COMMUNICATION

Beer’s law – why absorbance depends (almost) linearly on

concentration
Thomas G. Mayerhöfer *[a], Jürgen Popp [b]

Abstract: Beer’s law assumes a strictly linear dependence of the
absorbance from concentration. Usually, chemical interactions and
instrumental imperfection are made responsible for experimental
deviations from this linearity. In this contribution we show that even in
the absence of such interactions and instrumental errors, absorbance
should be only approximately proportional to concentration. This can
be derived from the quadratic dependence of the complex refractive
index, and, by that, of the molar attenuation coefficient, from the
dielectric constant and its frequency dispersion. Following dispersion
theory, it is the variation of the real and the imaginary part of the
dielectric function that depends linearly on concentration in the
absence of interactions between the oscillators. We show that this
linear correlation translates into a linear dependence of the
absorbance for low concentrations or molar oscillator strengths based
wherein  is the reduced mass of the charges and x their
displacement, and we have assumed that the restoring force is
proportional to this displacement (Hook’s law). Furthermore,
absorption is introduced via the damping of the vibration in form
of a term proportional to the velocity of the charged mass dx dt
and a damping constant . Finally, the vibration is forced by an
applied electric field E0 exp  it  , which is varying in time with
the frequency  and acts on the charge q . If the vibration were
unforced, its eigenfrequency would be 0  k  , wherein k is
the force constant from Hook’s law.
The solution for the displacement in dependence of time derived
from eqn. (1) is well-known:
q
x t   E t  . (2)
  0   2  i 
2

on an approximation provided by Lorentz in 1906. Accordingly, Beer’s

law can be derived from dispersion theory.
Newton seems to have been the first trying to explain the
phenomenon of dispersion.1 More successful attempts to
understand the change of optical properties with frequency
originally did not include absorption.2, 3 Nevertheless, even before
Maxwell’s equations gained general acceptance, it was already
established that absorption plays an important role to elucidate
dispersion.4, 5 Later on, Hendrik Antoon Lorentz,6 and Paul
Drude,7 among others, included electromagnetic theory into the
dispersion formulas. Since then, dispersion is seen as the
consequence of a vibration of opposing charges under the
influence of an electric field. Accordingly, the basic equation of
this type of motion is given by,
d 2x dx
 2
   02 x  qE0 exp  it  ,
dt dt
If this displacement is multiplied with the charge, we obtain the
dipole moment p which is linked to the macroscopic polarization
P via,
P  qx
N . (3)
p
Here, N is the number of dipole moments per unit volume. It might
not be obvious at this point, but it is eqn. (3) that ultimately leads
to Beer’s law, independent from what model we derive x(t). In line
with that and in order to derive the macroscopic polarization, we
explicitly assume that there are no interactions between the
oscillators, which is synonymous to the assumption that atoms or
molecules do not interact “chemically”. To establish the
connection to Beer’s law, we deviate slightly from the usual
course in deriving the dielectric function by using that the ratio of
(1) the number of dipole moments per unit volume and the Avogadro
constant results in the (molar) concentration: N N A  c .
Therefore, we can write eqn. (3) alternatively as:
P  qxN A c . (4)

It is well-known that for the

macroscopic polarization
[a] PD Dr., T.G. Mayerhöfer P   0   r 1 E holds, wherein 0
is the vacuum permittivity.
Spectroscopy/Imaging
Using this equation and combining it with eqs. (2) and (4), we can
Leibniz Institute of Photonic Technology
Albert-Einstein-Str. 9, Germany derive a formula for the relative dielectric function  r in
And Institute of Physical Chemistry and Abbe Center of Photonics dependence of the concentration c:
Friedrich Schiller University, Jena N Ac q2
D-07743, Helmholtzweg 4, Germany  r  1 . (5)
E-mail: Thomas.Mayerhoefer@ipht-jena.de  0 0   2  i
2

