporatively cool the confined atoms
(by leaking out the hotter ones) or to measure their temperature (by dumping the entire
contents of the trap). The atoms that escaped from the trap eventually stuck to the walls,
where they recombined to molecular hydrogen. The heat of recombination was measured
by a bolometer.
Starting with atoms of known energy distribution, the radial confining field could be
lowered suddenly to let the atoms interact with the helium-covered walls. It should be
noted that, because of the experimental design, the H atoms predominantly interacted
with the upper portion of the cell walls. Because the cell was over two feet tall, only a
thin helium film is expected to be present on the walls, even if bulk liquid existed at the
bottom of the cell [36]. After specified time intervals, the confining field was restored, and
the energy distribution measured again. Comparison with the original distribution yielded
the decay rate of the population in the trap. The population decayed because of sticking
events. (Spin relaxation, which would also lead to population decay, was determined to
take place only after sticking occurred.) The population decay rate at energy E, T(E)-~,
was used to extract the sticking coefficient s(E) through the relation
T(E)-1 = 7) (3.5)
with D the cell diameter, M the atomic mass, and n a geometric factor, determined by
Monte Carlo simulations of atomic trajectories in the cell. For energies above 1 mK,
a different method was used to determine s(E). It involved letting a pulse of atoms
escape from the barrier at the bottom of the trap, and detecting the heat of adsorption
bolometrically. This method involved a significant uncertainty in the factor 7, and resulted
in larger errors.
The sticking coefficient, measured over a range of energies 100pK 5 E 5 20mK [18],
is plotted in Fig. 3. The MIT results clearly show that s(E) rises as the energy is lowered
below 1 mK. This behavior is incompatible with the the theory of sticking on bulk liquid
![Page 22: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/22.jpg)
82 C. Carraro and M. W. Cole
helium [37]. It was soon realized [16,17] that quantitative agreement between theory and
experiment is recovered by taking proper account of the finite depth of the He liquid, and
thus of the presence of a substrate. Hijmans et al. [16] have carried out calculations of the
sticking on thin He films in the DWBA. Carraro and Cole [17] have addressed the effect of
the substrate through a fully nonperturbative calculation, following the method described
in 52.E.
1
10”
10-2
10-3 10-6 10-5 10’4 10-a 10-z 10-t 1
E WI
Fig. 3. Sticking coefficient of H atoms on He films. Circles and crosses:
experimental data from ref. [18]. Solid line: nonperturbative theory; dot-dashed
line: distorded wave Born approximation; dashed line: predicted sticking coeffi-
cient on bulk helium (theoretical results from ref. [17]). See text for details.
Figure 3 shows that both the effect of the substrate and nonperturbative corrections
are quite significant. The dashed line is the result of a calculation of the sticking coefficient
on bulk films, using the potential VII of Appendix A with p = 0.587 A-i. This potential
supports a single bound state. The dot-dashed line instead is calculated with potential
VIII of Appendix C, for a film of thickness d = 50 A on a substrate with C, = 5000 KA3
![Page 23: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/23.jpg)
Quantum Reflection 83
(this is a reasonable value to assume, lacking better characterization of the experimen-
tal conditions; for reference, graphite and copper substrates would give C. = 4400 and
5500 K A3 respectively [38]). There is now qualitative agreement between experiment
and calculation. Quantitative agreement is obtained when higher order corrections to the
DWBA are included, as in (2.38). This is shown by the solid curve. Further DWBA calcu-
lations, where the film thickness was taken as a variable, showed that effect of the substrate
on s(E) is felt for films as thick as 15011. This is because the helium-hydrogen interaction
is extremely weak, since He is the least polarizable material. Thus, any substrate below
a He film enhances the wall attraction. For sufficiently thin films, a second bound state
appears, implying that the sticking coefficient on He films is dominated by the presence of
a low-lying resonance.
C. Hydrogen Atom Scattering from Liquid Helium
In order to clarify the relation to theory for infinite bulk LHe (and implicitly to avoid
the resonance near zero energy that shows up for thin helium films), a new experiment
was designed by the MIT group, where the interaction between H atoms and helium
occurred near the bottom of the cell. Thus, in the presence of saturated liquid, this
experiment was able to measure the sticking coefficient on the bulk liquid, and also to
perform measurements as a function of film thickness in subsaturated conditions.
