Академический Документы
Профессиональный Документы
Культура Документы
201700688 Articles
When band ratios in infrared absorbance spectra of films are waves in the film. We further investigate how transmittance
compared (which had been converted from transmittance and reflectance, as well as absorbance and the (from absor-
spectra), it can be noted that even after background correction bance) regained index of absorption, depend on the thickness
and removal of interference fringes these band ratios change of the film and how these parameters influence the positions
with the thickness of the films. The main goal of this work is to of bands. We compare the results with those for the incoher-
show that this effect is a consequence of an electric field ent case and the case where a single pass of light through the
standing wave based on the coherent superposition of light film without reflection loss is assumed.
1. Introduction
Measuring transmission is the standard modus operandi in in- as this is implied in Beer’s law (at least as it is used nowadays;
frared spectroscopy, since it requires neither sophisticated while Beer did not take into account the occurrence of multi-
sample preparation, nor additional accessories and the spectra ple reflections between the interfaces of a sample, he correct-
seem to be very simple to evaluate.[1] Being seemingly simple ed not only for reflection at the first, but also at the second in-
to evaluate is probably also the reason why certain other terface,[12] something that is currently not best practice) cannot
modes like attenuated total reflection and transflection are account for the peculiarities. Interpreting transflection spectra
quite popular too as they seem to give spectra comparable to and their dependence on sample thickness, it was, aside from
those obtained by transmission measurements. Recently, how- multiple reflections and coherent superposition of the waves
ever, it was shown that even though transflection spectra may inside the sample, assumed that an electric field standing
look like they resemble transmission spectra, they can be wave effect (EFSW) is responsible for the change of band
rather complex to evaluate under certain circumstances.[2–10] ratios.[5–9] This effect is responsible for the fact that for very
For films e.g., band ratios strongly change with film thickness thin films with thicknesses on the order of one to a few nm
and band positions can considerably shift.[11] Since transflection absorption is suppressed for normal incidence, since there are
spectroscopy is quite popular for biomedical applications,[5–9] no normal components of the electric field on the surface of a
because reflective substrates are available which are less costly metal (at least, if the metal is a perfect electric conductor).[13]
compared to infrared transparent substrates like for example, This is true independent of frequency (of course, not strictly
CaF2, in particular recently several studies were undertaken to for real metals, but for well-reflecting metals this holds well in
reveal and understand these peculiarities. Generally, literature the infrared spectral range), but the electric field shows a spa-
is consistent concerning the fact that approaches based on tial modulation perpendicular to the surface of the metal. Very
Maxwell’s equations which take into account proper optical recently it has been shown theoretically, that this EFSW effect
models can fully explain and quantitatively predict all these cannot be responsible for the peculiarities in IR transflection
peculiarities. From these approaches it is obvious that a single spectra.[10] Even more recently we discovered that there is
pass of light through an absorbing medium without reflection indeed an EFSW effect, different from the original one, that is a
coherence-based effect and can explain the changes of band
[a] Dr. T. G. Mayerhçfer, Prof. Dr. J. Popp intensity in terms of an interference enhanced and decreased
Spectroscopy/Imaging absorption.[14] In contrast to the original postulated EFSW
Leibniz Institute of Photonic Technology effect, the new effect does not only show a spatial modulation
Albert-Einstein-Str. 9, Jena, D-07745 (Germany)
but also a spectral variation of the electric field. Furthermore,
and
Institute of Physical Chemistry and Abbe Center of Photonics while the metal behind the layer enhances this EFSW effect, its
Friedrich Schiller University occurrence is not necessarily coupled to the presence of the
Helmholtzweg 4, Jena, D-07743 (Germany) metal or its properties. This suggests that this effect should
E-mail: Thomas.Mayerhoefer@ipht-jena.de
also influence measurements of films in transmittance mode.