[b] Prof. Dr., J. Popp

Spectroscopy/Imaging To obtain the final form of the dispersion relation, we define a
Leibniz Institute of Photonic Technology
molar oscillator strength S* via S  q N A /   0  :
*2 2
Albert-Einstein-Str. 9, Germany
AND Institute of Physical Chemistry and Abbe Center of Photonics S *2
Friedrich Schiller University, Jena r  1 c . (6)
0 2   2  i
D-07743, Helmholtzweg 4, Germany
E-mail: Juergen.Popp@ipht-jena.de
Supporting information for this article is given via a link at the end of Therefore, there is a linear correlation between the changes of the
the document. dielectric function, both real and imaginary part, due to the forced
COMMUNICATION
oscillation, and the concentration of the vibrating species. At this 4 1 S *2  2
 *    . (13)
point, it is important to realize that this linear relation is not ln10 n     2  2 2  2 2
0
generally transferred to the index of refraction. The obvious
reason is, as Maxwell’s equations show for scalar media and has
been established experimentally numerous times in the early There is, however, one difference between empirical and derived
days of spectroscopy,8 that law, which is that according to the derivation the molar attenuation
coefficient is inversely proportional to the index of refraction and,
 r  n2 . (7)
thereby, itself a function of concentration, molar oscillator strength,
damping and oscillator position. For diluted gases, however, the
Based on eqn. (7), it can already be concluded that the complex
index of refraction is nearly unity, therefore dispersion will alter
index of refraction n is, in contrast to the dielectric function, not a
the absorbance only at higher concentrations and oscillator
property of matter, but a property of the waves inside. This seems
strengths.
to be an artificial difference, but this difference becomes much
In fact, if we start from eqn. (8), follow Lorentz’s path and take into
clearer in anisotropic media, where the dielectric function turns
account that for x << 1, 1  x  1  x 2 :6
into a tensor and, respectively, two indices of refraction depend
on the direction of the waves in a complex manner, described by S *2 2
n  1 c . (14)
the so-called Booker quartic.9 Furthermore, we will see later on 0   2  i
2

that eqn. (7) leads to a kind of destructive interference, which

distorts the absorbance through the imaginary part of n for higher
concentrations, while the shape of the dielectric function remains
unchanged. Therefore, it might not be surprising that doubling the
concentration will not generally lead to twice the original
absorbance. After all, dealing with light means also dealing with
waves, so some counterintuitive results, very much like in
quantum mechanics, are to be expected. One would assume,
though, that there should also be a kind of correspondence
principle. In contrast, eqn. (7) seems to reject any connection
between Beer’s law and the dispersion relation at first view, since
the complex index of refraction function, and with it its imaginary
part, the index of absorption function seems not at all to depend
on the concentration linearly:
S *2
n  1 c . (8)
0 2   2  i
For the moment we do not follow this path further. Instead, we
separate real and imaginary part in eqn. (6) and obtain for the
imaginary part of the dielectric function:
S *2

 r  c . (9)
This approximation is what lets us presume that the contributions
from each individual oscillator, be it that they all share the same
eigenfrequency  0 or not, are additive concerning the index of
absorption function and, ultimately, the absorbance (on a side
note, the linear concentration dependence of the real part is
exploited to determine concentration in non-absorbing spectral
regions by refractive index sensing). The above approximation
eliminates the dependency of the molar attenuation coefficient
from the index of refraction:
4 S *2  2
 *    . (15)

ln10  2  2 2  2 2
0 
We have recently investigated how eqn. (14) performs as
approximation of eqn. (8) and found that the differences are very
small as long as the oscillators are comparably weak, like for
organic and biological matter in the infrared spectral range.10
Accordingly, we can predict that the approximation/assumption of
additivity, on which Beer’s law and its standard derivation are
based, is excellent in these cases.
2.0

 0    2 2

2 2 2

c·d/(m·mol·l-1)
0.05·1000
In the next step, we use that for  r the following relation holds, 1.5
0.5·100
 r  2 n k , put this relation into eqn. (9) and divide both sides by
Absorbance