The measured sticking probability on bulk helium is shown in Fig. 4. These measure-
ments provide the first clear experimental verification of the onset of quantum reflection
in atom-surface scattering. Also shown in Fig. 4 are the predictions of Hijmans et al. [16]
and of Carraro and Cole [17]. There is excellent agreement between the latter theory and
experiment.
The discrepancy between the calculation of Hijmans et al. [16] and that of Carraro
and Cole [17] can be traced to a difference in the form of the adsorption potential at short
distances (in the case of sticking on bulk liquid, nonperturbative effects accounted for in
ref. [17] are negligible). Basically, the potential used by Carraro and Cole switches off
the short range (many body) effects at a distance of about 25A, which is 25% larger than
![Page 24: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/24.jpg)
84 C. Carraro and M. W. Cole
that used by Hijmans et al., and larger than we would have guessed prior to the sticking
coefficient measurements on thin films [18]. In our opinion, this fact points to the existence
of interesting, and not understood, many body effects affecting the interaction of H atoms
with He films at distances that are somewhat larger than the known width of the helium
surface [39].
0.1
?= x
0.01
0.1 1 10 100 1000
T [mKl
Fig. 4. Sticking coefficient of H atoms on liquid He (Reprinted from ref. [15]
with permission). Circles represent the ultra-low temperature MIT data [15],
while triangles are data from the earlier capillary flow experiment [33]. The solid
lines are the prediction of ref. [17] assuming binding energies of 1.1 and 1.0 K.
The broken line is the prediction of ref. [16].
Finally, Fig. 5 shows how s(E) varies as the 4He pressure is increased towards sat-
uration, i.e., for increasing film thickness. Although precise values of film thickness were
not determined in the experiment, the observed trend in the data is consistent with the
predictions of ref. [17].
We should stress that both the theory of Hijmans et al. [16] and that of Carraro and
Cole [I71 make a very important prediction regarding the effect of relativistic retardation
of the van der Waals potential. (This prediction concerns the form of the potential at
very large distances, of order hundreds of A; hence, it is unaffected by small discrepancies
in the potential at short length scales.) The prediction is the the sticking coefficient in
![Page 25: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/25.jpg)
Quantum Reflection 85
the microkelvin regime is extremely sensitive to the form of the potential at those very
large distances at which retardation is expected to be important. Omitting the effect of
retardation of the van der Waals potential leads to a prediction of the sticking coefficient
which is at least three times larger than experimentally observed, and cannot be reconciled
with the data by invoking uncertainties in binding energy, ripplon dispersion, substrate
polarizability, and the like. Indeed, the earlier measurements of s(E) on thin films [18]
are in reasonable agreement with a calculation of Goldman for bulk liquid, which omitted
retardation. Thus, the experiments done at different pressure, showing the continuous
crossover between thin film and bulk regimes, should be regarded as a striking confirmation
of the retardation of van der Waals forces.
1
g 0.1
0.01 0 5 10 15 20 25
Film Thickness [nm]
Fig. 5. Sticking coefficient of H atoms at 300 mK on He films of various thicknesses (Reprinted from ref. [15] with permission).
![Page 26: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/26.jpg)
C. Carraro and M. W. Cole
4. Summary and Conclusions
The phenomenon of quantum reflection has been demonstrated experimentally in the
case of ultra slow H atoms incident on liquid 4He. It would seem warranted to further
explore the idea by studying other cases. A few examples of such systems will indicate
that other problems are worth exploring. The case of H sticking to cold 3He surfaces
seems intriguing because of the role of ripplons on a fermion superfluid; to our knowledge,
these modes have not been explored to date. The case of 3He sticking to Cs is also worth
investigating because Cs is a solid, the number of bound states is small, and the recently
calculated potential [40] has relevance to the problem of wetting transitions. Equally inter-
esting, at least, would be the study of the low energy behavior of the sticking coefficient for
a charged particle, in which case the quantum reflection concept does not apply. We hope
that experimentalists are not deterred from such a study by the difficulties of achieving
low energy beams of charged particles.
The significance of the observation of quantum reflection and of the remarkable agree-
ment between experiment and theory is worth commenting on. At first sight, one might
say‘that a good agreement is not surprising in the case of H sticking to bulk liquid 4He,
since the interaction is weak and standard perturbation theory is valid and has never been
questioned. However, it should be noted that theory preceded the experiments: the the-
oretical curve of Fig. 4 is not a fit to the data! Rather, it is a prediction based on the
results of the thin film experiments, whose analysis requires a theory of the strong coupling
regime. We used a physically sensible one-body potential, whose parameters are either
known (i.e., the van der Waals dispersion coefficients, the H immersion energy in the liq-
uid, and, within a tight range, the binding energy) or were fixed by comparison to the thin
film data. This allowed us to predict the behavior of s(E) on thick films and bulk. The
fact that the prediction was confirmed by experiments attests to the validity of the theory
of sticking in both weak and strong coupling regimes.
