Center For Materials For Information Technology

An NSF Science and Engineering Center

X-ray Reflectivity Studies of

Thin Film

Prepared by

Feng Huang, Ph. D.

Postdoctoral Research Fellow

Center for Materials for Information Technology

The University of Alabama, Tuscaloosa, AL 35487
June 2005

════════════════════◊ Internal Report ◊════════════════════

 Research Notes

X-ray Reflectivity Studies of Thin Films

I. Introduction
X-ray reflectivity (XRR) technique is a powerful method for investigating monolithic
and multilayered film structures. XRR studies are highly sensitive to electron density
gradients irrespective of the crystalline nature of the system investigated [1]. It is also one
of the few methods that, with great accuracy, allow one to not only extract information on
the free surface and the interface(s), but also determine the mass density and the thickness
of thin layers along the direction normal to the specimen surface.
A typical XRR profile is shown in Fig. 1 below. This profile contains (i) a critical
angle αc whose position gives the average electron density, and (ii) interference fringes
whose period gives the total thickness. The roughness values of the air/film and the
film/substrate interfaces can be extracted from fitting analysis of the profile. However, if
the interferences from several layers beat together, the situation is much more complicated.

500 Å Cu/SiO2
σf=5 Å; σs=5 Å
Intensities (a.u.)

Critical angle
∆α

Period -- Thickness;
Amplitude -- Density difference
and roughness;

0.0 0.5 1.0 1.5 2.0 2.5

Incident angle (deg)
Fig. 1. A calculated XRR pattern for a 500 Å Cu on SiO2 substrate.

II. Basic Principles

Both reflection and refraction are well known optical phenomena. For a condensed
matter irradiated by x-rays of a given wavelength, its refractive index depends on the
electron density of the volume irradiated. Hence, x-rays are reflected from, and refracted

Copyright © by Feng Huang 1/8

 Research Notes

at, interfaces between materials of differing electron density. X-ray reflectivity relies on
the interference of x-rays reflected at different interfaces, while x-ray diffraction results
from the interference of x-rays diffracted from periodic lattice points.
Determination of the mass density of the irradiated volume depends on the total
external reflection phenomenon, which results from the fact that the refractive index of a
condensed matter is less than unity.

2.1. Refractive index

For hard x-ray radiation, the refractive index of a condensed matter, n , is slightly
less than unity and can be written as [2,3]:
n = 1 − δ − iβ , (1)
with the dispersion term δ being
⎛ λ 2 ⎞⎟ ⎛ Z + f ′ ⎞⎟ ⎛ λ 2 ⎞⎟
⎜
δ = ⎜ ⎟ re N a ρ ⎜⎜ ⎟ = ⎜ ⎟r ρ , (2)
⎜⎝ 2π ⎠⎟ ⎜⎝ A ⎠⎟ ⎝⎜⎜ 2π ⎠⎟ e e
and the absorption term β being
⎛ λ 2 ⎞⎟ ⎛ f ′′ ⎞ ⎛ f ′′ ⎞⎟ λµ
β = ⎜⎜ ⎟ re N a ρ ⎜⎜ ⎟⎟⎟ = ⎜⎜ ⎟δ = , (3)
⎜⎝ 2π ⎠⎟ ⎜⎝ A ⎠ ⎝⎜ Z + f ′ ⎠⎟ 4π
where λ is the x-ray wavelength, re (=2.818× 10−15 m) the classical electron radius, N a the
Avogadro’s number, Z the atomic number, A the atomic mass, ρ the mass density, ρe
the electron density, µ the linear absorption coefficient, and f ′ and f ′′ are real
(dispersion) and imaginary (absorption) part of the dispersion corrections, respectively.
The magnitudes of δ and β are simply related to the electron density of the
material, and are on the order 10−5 to 10−7 [4]. The real part, n = 1 − δ , is connected to
the phase-lag of the propagating wave, the imaginary part, β , corresponds to the decrease
of the wave amplitude [2].

2.2. Critical angle

Since n <1, the x-rays are refracted away from the normal to the surface when they
enter the matter (the Snell’s law). Consequently, there exists a critical angle of incidence,
αc , below which total reflection of x-rays occurs. When the absorption can be neglected,
the critical angle αc can be related to the dispersion term δ by
αc = 2δ . (4)
For Cu K α radiation ( λ ~1.5 Å), typical values of αc lie in the range of 0.2-0.6°. Here,
the angle of incidence, α , is defined as the angle between the incident ray and the surface.

Copyright © by Feng Huang 2/8

 Research Notes

The critical angle reflection is not well defined for heavily absorbing materials. Fig. 2
shows the effect of absorption on the definition of the critical angle. Clearly, the I R / I 0
ratio at the critical angle is a function of β / δ .

Fig. 2. The effect of absorption on defining the critical angle. Taken from [5].