5·10
2n : 50
1.0
c S *2
 2
k    . (10)
n      2  2  2 2
2
0
0.5

If we switch to wavenumbers, the absorbance is defined as,

4
A    k   d . (11) 0.0
ln10 1600 1650 1700 1750 1800
-1
Wavenumber / cm
where d is the thickness and  the wavenumber. From eqs. (10)
and (11) we obtain:
4 c S *2
 2 Figure 1. Wavenumber dependence of the absorbance calculated for the
A    d. (12) hypothetical material and concentrations of 0.05, 0.5, 5 and 50 mol/l. For clarity,
ln10 n     2 2  2 2
 2
 the values have been multiplied by the factors 1000 (0.05 mol/l, black curve,
0
hidden behind the red curve), 100 (0.5 mol/l, red curve) and 10 (5 mol/l, green
curve), which is equivalent of assuming the product of concentration and
This is nothing else, but Beer’s law A     *   cd , when   
*
thickness c∙d to be constant.
is the molar attenuation coefficient given by:
COMMUNICATION
For stronger oscillators, however, we have to investigate the limits
of the approximation. Accordingly, we assumed an oscillator with
a molar oscillator strength of 70 (l cm-1)/mol, which would
resemble a medium strong oscillator of an inorganic material in
the infrared or in the UV (  *  103 l/(mol cm-1)), if this hypothetical
material would be neat at c = 50 mol/l.
We calculated four spectra for c = 0.05, 0.5, 5 and 50 mol/l and
assumed that the product of thickness d and c is constant: c∙d =
0.005 (mol∙cm)/l to clarify the spectral changes. The
corresponding spectra are shown in Fig. 1. Accordingly, the
absorbance peak is originally practically symmetric and located at
its oscillator position for the small concentrations c = 0.05 and 0.5
mol/l, suggesting that the approximation contained in eqn. (14)
and thereby Beer’s law should hold well in this range. As a
consequence, there is no noticeable difference between the
thousandfold absorbance for c = 0.05 mol/l (black curve) and the
hundredfold absorbance for 0.5 mol/l in Fig. 1 (red curve, which
practically covers the black curve completely).
For c = 5 mol/l the inverse proportionality of the molar attenuation
constant and the index of refraction begins to show and at c = 50
mol/l, this functional dependence deforms the absorbance
spectrum strongly. These deformations are nothing extraordinary
for strong oscillators. While   is a perfectly symmetric function
(at least for one oscillator), k can no longer be symmetric, since
  is divided by the index of refraction, which is, in contrast to   ,
an odd function. Accordingly, the deviations of k become grander
the larger the product of concentration c and oscillator strength S*
is (more examples covering most of the possible molar oscillator
strengths can be found in the supporting information).
2.0
CDHO
Beer's law
1.5 linear fit between 22.5 - 27.5 mol/l
linear fit between 45 - 50 mol/l
Absorbance
1.0
0.5
0.0
0 10 20 30 40 50
c / (mol/l)
Figure 2. Concentration dependence of the absorbance at the oscillator position
(1700 cm-1) for a hypothetical material based on the classical damped harmonic
oscillator (CDHO) model (black curve) compared to the Lorentz approximation
that leads to Beer’s law (red line). Furthermore, two linear fits for two regions of
the black curve are presented.
Deformations like the one for c = 50 mol/l are usually attributed to
either chemical interaction and/or missing linearity of the
instrument. Recently, also deviations from the (Bouguer-)Lambert
part of the (Bouguer-)Beer-Lambert law due to wave optics like
interference became a focus of attention in this context (“electric
field standing wave artefacts”).11-14 Respectively, blue shifts and