One may well ask what other significant physical questions have been answered or
remain to be answered by such tour de force experiments. In our opinion, a key issue is
the retardation of surface forces. That subject, part of the Casimir problem, is fundamental
![Page 27: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/27.jpg)
Quantum Reflection 87
to our understanding of both the quantum vacuum and the expansion of the universe [41].
Yet there exist few, if any, definitive tests of the concept. Low energy sticking was shown
to be incompatible with unretarded surface potentials; further experiments would be useful
for a definite quantitative validation of Casimir forces.
Acknowledgements
This research has been supported by the National Science Foundation. We would like
to thank W. Brenig, F. Goodman and T. Greytak for many stimulating conversations on
this subject.
A. The Elastic Potential
Appendix
The interaction between atomic He and H is known rather accurately. At long range,
the functional form of the potential follows from first principles. The leading term is the
dipole-dipole van der Waals interaction
?&&v(r) = -c&F (~4.1)
At distances so large that retardation effects become important (typically a few hundred
A), the potential falls off as -C7/r7. The dispersion coefficient depends on the polariz-
abilities of He and H; its value is Cs = 19500 K A6 [42]. At distances of several A and
below, the potential is obtained by fitting atomic beam phase shifts to phenomenological
functional forms. A first approximation then treats the He liquid as a semiinfinite contin-
uum extending from z = --oo to t = 0, where it is bounded by a sharp interface. This
results in a total potential V(z) on the H atom :
where V(r) is the He-H pair potential, p = 0.0218 AP3 is the number density of bulk He,
and R = (2, y). The van der Waals tail of the He-H potential determines the long range
behavior of V(z) [43]:
V(z) -+ -c3/z3; C3 = $KT6 = 223 KA3. (A.3)
![Page 28: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/28.jpg)
88 C. Carraro and M. W. Cole
Retardation has been shown in many cases to reduce this potential by a function well
approximated by [44]
(A-4)
with a characteristic length X which is typically 200 A. This is the correct form of the
potential when .z is large enough that the atomic structure of the liquid is not resolved. At
shorter distances, the interaction becomes a complicated many-body problem. However,
the observation that H scattering from the surface is predominantly elastic justifies a
description in terms of an effective one-body potential. This is also supported by the
many-body calculations of Mantz and Edwards [45], which provide the best variational
estimate to date of the binding energy.
To write the elastic potential in a simple analytic form, we start from the parameter-
ization of Goldman [29]:
VI(z) = po/(e BE + 1) - C&(Z6 + 2,“). (-4.5)
we fix ~0 = 37 K, Cc = 911 K A3, zc = 3.8 A as in ref. (291, but keep p as an adjustable
parameter, which eventually determines the binding energy. The latter has been measured
to be in the range 0.9 K 5 Eg 5 1.15 K [46,47], about 50% higher than the variational
estimate [45]. These values of EB most likely imply a single bound state for VI. As it
stands, VI does include van der Waals dispersion, but the value of the coefficient CG used
by Goldman is some four times higher than the known dispersion Ca. To correct for this,
and include the effect of retardation, we write
ti&) = %(')f(') -c3(z6+e;) (1 - f(4)-&) (-4.6)
f(z) = l/(1 + exp(z - V/a), i.e., we turn off the Goldman dispersion with a fermi-type function at a conservatively
large distance b = 25 A; a = 4 A.
Note that any effective one-body interaction can be thought of as generated by an
effective two-body He-H potential [29]:
V(T) = 1 82
--V(%) 27rrp 8.z2 z=T'
C-4.7)
![Page 29: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/29.jpg)
B. Coupling to Ripplons
Quantum Reflection 89
The He surface was treated as a flat boundary in (A.2). A realistic description must
take into account oscillations of the height ~(2) of the surface about the equilibrium
position. Quantization of the normal modes of such capillary-gravity waves, or ripplons
[48], allows the surface displacement to be written as
The free ripplon hamiltonian is
while the ripplon propagator is given by
D(q,w) = ftq 1
Mp w2 - (wq - iv)2 ebc - wq),
(4
(A.lO)
where n --* O+. Here Q = 0.27 KAe2 is the He surface tension, M is the mass of a He atom,
and we neglect the effect of gravity. We assume an upper cutoff for the ripplon spectrum
at qc = 1 A-‘, corresponding to LC x 12 K [49].