III. Direct Analysis of XRR Patterns for Monolithic Films

XRR patterns for monolithic (single-layer) films can be directly analyzed without
invoking tedious fitting analysis.

3.1. A quick estimation of thickness

For a clearly identified single periodicity, the film thickness t can be estimated
quickly yet rather accurately from the fringe spacing ∆α at high angles ( ≥ 4αc ) by
λ 2π
t= = . (5)
2∆α ∆q
where q =4 π sin α / λ . The advantage of Eq. (5) lies in its simplicity. In many cases, it is
enough for determining the deposition rates. However, its accuracy, though much better
than obtained from profilometry, is not so good as Eq. (7) below.

3.2. Modified Bragg equation

For monolithic films, the film thickness t and density ρ can be obtained by directly
reading the maxima and minima of the interference fringes of the reflectivity patterns
through the so-called modified Bragg equation as follows [6]:
qm2 = 2δ(4π / λ)2 + (m + ∆m )2 (2π / t )2 , (6)
where m is an integer indicating the reflection order, αm the observed angular position
corresponding to the maximum or minimum intensity of the m th order of reflectivity

Copyright © by Feng Huang 3/8

 Research Notes

oscillation, ∆ m has the value of 1/2 or 0. If the mass density of the film is higher than
that of the substrate, then ∆ m =1/2 for the maximum and ∆ m =0 for the minimum. For
the reversed situations, ∆ m =0 for the maximum and ∆ m =1/2 for the minimum. Eq. (6)
indicates that a qm2 versus (m + ∆m )2 plot should consist of points falling on a straight
line. Fig. 3 shows such a plot.
The slope of the line can be used for precise determination of the film thickness t by:
2π
t= . (7)
slope
Likely, the dispersion term δ , and accordingly the mass density ρ , can be determined
from the intercept with the qm2 axis.
8

maxima
minima
6
qm (100×)

(2tπ )
2
2

t = 511.6 Å
-5
δ = 1.334x10
0

0 100 200 300 400 500

2
(m+∆m)

Fig. 3. A typical qm2 versus (m + ∆m )2 plot.

IV. Reflectivity Calculations

For monolithic films, the thickness determined from Eqs. (5) or (7) is accurate enough
for most cases. However, if still higher accuracy is expected for thickness, density, and
roughness, fitting analysis — fit the experimental reflectivity patterns to simulated ones —
is needed. The fitting analysis, generally a very tedious process, is the most reliable way
to extract information from reflectivity patterns. Also, if the film structure is composed of
more than one layer, direct analysis of the pattern is very difficult.

4.1. Reflectivity from an ideal surface: Fresnel Reflectivity

At grazing incidence, there is no distinction between the σ and the π polarizations.
Consequently, the Fresnel reflection coefficient for the interface between the air and a
matter with a refractive index of n = 1 − δ − i β is

Copyright © by Feng Huang 4/8

 Research Notes

αi − αi2 − 2δ − 2i β
rF = , (8)
αi + αi2 − 2δ − 2i β
where αi is the angle of incidence with respect to the film surface. Accordingly, the
Fresnel reflectivity is given by
2
αi − αi2 − 2δ − 2i β
RF = . (9)
αi + αi2 − 2δ − 2i β
For incidences at αi > 3αc , Eq. (9) can be approximated by
4
α 1
RF ≈ c ∝ . (10)
2αi αi4
Therefore, at higher incidence angles, the reflectivity drops by a factor of αi-4 .

4.2. Reflectivity from ideal interfaces

A. Definitions
For a multilayered film structure consisting of n layers, the total number of
interfaces involved is n + 1 . Starting from the air/film interface (the film surface), these
interfaces are labeled successively as interface 0, 1, " , n (Fig. 4). Hence, layer j is
bound by the interfaces j − 1 and j . If αj is defined as the incidence angle with respect

Fig. 4. A schematic diagram showing the definitions used in a multilayered film.

Copyright © by Feng Huang 5/8

 Research Notes

to the interface j , then α0 = αi , the incidence angle with respect to the film surface. Also,
the refractive index with respect to the vacuum for the layer j is defined as n j given by
n j = 1 − δj − i βj . (11)
The Fresnel reflection coefficient from interface j is given by [5]:
f − fj +1 q − q j +1
Fj = j = j , (12)
fj + f j +1 q j + q j +1
where
⎧
⎪ fj = α02 − 2δj − 2i β j
⎪
⎪
⎨ . (13)
⎪
⎪q j = 2k0 f j = 4π fj / λ
⎪
⎩
Clearly, f0 = α0 and all f j ≠0 take a complex value. Similarly f j +1 can be obtained.