asymmetric broadening in absorbance spectra seem to be
nothing out of the ordinary, cf., e.g., Fig. 1 in the excellent tutorial
about the Beer-Lambert law by Mäntele and Deniz.15 However,
how much the individual influences contribute to the spectral
deformation, would have to be investigated in each particular case.
Nevertheless, with all progress in instrument development and
the fact that the tendency to interact should be strongly diverse
for chemically different solutes, it would be somewhat surprising
that we still see the same deviations from linearity for Beer’s law,
except if other factors played a role. Speaking of these deviations,
Fig. 1 of 15 also contains an inset in which the absorbance at a
fixed wavelength, the peak maximum for smaller absorbances, is
plotted against the concentration. Also, such plots are still very
common, and, likewise, seem to be uninfluenced from progress
in instrument development. Accordingly, it is still actively
searched for improved mathematical instruments to analyze the
deviations from linearity of Beer's law.16 A corresponding plot is
provided in Fig. 2 of our paper. As can be seen, similar deviations
from linearity occur due to the non-additive nature of absorbance
and may provide an alternative or additional cause for the
occurrence of such deviations in experiments.
Nevertheless, up to relatively large concentrations, the linear
dependence of absorbance from concentration according to
Beer’s law is an excellent approximation (note that, the deviations
of the same order of magnitude should occur as long as the
product  *  c is constant. E.g. for the highest molar attenuation
coefficients in UV-Vis spectroscopy of  *  105 l/(mol cm-1), a
hundredfold smaller concentrations would suffice to introduce the
same deviations). Even more interesting, also at higher
concentrations, ranges can be found where the absorbance is
approximately linear, albeit with a decreased  * compared to that
at lower concentrations. These findings are of particular
importance since recently a method has been introduced by Zhao
et al., which allows molar range detection.17 Interestingly, the
authors took for each presented example the so-called sideband
(actually, it is not really a band. Instead, an inverted peak
develops based on changes in the flank of the band and the
remaining detector saturation around the peak maximum) at the
high-wavenumber side of the peak maximum. In our view, this is
the natural selection, since due to the blue shift of the band with
increasing concentration, it can be concluded that the high
frequency/wavenumber region should show much stronger
deviations from Beer’s law than its counterpart. This is
demonstrated in Fig. 3, where we have plotted the absorbance
dependence on the concentration 80 cm-1 away from the oscillator
position. Note, that eqn. (15) is not the usual Lorentz-profile (for
which two more approximations have to be invoked, cf. 6),
therefore the band is not fully symmetrical and the lines at 1620
and 1780 cm-1, which display Beer’s law conform behavior, have
a slightly different inclination. Thanks to the band shift, at 1620
cm-1 the deviations from Beer’s law and linearity are comparably
small, whereas at 1780 cm-1 the deviations increase enormously
and are positive right from the start.
COMMUNICATION
Acknowledgements
0.06
@1620 cm-1
CDHO Financial support of the EU, the ”Thüringer Ministerium für
Beer's law Wirtschaft, Wissenschaft und Digitale Gesellschaft”,
@1780 cm-1 the ”Thüringer Aufbaubank”, the Federal Ministry of Education
0.04
Absorbance