The presence of ripplons affects the He substrate-H atom interaction, which is now
obtained from (A.2) by replacing the upper limit of the z’ integration with ~(2). Ex-
panding in powers of IL, the zeroth order term yields the elastic interaction V(z); the linear
term in the ripplon operators gives the H atom-ripplon coupling
(A.ll)
and V(T) is related to V(z) through (A.7).
This coupling describes processes in which a H atom emits or absorbs a ripplon;
momentum parallel to the unperturbed interface is conserved.
![Page 30: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/30.jpg)
90
C. Effect of Substrate
C. Carraro and M. W. Cole
If the He is actually a film of thickness d on a substrate with dispersion coefficient
C,, the results we have derived above must be modified to account for the presence of the
substrate. The latter affects the physics in two ways: it changes the elastic potential V(z)
and the ripplon frequency spectrum wq.
For d of order a few tens of A or larger, only the long range behavior of V(z) is
changed. Putting (’ = z + d, we replace (A.7) by
VIII(Z) = vi(t)f(z) - c3 (z6~zg) - f) (1 - fb>)r(4 - +o. (A.12)
Thus the substrate can considerablyenhance the long range attraction, provided C, >>
C’s, which is always the case, since He provides the weakest van der Waals force of any
condensed medium.
The dispersion relation for the capillary-gravity waves on the surface of a film is [48]
wi = (gq + zq3)tanhqd, MP
(A.13)
where the acceleration in the substrate potential is g = Cs/d4 (neglecting retardation). It
turns out that gravity effects are important only for d 5 10 A. For much thicker films the
correction is negligible. The same is true of the hyperbolic tangent term, which can be put
equal to unity in the relevant range of qd.
References
1. J. E. Lennard-Jones and A. F. Devonshire, Proc. R. Sot. London A156, 6 (1936).
2. F. 0. Goodman, J. Chem. Phys. 55, 5742 (1971).
3. J. Boheim, W. Brenig, and J. Stutski, Z. Phys. B48, 43 (1982).
4. A. P. Mills, E.D. Shaw, M. Leventhal, P.M. Platzman, R.L. Chichester, Th. Martin
and R.R. Lee, Phys. Rev. Lett. 66, 735 (1991).
5. T. Martin, R. Bruinsma and P. M. Platzman, Phys. Reps. 223, 135 (1993).
6. W. Kohn, Surf. Rev. Lett. 1, 129 (1994).
![Page 31: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/31.jpg)
Quantum Reflection 91
7. W. Brenig and R. Russ, Surf. Sci. 278, 397 (1992).
8. W. Brenig, Z. Phys. B36, 227 (1980).
9. At high energy, electronic excitations are also important, of course.
10. Yu. Kagan and G. V. Shlyapnikov, Phys. L&t. A 95, 309 (1983).
11. A. F. G. Wyatt, M. A. H. Tucker, and R. F. Cregan, Phys. Rev. Lett. 74, 5236 (1995).
12. C. Carraro and M. W. Cole, Z. Phys. B 98, 319 (1995).
13. V. U. Nayak, D. 0. Edwards, and N. Masuhara, Phys. Rev. Lett. 70, 990 (1983).
14. D. P. Clougherty and W. Kohn, Phys. Rev. B 46, 4921 (1992).
15. I. A. Yu, J.M. Doyle, J.C. Sandberg, C.L. Cesar, D. Kleppner and T.J. Greytak, Phys.
Rev. Lett. 71, 1589 (1993).
16. T. W. Hijmans, J. T. M. Walraven, and G. V. Shlyapnikov, Phys. Rev. B 45, 2561
(1992).
17. C. Carraro and M. W. Cole, Phys. Rev. B 45, 12935 (1992).
18. J. M. Doyle, J.C. Sandberg, I.A. Yu, C.L. Cesar, D. Kleppner and T.J. Greytak, Phys.
Rev. L&t. 67, 603 (1991).
19. Th. Martin, R. Bruinsma, and P. M. Platzman, Phys. Rev. B 39, 12411 (1989).
20. G. Iche and P. Nozieres, J. Phys. (Paris) 37, 1313 (1976).
21. E. P. Wigner and L. Eisenbud, Phys. Rev. 72, 29 (1947).
22. C. Mahaux and H. A. Weidenmiiller, Shell Model Approach to Nuclear Reactions,
North-Holland, Amsterdam (1969).