[b] Dr. H. Mutschke
Indeed, deviations from Beer’s law in transmission experiments
Astrophysical Institute and University Observatory, AIU
Schillergßchen 2-3, D-07745 Jena (Germany) due to optical effects were already discussed earlier in the liter-
The ORCID identification number(s) for the author(s) of this article can ature,[3, 15–26] but apart from the fact that these deviations too
be found under https://doi.org/10.1002/cphc.201700688. can be fully predicted by applying Maxwell’s equations and a
ChemPhysChem 2017, 18, 1 – 9 1 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &
& ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 2 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 1. a) Electric field distribution in the system vacuum/layer/vacuum. The layer extends from 2.5 to 2.5 mm (indicated by black lines) and the layer is
non-absorbing with n2 = 1.5. b) Calculated reflectance, transmittance and absorptance spectra of the system above. c) Electric field distribution in the system
vacuum/layer/vacuum. The layer has the same thickness as in (a) but an index of refraction of n2 = 4. d) Calculated reflectance, transmittance and absorptance
spectra of the system in (c).
In the following we use polyethylene as an example, a materi- e1 = 2.25 Sj [cm1] gj [cm1] ñ [cm1]
al, which we already employed in [17] to illustrate our findings. #1 40.8 24.245 726.4
In order to allow the reader to reproduce our results we have #2 27.2 26.279 1367.9
used dispersion analysis based on Equation (3) to parameterize #3 52.2 18.164 1471.4
#4 205.1 23.903 2858.5
the optical constants, which were originally taken from [29].
#5 286.9 27.120 2921.7
The obtained oscillator parameters are listed in Table 1.
We also want to compare our results with those that would
be obtained assuming an incoherent superposition of the back
and forth reflected waves inside the layer. The incoherent T = 1, which means that reflectance must have been reduced
counterpart to Equation (1) is given by [Eq. (5)]:[17] by interference enhancement. The curve for the incoherent
case starts at about T = 0.9 due to the for this case non-vanish-
R12 ð1 R12 Þ2 e2md ð1 R12 Þ2 emd ing reflectance losses. Those losses are neglected in the Beer–
R ¼ R12 þ ; T ¼
1 R212 e2md 1 R212 e2md ð5Þ Lambert law, which is why the corresponding curve starts ex-
R12 ¼ j1j2 ; m ¼ 4 p~
nk 2 actly at T = 1. All three curves converge to zero and the inter-
ferences become increasingly damped by absorption. Obvious-
Furthermore, due to the definition of transmittance absor- ly, this kind of oscillation is caused by the interference fringes
bance (which will be in the following for simplicity only be very much the same way like the height of an object swim-
called “absorbance”) as A lg T and the Bouguer-Beer–Lam- ming on a water surface is periodically elevated and lowered
bert’s law (in the following referenced as Beer–Lambert law) by waves. If a minimum in the transmittance due to the inter-
according to which A = lg e 4 pñk2d, transmittance can be ex- ference fringes concurs with the minimum caused by absorp-
pressed as Equation (6): tion, the minimum value becomes smaller while it becomes
larger if the peak is at a position where the background has a
T ¼ Exp ½4 p~
nk2 d: ð6Þ maximum. For the stronger oscillator 4 this behavior is less
well developed, since absorption leads to a damping of the in-
In addition to T (d) we also investigated the wavenumber posi- terference fringes. Nevertheless, periodic changes of the trans-
tion of the minimum of the transmittance for oscillator 1 mittance are visible which become increasingly damped with
and 4. The results are depicted in Figure 2. increasing layer thickness.
For the weak oscillator 1 it is obvious that the transmittance Concerning the oscillator position it is known and can easily
for the coherent case oscillates around the values for the inco- be verified, that oscillator position and minimum in transmit-
herent case. Obviously, the starting point is a value close to tance are not identical, since the oscillator generates a maxi-
ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 3 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &
Figure 2. Left upper: Dependence of the position of the transmittance minimum on the thickness of the sample for oscillator 1. Left lower: Dependence of
the transmittance from the thickness of the sample for oscillator 1. Right upper: As for upper left but for oscillator 4. Right lower: As for lower left but for os-
cillator 4.