CDHO and Research, Germany (BMBF), the German Science

Beer's law
Foundation, the “Fonds der Chemischen Industrie” and the Carl-
Zeiss Foundation is gratefully acknowledged.
0.02

Keywords: Dispersion • Absorbance • Beer’s law

0.00
0 10 20 30 40 50
c / (mol/l)
Figure 3. Concentration dependence of the absorbance at two wavenumbers
80 cm-1 away from the oscillator position (1700 cm-1) for the hypothetical
material based on the CDHO model (black/green curves) compared to the
Lorentz simplification that leads to Beer’s law (red/blue line).
To summarize, we have established the connection between
Maxwell’s equations, the dispersion relations and Beer’s law and
derived the latter from the former employing an approximation
introduced by Lorentz in 1906. While it is in theory known since
then, though not well-known, that absorbance is not additive, the
intuitive assumption of a linear dependence of absorbance from
concentration is in wide concentration ranges justified.
The advent of much brighter light sources (like quantum cascade
lasers in the mid-infrared) and more sensitive detectors, as well
as methods to determine concentrations in the molar range with
standard equipment, might, however, be a game changer and
demand an understanding of the dependence of the optical
properties way beyond the limits of Beer’s law.
As a note after acceptance, it was hard to believe for us that the
natural connection between dispersion and absorption had not
been established previously. In fact, while this manuscript was
under review, we kept looking for an older work on this topic and,
finally, made a find. Accordingly, Max Planck derived his own kind
of dispersion relation in 1902.18 One year later, he published a
paper where he discussed, among other things, the principal
influence of the number of dipole moments per unit volume
(“vertheilungsdichte”) on the shape of the bands of the absorption
index.19 He discussed limiting cases based on an analogue of eqn.
(14), and came to the same principal conclusions as we did, i.e.
Beer’s law should hold good for small “vertheilungsdichten” and
lose its justification for stronger absorptions and higher
concentrations. Even if Planck did neither derive Beer’s law
explicitly nor an expression for the molar attenuation constant, we
think that his insights should have deserved much more than
having gone unheeded for so long.
In addition, we recently succeeded in deriving Beer’s law in strict
form directly from basic electromagnetic theory without the detour
to dispersion theory. As a consequence, it is not necessary to
make any assumption about a specific model oscillator/lineshape
function and it can be seen as a generalization of this work.
Accordingly, the principle results, like the dependence of the
molar attenuation constant from the refractive index, and the
corresponding deviation from the empirical law for higher
concentrations, remain unchanged.20
References
[1] E. Hecht, Optics,4/e, Pearson Education, 2002.
[2] W. Sellmeier Annalen der Physik. 1872, 223, 386-403.
[3] W. Sellmeier Annalen der Physik. 1872, 223, 525-554.
[4] H. Helmholtz Annalen der Physik. 1875, 230, 582-596.
[5] E. Ketteler Annalen der Physik. 1887, 266, 299-316.
[6] H. A. Lorentz Koninkl. Ned. Akad. Wetenschap. Proc. 1906, 8, 591-611.
[7] P. Drude Annalen der Physik. 1893, 284, 536-545.
[8] C. Schaefer, F. Matossi, Das Ultrarote Spektrum, Verlag von Julius
Springer, Berlin, 1930.
[9] T. G. Mayerhöfer, J. Popp Journal of Optics A: Pure and Applied Optics.
2006, 8, 657-671.
[10] T. G. Mayerhöfer, J. Popp Chemphyschem : a European journal of
chemical physics and physical chemistry. 2018, accepted, DOI:
10.1002/cphc.201800961.
[11] T. G. Mayerhöfer, H. Mutschke, J. Popp Chemphyschem : a European
journal of chemical physics and physical chemistry. 2016, 17, 1948-1955.
[12] T. G. Mayerhöfer, H. Mutschke, J. Popp Chemphyschem : a European
journal of chemical physics and physical chemistry. 2017, 18, 2916-2923.
[13] T. G. Mayerhöfer, J. Popp Spectrochimica Acta Part A: Molecular and
Biomolecular Spectroscopy. 2018, 191, 283-289.
[14] T. G. Mayerhöfer, S. Pahlow, U. Hübner, J. Popp The Analyst. 2018, 143,
3164-3175.
[15] W. Mäntele, E. Deniz Spectrochimica acta. Part A, Molecular and
biomolecular spectroscopy. 2016.
[16] A. Y. Tolbin, V. E. Pushkarev, L. G. Tomilova Chemical Physics Letters.
2018, 706, 520-524.
[17] L. C. Zhao, M. H. Guo, X. D. Li, Y. P. Huang, S. H. Wu, J. J. Sun
Analytical chemistry. 2017, 89, 13429-13433.
[18] M. Planck Sitzungsberichte der Königlich Preussischen Akademie der
Wissenschaften. 1902, I, 470-494.
[19] M. Planck Sitzungsberichte der Königlich Preussischen Akademie der
Wissenschaften. 1903, I, 480-498.
[20] T. G. Mayerhöfer, J. Popp submitted. 2018.
COMMUNICATION
Entry for the Table of Contents

COMMUNICATION
Beer’s law – why absorbance Author(s), Corresponding Author(s)*
depends (almost) linearly on
concentration: Page No. – Page No.
Absorbance is, counterintuitively, not per
Absorbance
law
Title
se an additive quantity. This is a

's
hereditary from the absorption index,
which is a wave, rather than a materials
er
Be
property, and extends to a non-linear
concentration dependence. For weak
O
CDH
oscillators the deviations from linearity
are very weak, otherwise the non-
additivity may account for persistent
experimental errors.
Concentration