23. I. Tamm, J. Phys. USSR 9, 449 (1945).
24. S. M. Dancoff, Phys. Rev. 78, 382 (1950).
25. E. R. Bittner and J. C. Light, J. Chem. Phys. 102, 2614 (1995).
26. T. J. Greytak and D. Kleppner, in New Y&ends in Atomic Physics Vol. II, G. Grynberg
and R. Stora, (Eds.), Elsevier, Amsterdam (1984), p. 1125.
27. I. F. Silvera and J. T. M. Walraven, Prog. Low Temp. Phys. 10, 139 (1986).
28. D. S. Zimmerman and A. J. Berlinsky, Can. J. Phys. 61, 508 (1983).
29. V. V. Goldman, Phys. Rev. Lett. 56, 612 (1986).
![Page 32: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/32.jpg)
92 C. Carraro and M. W. Cole
30. It has been pointed out by Yu. Kagan et al. [Phys. Lett. A 143, 477 (1990)] that if
the adsorbate density on the surface of the liquid is high, the adsorption process is
not dominated by ripplon-assisted processes, but rather, by adsorption into a surface
condensate with simultaneous emission of a phonon. This interesting observation is
not relevant to the experiments analyzed in detail in 3 3B & C [15,18].
31. R. Jochemsen et al., Phys. Rev. Lett. 47, 852 (1981); see also M. Morrow and W. N.
Hardy, Can. J. Phys. 61, 956 (1983).
32. J. Helffrich, M.P. Maley, M. Krusius and J.C. Wheatley, Phys. Rev. B 34, 6550 (1986).
33. J. J. Berkhout, E. J. Wolters, R. van Roijen, and J. T. M. Walraven, Phys. Rev. Lett.
57, 2387 (1986).
34. H. F. Hess, Phys. Rev. B 34, 3476 (1986).
35. N. Masuhara, J.M. Doyle, J.C. Sandberg, D. Kleppner and T. J. Greytak, Phys. Rev.
Lett. 61, 935 (1988).
36. To good approximation, the film surface is an equal potential energy surface, so that
mgh+V(d) is a constant. Here, V(d) is the potential energy of a He atom at distance d
from the wall surface, and h is the height along the cell wall. Hence, the film thickness
decreases rapidly with height.
37. C. Carraro and M. W. Cole, Phys. Rev. Lett. 68, 412 (1992).
38. G. Vidali, G. Ihm, H.-Y. Kim, and M. W. Cole, Surf. Sci. Reps. 12, 133 (1991).
39. L. B. Lurio, T.A. Rabedeau, P.S. Pershan and I.F. Silvera, Phys. Rev. B 48, 9644
(1993).
40. A. Chizmeshya, M. W. Cole, and E. Zaremba, J. Low Temp. Phys. 110, 677 (1998).
41. S. Weinberg, Rev. Mod. Phys. 61, 1 (1989).
42. J. M. Standard and P. R. Certain, J. Chem. Phys. 83, 3002 (1985).
43. To be precise, there are many-body screening corrections to this result, which are of
order a few percent for He; see ref. [44].
44. E. Cheng and M. W. Cole, Phys. Rev. B 38, 987 (1988).
45. I. B. Mantz and D. 0. Edwards, Phys. Rev. B 20, 4518 (1979).
![Page 33: Sticking coefficient at ultralow energy: Quantum reflection](https://reader036.vdocument.in/reader036/viewer/2022082718/575021d91a28ab877ea1d675/html5/thumbnails/33.jpg)
Quantum Reflection 93
46. A. P. M. Matthey, J. T. M. Walraven, and I. F. Silvera, Physica B+C 108, 1499
(1981).
47. M. Morrow, R. Jochemsen, A. J. Berlinsky, and W. N. Hardy, Phys. Rev. Lett. 46,
195 (1980); Erratum Phys. Rev. Lett. 47, 455 (1981).
48. M. W. Cole, Phys. Rev. B 2, 4239 (1970).
49. Analysis of neutron scattering and surface tension data, plus microscopic theory, yields
corrections to the hydrodynamic spectrum; see S. Sun and D. 0. Edwards, Bull. Am.
Phys. Sot. 36, 869 (1991) for references. The present results are not significantly
affected.