B. Fresnel reflectivity from a single layer

The Fresnel reflectivity for a single layer bonded on a semi-infinite substrate is [3]:
2
F + F1 exp(−iq1t1 )
RF = 0 , (14)
1 + F0F1 exp(−iq1t1 )
where F0 is the Fresnel reflection coefficient of the free surface (air/film interface), F1 the
Fresnel reflection coefficient of the film/substrate interface, q1 is given by Eq. (13), and t1
the layer thickness [7]. Note that the denominator of Eq. (14) differs from unity by a term
which corresponds to multiple reflections in the layer [3].
A direct calculation of Eq. (14) is very tedious, thereby preventing simple evaluation
of reflectivity patterns. Significant simplifications, however, can be made at relatively high
angles where α0 / 3 > max( 2δ1 , 2δ2 ) . Eq. (14) can then be simplified as
δ12 + (δ2 − δ1 )2 + 2δ1(δ2 − δ1 )cos[Re(q1 )t1 ]
RF ≈ . (15)
4α04
Here, δ1 and δ2 − δ1 indicates the contrasts in electron density across the two relevant
interfaces. From Eq. (15), the period of the reflectivity pattern is given by
2π λ λ
= ≈ . (16)
Re(q1 ) 2 α0 − 2δ1
2
2α0
Therefore, when multiplied by α04 , the reflectivity at high angles oscillates by a period of
λ / 2α0 , with the amplitude proportional to δ1(δ2 − δ1 ) .

C. Fresnel reflectivity from a multilayer

The reflectivity for a system having n thin layers can be obtained by a recursive
method attributed to Parratt, who showed that, for an arbitrary layer j that is bound by

Copyright © by Feng Huang 6/8

 Research Notes

interfaces j − 1 and j , the total reflectivity at the two interfaces, rjF−1 and rjF , are related
to Fj −1 by [5,8]
Fj −1 + rjF exp(−iq j t j )
rjF−1 = . (17)
1 + Fj −1rjF exp(−iq jt j )
In Eq. (17), Fj −1 and q j can be found from Eqs. (12) and (13), while rjF is the total
reflection coefficient at interface j , defined as the ratio of outgoing to incoming electric
wave amplitudes. Note that except at the interface n (the film/substrate interface), rjF is
different from Fj in that the former is composed of the superimposed reflections from
interface j and all the interfaces beneath.
The boundary condition for Eq. (17) is found at interface n where rnF = Fn , since the
substrate is semi-infinite [5]. Therefore, starting from j = n , each rjF−1 down to r0F can be
found by a recursive application of Eq. (17). The reflected intensity RF measured by the
2
detector can be found by RF = r0F .
For a semi-infinite substrate, Eq. (8) is recovered by noting n =0 and r0F = F0 . For a
homogeneous layer bonded to a semi-infinite substrate (n =1), it immediately follows
r1F = F1 and
F0 + F1 exp(−iq1t1 )
r0F = . (18)
1 + F0F1 exp(−iq1t1 )
Accordingly, Eq. (14) is recovered. The Parratt formalism boasts several notable
advantages. Among them, the most remarkable one is its freedom from approximations,
which makes it applicable for all the scattering regions that span from less than the critical
angle to much higher incidences [1]. In addition, it takes into account the absorption, and
can deal with an infinite number of layers (from n = 0 to n = ∞ ) [1].

4.3. Reflectivity from rough interfaces

Ideal interfaces have been assumed in deriving the Parratt formalism. The non-
ideality of the interface should be taken into account in real reflectivity calculations since
most interfaces are not ideal on the length scale probed by the x-rays. For a non-ideal
interface, the average electron density profile along the specimen normal can be modeled
by an error function (the Névot-Croce roughness [9]). Accordingly, the rjF terms in Eq.
(17) can be replaced by rjN −C given by [8,10]
rjN −C = rjF exp(−2σ j2k 02 f j fj +1 ) , (19)
where k0 = 2π / λ is the wavevector in the free space.

Copyright © by Feng Huang 7/8

 Research Notes

References

[1] H. Zabel, Appl. Phys. A 58:159 (1994).

[2] G. Bauer and W. Richter, Optical Characterization of Epitaxial Semiconductor
Layers, p.373 (Springer-Verlag, Berlin, 1996).
[3] J. Daillant and A. Gibaud, X-ray and neutron reflectivity: principles and
applications (Springer, Berlin, 1999).
[4] E. Chason and T. M. Mayer, Crit. Rev. Solid State Mater. Sci. 22:1 (1997).
[5] L. G. Parratt, Phys. Rev. 95:359 (1954).
[6] A. Segmüller, Thin Solid Films 18:287 (1973).
[7] V. Holý et al, High-resolution x-ray scattering from thin films and multilayers
(Springer, Berlin, 1999).
[8] D. Bahr et al, Phys. Rev. B 47:4385 (1993).
[9] L. Névot and P. Croce, Rev Phys Appl 15:761 (1980).
[10] S. K. Sinha et al, Phys. Rev. B 38:2297 (1988).

Copyright © by Feng Huang 8/8