mum in the imaginary part of the dielectric function which tions might be less intense, at least for oscillator 1, they are
does not coincide with the maximum of k.[22] The spectral dif- still clearly discernable. We further observe that the common
ference between both maxima is increasing with oscillator starting point of the R + T spectra is now unity (since the Beer–
strength. For the weak oscillator 1 the maximum of k and the Lambert law, as it is used today, neglects reflection, T is equal
imaginary part of the dielectric function is not very different, to R + T in this case).
which is why true value and minimum of the transmittance However, the limiting value for R + T for the Maxwell’s equa-
nearly coincide. This is different for oscillator 4 (but the differ- tions compatible solutions is the reflectance of an infinitely
ences are still small compared to those that exist for strong os- thick sample and, therefore, not zero. Concerning the oscillator
cillators in inorganic materials!). positions, since we removed the influence of reflection for very
In the incoherent case the minimum of the transmittance is thin samples, the starting point for the incoherent case is
comparably far away from the oscillator position for small nearly identical to that for the Beer–Lambert line, but the
thicknesses. The reason for this behavior is that for small thick- values are increasing for higher thicknesses and developing to-
nesses the spectra look “dispersion-like” (the related effect is wards the minimum in reflectance, now for larger thicknesses.
called “dispersion distortion” in [22]) since it is mainly reflec- In the coherent case, the minimum in R + T is oscillating
tance and not absorption losses that shape the transmittance around the value for the incoherent case.
spectra[16] and the shape of the reflectance spectrum is mainly The most important observation, however, is the fact that
influenced by the real-part of the index of refraction. In con- we did only partly remove the oscillation in the R + T spectra
trast, in the coherent case, the minimum of the transmittance despite removing the interference fringes which is in line with
is not influenced by the reflectance spectrum (as reflectance is observations and conclusions in [29, 30]. What is responsible
very small for small thicknesses, at least for oscillators 1 and 4 for the remaining oscillation?
in polyethylene), and the minimum is close to the oscillator To comprehensively investigate this question, and to better
position. For increasing thickness, all three curves are converg- illustrate the results, we change our model system to one that
ing. is similar to the ones used in [6, 7, 10]. The model system is
A last point that needs to be emphasized is that the perio- now a fictive material having oscillators of the same compara-
dicity of the variations of the spectral position of the transmit- bly low strengths and the same damping constants every
tance minimum and of the transmittance itself are identical, 200 cm1 starting with 800 cm1 and ending at 3000 cm1
pffiffiffiffiffiffi
therefore both effects have the same origin (this situation is ( e1 ¼ nVis ¼ 4, Sj = 100 cm1, gj = 20 cm1 and ñj = (800 +
the same as in transflection).[14] Removing the interference 200j) cm1, j = 0,1,2,3…11.). The corresponding index of reflec-
fringes from the transmittance spectra (or reflectance spectra) tion and absorption coefficient functions are displayed in
is very easy for calculated spectra, as can be immediately seen Figure 4.
from Figure 1 and straightforwardly concluded from energy The low oscillator strengths lead to comparably small varia-
conservation, that is, R + T + A = 1. Simply, transmittance and tions of the index of refraction and to virtually the same value
reflectance must be added, and in non-absorbing regions R + of the absorption coefficient a (a = lg e·4 pñk2) at its maxima.
pffiffiffiffiffiffi
T = 1 holds since A = 0.[17] Therefore, we next investigate the The high value of e1 ¼ nVis ¼ 4 serves to emphasize the ef-
behavior of the reflectance corrected R + T spectra which fects and allows a comparison with the graphs of Figure 1 in
should be very similar to baseline-corrected transmittance the right panels, which we, for easier comparison, reproduce
spectra, as long as the sample thickness is not too small. The on the left side of Figure 5, albeit slightly altered: In addition
first observation by inspecting Figure 3 is that while the oscilla- to transmittance, reflectance and absorptance also hE2i is
& ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 4 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 3. Left upper panel: Dependence of the position of the R + T minimum on the thickness of the sample for oscillator 1. Left lower: Dependence of R + T
from the thickness of the sample for oscillator 1. Right upper: As for upper left but for oscillator 4. Right lower: As for lower left but for oscillator 4.
Figure 4. Index of refraction function (left: Real part in black, imaginary part in red) and absorption coefficient function (right) of the fictive material used to
study the enhancement and attenuation of bands by the EFSW effect.
shown, which is the averaged electric field intensity j E j 2/ j E0 j 2 curves share only the point at the origin). The cause of the de-
inside the layer. Based on the definition of transmittance, it is viations is the reflectance, which is not accounted for, when
not at all surprising that it scales with hE2i. If we introduce ab- absorbance is calculated according to the Beer–Lambert law
sorption, it is seen in Figure 5, right panels, that the electric (A = lg e·4 pñk2d). The curves for the incoherent case resemble
field is lowered in close proximity to the oscillator positions. those that are obtained if stray light is considered as the
The oscillators are located in spectral regions with different source for deviations from the Beer–Lambert law,[1] since it
hE2i. Those in regions of lower electric field strengths show does not matter in the end if the light does not reach the de-
lowered absorptance with the peak height scaling with hE2i. tector because it has been reflected or gone astray.
Accordingly, the corresponding EFSW effect could also be From the absorbance curves, the starting point, at least as
termed “interference enhanced and decreased absorption”. far as the absorption index is concerned, could be recovered
This effect fully explains the remaining oscillations in the R + T by k2 = A/(lg e·4 pñd).[31] In Figure 7 we show the dependence
spectrum since A is nothing else but 1(R + T). With changing of the peak values of k2 for the five oscillators of polyethylene
layer thickness the spectral position of higher and lower hE2i ratioed to the corresponding starting values. Accordingly, if the
are periodically altered [cf. Eq. (4)]. These alterations obviously Beer–Lambert law would be correct, all curves should be
cannot be removed by removing interference fringes by straight lines through unity parallel to the x-axis.
adding R + T or by any other kind of baseline correction. Indeed, for the very weak oscillator located at 1368 cm1 the
Going back to polyethylene, it is very instructive to examine deviations from unity are comparably small. In contrast, since
how the absorbance depends on the thickness of the layer, the absorbance reaches the limiting constant value given by
that is, to investigate the adherence of the absorbance to line- lg R12 (a sample that is thick enough so that T is virtually
arity. This dependence is depicted in Figure 6. As can be ob- zero) as can be seen in Figure 6 early in case of oscillator 4
served in this Figure, absorbance starts to deviate very early (and even earlier for oscillator 5) k2 calculated according to
from linearity (to be more precise, for the incoherent case, the Beer–Lambert quickly decays. In the coherent case, the weak
ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 5 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &
Figure 5. a) Electric field distribution in the system vacuum/layer/vacuum assuming the layer is transparent and has an index of refraction of n2 = 4. The layer
extends from 2.5 to 2.5 mm. b) Simulated reflectance, transmittance and absorptance spectra of the system above. In addition, the averaged squared abso-
lute value of the electric field hE2i is shown, which is proportional to the transmittance. c) Electric field distribution in the system vacuum/layer/vacuum. The
layer is the same as in (a) but absorbing with the optical constants depicted in Figure 4. d) Simulated absorptance spectra of the same system as in (c). Addi-
tionally, the averaged squared absolute values of the electric field hE2i are presented.
Figure 6. Variation of the absorbance for polyethylene assuming coherent (black) and incoherent (red) superposition of the waves inside the layer. The green
line is the corresponding absorbance for a layer according to the Beer–Lambert law. Left panel: Oscillator 1. Right panel: Oscillator 5.
oscillators start at values between 1.4 and 1.5 and then oscil- but as in A = lg (Tc/Tc = 0), wherein Tc is the transmittance with
late comparably strongly. a not too highly concentrated solution and Tc = 0 is the transmit-
For the stronger oscillators, this oscillation becomes rapidly tance of the pure solvent. The latter works, because this defini-
damped. The shift in the positions of the maxima and the oc- tion of the transmittance absorbance conforms, to a good ap-
currence of second maxima is related to the transformation of proximation, with the solutions based on Maxwell’s equations
spectra that resemble the index of absorption to those that for the incoherent case.[17]
are dispersion like as already discussed. Unfortunately, in particular in recent literature there is a
Overall, our conclusion is the same as in [17]. Accordingly it trend to use the easy path when absorption indices of pure
is not advisable to calculate k2 from absorbance spectra substances are to be calculated, even though a comparably
except if there is a situation as in spectrophotometry where simple but Maxwell’s equations compatible method like disper-
absorbance is actually not used as in A = lg (I/I0), wherein I is sion analysis[32] is available or iterative methods[3, 22] can be
the irradiance after and I0 is the irradiance before the sample, chosen, which are, however, somewhat more complex.
& ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 6 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 7. Upper panel: Spectral positions of the peak maxima relative to the positions of the peak maxima in the original k2. Lower panels: Values of the peak
maxima ratioed to those of the original k2. Left: Coherent superposition of the waves. Right: Incoherent superposition of the waves.
We investigated the thickness dependences of transmittance, Financial support of the EU, the “Thringer Ministerium fr Wirt-
absorptance, absorbance and index of refraction determination schaft, Wissenschaft und Digitale Gesellschaft”, the “Thringer
according to the Beer–Lambert law on the example of a free- Aufbaubank”, the Federal Ministry of Education and Research,
standing film sandwiched by vacuum. In thin films, interfer- Germany (BMBF), the German Science Foundation, the Fonds der
ence leads to periodic variations of the transmittance as well Chemischen Industrie and the Carl-Zeiss Foundation are greatly
as to the well-known occurrence of interference fringes in acknowledged.
spectra. The latter can be removed by adding reflectance and
transmittance. The resulting spectrum is given by 1—absorp-
tance. Despite removing the interference fringes, still periodic Conflict of interest
changes in the thickness dependence of the absorptance can
be observed. To determine their origin, we analyzed the elec- The authors declare no conflict of interest.
tric field distribution in- and outside of the layer and found
that there exists an electric field standing wave effect inside Keywords: absorbance · electric field enhancement · electric
the layer that has both, a spatial and a spectral variation. Due field standing wave effect · infrared spectroscopy ·
to the spectral variation, also the electric field intensity and transmission
thereby absorption varies. We further investigated the adher-
ence of the absorbance to linearity. We found that with in- [1] J. M. Chalmers, P. R. Griffiths, Handbook of vibrational spectroscopy,
creasing reflectance of the sample absorbance stronger devi- Wiley, New York, 2002.
[2] H. Brooke, B. V. Bronk, J. N. McCutcheon, S. L. Morgan, M. L. Myrick,
ates from linearity. For polyethylene, which has a comparably
Appl. Spectrosc. 2009, 63, 1293 – 1302.
low reflectance, the values begin to noticeably deviate from [3] B. J. Davis, P. S. Carney, R. Bhargava, Anal. Chem. 2010, 82, 3474 – 3486.
linearity at absorbance values of 0.5. A consequence of this be- [4] B. J. Davis, P. S. Carney, R. Bhargava, Anal. Chem. 2010, 82, 3487 – 3499.
havior is the in former times well-known fact that a determina- [5] J. Filik, M. D. Frogley, J. K. Pijanka, K. Wehbe, G. Cinque, Analyst 2012,
137, 853 – 861.
tion of the index of absorption by applying the Beer–Lambert
[6] P. Bassan, J. Lee, A. Sachdeva, J. Pissardini, K. M. Dorling, J. S. Fletcher, A.
law is generally not advisable. Henderson, P. Gardner, Analyst 2013, 138, 144 – 157.
[7] T. P. Wrobel, B. Wajnchold, H. J. Byrne, M. Baranska, Vib. Spectrosc. 2013,
69, 84 – 92.
[8] D. Perez-Guaita, P. Heraud, K. M. Marzec, M. de la Guardia, M. Kiupel,
B. R. Wood, Analyst 2015, 140, 2376 – 2382.
[9] E. Staniszewska-Slezak, A. Rygula, K. Malek, M. Baranska, Analyst 2015,
140, 2412 – 2421.
ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 7 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &
& ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 8 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemPhysChem 2017, 18, 1 – 9 www.chemphyschem.org 9 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &