You are on page 1of 7

In the Classroom

edited by
Computer Bulletin Board    Steven D. Gammon
Western Washington University
Bellingham, WA  98225

Beyond 𝛌max Part 2: Predicting Molecular Color


Darren L. Williams,* Thomas J. Flaherty, and Bassam K. Alnasleh
Department of Chemistry, Sam Houston State University, Huntsville, TX 77340; *williams@shsu.edu

The chemistry curriculum typically describes the color of but the major goal of “increasing student engagement with the
chemical species in generic terms. Color wheels are used to ex- material” was obviously achieved as evidenced by their desire to
plain the apparent color of an object based upon the absorption come forward after lecture to “play” with the spreadsheet tool.
of a complementary color (1). Some introductory texts contain This same response occurred among faculty and high school
tables of wavelengths of maximum absorption (λmax) with the instructors when early versions of this work were presented.
complementary observed colors and the advanced-level texts are The most common question was, “What happens if there are
not much better (2). These explanations are appropriate for the two peaks?”
introductory course, but color could be used to introduce the In the physical chemistry curriculum this work is used to
students to the general spectroscopic concepts of absorbance and strengthen student understanding and ability in the following
transmittance. Then, a substantial effort can be made to familiar- areas:
ize the chemistry major with standard colorimetry during the
• Computational chemistry and the electronic structure of
physical chemistry course.
molecules
The Standard Colorimetric Observer (3) analysis published
by the Commission Internationale de l’Éclairage (CIE) (4) • Molecular orbital theory and spectroscopic transitions
has been the objective benchmark for color specification for • The transformation of absorbance, reflectance, and trans-
over 75 years. International standards have been developed to mittance spectra
calculate the CIE tristimulus values and 24-bit standard RGB • Spectral line shapes of vibrational–electronic (vibronic)
(sRGB) values from visible spectra (5–7), and an outline of these transitions
procedures was published in this Journal as part one (8) of this
• Molecular structure changes and concomitant color
two-part series of articles.
shifts
The technical details behind color prediction are fairly
intimidating, but the color prediction spreadsheet (described Electronic structure calculations are being used more and
herein) is suitable for use in introductory chemistry lectures on more in the physical chemistry sequence. A recent article in this
color. All of the theoretical details do not need to be explained Journal (10) provides adequate instructional content to enable
to see what color results from a given λmax and peak shape. It has students to compute the geometry and visible spectroscopic
been our experience that after the students see the color ranges transitions of several azulene derivatives. We extend the analysis
of different λmax values, they begin asking questions about the into a specific prediction of the range of colors available from the
displayed spectra. Here, the qualitative descriptions of absorbed computed spectroscopic transitions. The students are fascinated
and transmitted light were found to be sufficient. by color and are eager to learn how one can tease the color of a
For the physical chemistry curriculum, this article provides molecule out of the computational results.
a systematic process for generating absorbance and transmittance Our scheme serves as a nice capstone to the teaching pro-
spectra of a molecular species over a range of theoretical concen- gression of electronic spectroscopy from MO diagrams of di-
trations. The theoretical transmittance spectra are manipulated atomic molecules, to HOMO–LUMO transitions, to this work
using the scheme described earlier (8), and the sRGB values are where the transition dipole approximation is used to predict the
visualized using a macro within a spreadsheet. Students who use wavelength and intensity of the spectroscopic lines.
this colorimetry spreadsheet in the laboratory have experienced The importance of the Beer–Lambert law to the absorbance
firsthand how inadequate a simple λmax description of color can spectrum is illustrated by the concentration multiplier, and the
be, and they have demonstrated an improved understanding student can easily see the linear dependence in the simulated
of the importance of line shape, weakly absorbing bands, and series of absorbance spectra. The importance of spectral satura-
spectral saturation in spectroscopy and colorimetry. tion is observed in the transmittance spectrum, where certain
The accuracy of this predictive colorimetry process is il- portions of the visible spectrum are completely blocked as con-
lustrated by predicting the colors of phenolphthalein and of centration increases. Both of these play a role in the resulting
nine azulene derivatives published in this Journal (9, 10) and color prediction, and it is very interesting to see the change of
elsewhere (11). hue as the concentration multiplier increases.
The broad peaks and the asymmetric line shapes used in
Pedagogical Approach this project are useful for connecting vibrational and electronic
spectroscopies. Physical chemistry students are typically curi-
We make use of this spreadsheet tool in the introductory ous about the asymmetric line shape, and their questions often
chemistry course when spectroscopy is introduced with the line lead to a discussion of selection rules, force constants, and the
spectra of hydrogen and during the coordination chemistry electronic potential energy surfaces (PES) that give rise to the
chapter. Statistical data on student learning are not available, different line shapes. We find it useful to refer to the diatomic

© Division of Chemical Education  •  www.JCE.DivCHED.org  •  Vol. 86  No. 3  March 2009  •  Journal of Chemical Education 333
In the Classroom

S  0.5 S  2.0
color prediction are sensitive to the geometric parameters of
Conc’n Conc’n electron-rich moieties, and low-level optimizations may yield
Factor R G B Factor R G B
different geometries from higher-level (presumably more ac-
1 255 237 239 1 242 240 243
2 255 218 237 2 231 224 245
curate) calculations.
4 255 186 232 4 210 198 248 The visible spectral transitions are computed using
8 255 136 223 8 174 157 254 Zerner’s method of intermediate neglect of differential overlap
16 255 76 203 16 120 109 255
(ZINDO) (19). This computational package is distributed
Relative Absorbance

32 255 28 167 32 55 72 255


64 255 7 114 64 0 56 255 within many computational chemistry software packages, and
255 4 58 0 51 255
128
256 255 5 16
128
256 0 47 255
it is fast due to its semi-empirical nature. The method used in
512 255 6 0 512 0 45 255 this article is traditionally denoted as ZINDO//B3LYP/6-
1024 255 5 0 1024 0 43 255 31G(d), which reads, “a ZINDO calculation at the B3LYP/6-
31G(d)-optimized geometry”. The ZINDO output produces
a table of transition wavelengths (λi) and oscillator strengths
( fi ). The same output can be obtained with other, more com-
putationally expensive methods such as time-dependent DFT,
TD/B3LYP/6-31G(d)//B3LYP/6-31G(d), or configuration-
400 500 600 700
interaction with single-electron excitation, CIS/6-31G(d)//
Wavelength / nm B3LYP/6-31G(d) (17).
Figure 1. The lognormal lineshape (eq 2) displayed with fi = 0.05, The oscillator strength is directly proportional to the
λmax = 550 nm, w = 60 nm, and two different asymmetry parameters. integrated intensity of an absorption peak (20). If the spectral
The colors of each are displayed for concentration multipliers ranging full width at half-maximum (FWHM) is similar among the
from 1x to 1024x. The solid line goes with the solid-outlined box, and transitions, then the maximum molar absorptivity (εmax) is ap-
the dashed line goes with the dash-outlined box.
proximately proportional to the oscillator strength:
F max M i
u fi (1)
PES that the students use in the iodine visible absorption labora-
tory, with an explanation that there are three more dimensions To more accurately model the visible spectrum, the stick spec-
(degrees of freedom) added to this surface with each additional trum is convolved with a lognormal line shape as shown in eq
atom. Even in the simple case of I2, the asymmetric line shape 2. The lognormal line shape is used because visible absorption
is needed to model the band head in the rotation–vibrational line shapes are often asymmetric (21) due to a multitude of
manifold and the overall shape of the vibronic manifolds. unresolved vibronic transitions.
Finally and most importantly, the scheme provides a
seamless connection between the molecular structure and the A M

molecular color (i.e., the color of a single molecule in space with 2

2
no external fields). There are no solvent effects, crystal fields, or ln 2 M  M i
S2  1 (2)
particle-size effects, but there is a concentration multiplier that fi exp  ln 1
illustrates the effects of spectral saturation and the resulting ln S
2 wS
changes in color intensity and hue. There is a common mis-
perception that the hue does not change with concentration, where
but the hue changes with strongly absorbing species and highly
concentrated solutions that result in spectral saturation. The HWHM
high M high  M max
S  
spreadsheet clearly shows this effect and can be used to explain HWHM
low M max  M low
the concentration-dependence of color.
The w in eq 2 is the user-defined FWHM, and A(λ) is the
Computational Scheme effective absorption intensity at wavelength λ (22). The asym-
metry parameter or half-width at half-maximum (HWHM)
The equations in this computational scheme are fully imple- ratio (ρ in eq 2) determines the skewness of the peak. The peak
mented in the color simulation spreadsheet that is provided shape is skewed to short wavelengths (ρ < 1) when the excited
in the online material. The student builds the molecules in a electronic state PES has a minimum directly above the lower
graphical user interface such as GaussView (12). Other inter- electronic state PES minimum and vibrational excitation of the
faces that have been tested are WebMO (13), Hyperchem (14), lower state is low. This particular PES arrangement can produce
ArgusLab (15), and Spartan (16). WebMO and ArgusLab are peak shapes skewed to long wavelengths (ρ > 1) if vibrational
licensed to academic institutions without cost. excitation of the lower state is plentiful. Symmetric line shapes
The optimized molecular geometry is obtained using (ρ ≈ 0.999) can arise from the typical Franck–Condon factor
Gaussian 03W (17) with the hybrid density functional (DFT) intensities of the vibrational manifold if the PES minima are
model chemistry of B3LYP and the basis set of 6-31G(d). This not directly above each other and there is minimal vibrational
model chemistry—denoted as B3LYP/6-31G(d)—exhibits excitation in the lower state. The lognormal distribution is not
moderately accurate thermochemical results (18) with reason- defined when ρ = 1, but ρ can be very close to one to model
able speed on standard desktop PCs with 2 GHz processors symmetric peaks.
with 1 GB of RAM. The semi-empirical geometry optimizations The asymmetry parameter captures the essence of the arti-
described in ref 10 should be used for molecules with more than cle’s title “Beyond λmax”. Two peaks with the same λmax and w are
50 atoms. However, the electronic spectrum and the resulting shown in Figure 1. The color-prediction spreadsheet (Figure 1)

334 Journal of Chemical Education  •  Vol. 86  No. 3  March 2009  •  www.JCE.DivCHED.org  •  © Division of Chemical Education 
In the Classroom

was used to generate the range of colors that could result from
this λmax for two different asymmetry parameters. This figure
demonstrates that tables of λmax and absorptivity are inadequate
to fully describe the color of a substance. Ideally, the width and
peak shape should also be reported, or a specific sRGB range
should be used to report color in the literature.
The choice of the asymmetry parameter and peak width
is problematic in a purely theoretical situation. A full PES scan
of the ground state and all relevant excited states is impracti-
cal. It is a good start to simply choose a symmetric line shape
(ρ = 0.999), but it is often useful to match the peak width and
line shape of an experimental spectrum of a similar molecular
system (if available).
The transition wavelengths (λmax), oscillator strengths ( fi ), Figure 2. The simulated absorption spectra, transmittance spectra,
and estimates of peak width (w) and shape (ρ) are entered into chromaticity diagram, and color of azulene using the ZINDO//
the spreadsheet (Figure 2) by the user, and the rest of the data B3LYP/6-31G(d) model.
manipulation is performed automatically by the spreadsheet
formulas. Each transition is convolved with the lognormal line
shape at 5 nm intervals over the wavelength range of 400 to
700 nm. The intensity of absorption from all peaks is summed 2 O O O
at each wavelength to create the overall absorption spectrum 1 3 F F
of the molecule. Conversion of this absorption spectrum to
24-bit sRGB values (via the transmittance spectrum) often
4
yields a colorless result because the ZINDO-calculated oscil- 8

lator strengths ( fi ) are typically small. The effective absorption 7 5


spectrum based upon these fi is multiplied by a concentration 6
multiplier to give a series of absorbance spectra representing
I II III IV
increasing concentrations. azulene 1-CHO 1,3-diCHO
1,3-diF
The concentration multiplier is an artificial way to intensify
the simulated spectrum of a single molecule so that it represents O
the spectrum of an ensemble of molecules with no intermolecu-
lar interactions. In the color prediction spreadsheet, absorbance F
spectra at ten concentration increments are converted to trans- O
mittance spectra, and the transmittance spectra are converted to
24-bit standard sRGB color values using the scheme described
earlier (8). A macro button has been placed on the spreadsheet
(Figure 2) that displays the colors of the sRGB values for the
whole series of concentrations. O O
The color bars, absorbance spectra, transmittance spectra, V VI VII VIII IX
and chromaticity diagram are all on the same page so that com- 1-F 2-CHO 4-CHO 5-CHO 6-CHO
parisons can be made. The results of concentration are evident
in the transmittance spectra in Figure 2 where the peaks saturate Figure 3. Compounds used to predict the colors.
and the edges of the transitions become important. The effects of
this “saturation series” of spectra can be seen in the chromaticity
diagram where the hue begins to curve. Saturation is also seen in
the sRGB color bars as they become more and more opaque.
9–11. The colors of other derivatives (V–IX) found in ref 11 are
Results also predicted for comparison purposes.
The B3LYP/6-31G(d)-optimized structures had the ex-
This method has been applied successfully to many colored pected planar geometry. The ZINDO visible transitions are
species such as nitroaromatics, phenylazo dyes, phenolphthalein shown in Table 1 for all nine structures. Line widths of 60 nm
(protonated and deprotonated), and various azulene derivatives. and asymmetry parameters of 0.9 were used for all spectroscopic
The azulenes are presented here because they illustrate substitu- transitions, and the resulting absorbance spectra qualitatively
ent effects on color, and phenolphthalein is presented to show match the experimental spectra of these four species reported in
an example of structural effects on color. ref 9. The concentration increment of 2× (Figure 2) was used to
compute the range of colors possible from these spectral transi-
Azulenes Color Predictions tion values, and the sRGB results for each species are included in
The analysis of azulene (I), 1,3-difluoroazulene (II), Table 1. The background color of the sRGB cells represents the
1-formylazulene (III), and 1,3-diformylazulene (IV) in this simulated color of each compound. The sRGB values are given
article (Figure 3) provides a complement to the earlier inter- so that readers with black-and-white copies of this article can
pretation of the visible spectra of these molecules found in refs reproduce the colors on their own computer systems.

© Division of Chemical Education  •  www.JCE.DivCHED.org  •  Vol. 86  No. 3  March 2009  •  Journal of Chemical Education 335
In the Classroom

Table 1. The Simulated Lowest-Energy Singlet Transition Wavelengths, Oscillator Strengths, and Colors of Species
Species λ2/nm f2 λ1/nm f1 Multiplier sRGB
I Azulene 386 0.037 583 0.024 64x [22 106 189]
II 1,3-diF 390 0.086 618 0.029 64x [0 212 135]
III 1-CHO 395 0.160 553 0.020 64x [186 31 92]
IV 1,3-diCHO 395 0.191 532 0.017 64x [255 67 65]
V 1-F 388 0.062 602 0.027 64x [0 167 160]
VI 2-CHO 385 0.001 609 0.034 64x [0 166 255]
VII 4-CHO 403 0.016 643 0.032 64x [0 255 151]
VIII 5-CHO 414 0.186 567 0.021   8x [204 221 91]
IX 6-CHO 396 0.001 644 0.028 64x [12 249 237]
Note: The data are from the model ZINDO//B3LYP/6-31G(d) using w = 60 nm and ρ = 0.9.

Table 2. The Color Predictions Using the λmax and Absorptivity Data from Ref 11
ε 2/ ε1/
Species λ2/nm w/nm λ1/nm w/nm Multiplier sRGBa
(10–5 M–1 cm–1) (10–5 M–1 cm–1)
I Azulene 341 0.0401 60 580 0.0348 120 64x [0 20 226]
II 1,3-diF 342 0.0329 60 670 0.0144 120 512x [0 235 138]
III 1-CHO 385 0.0100 60 542 0.0467 120 8x [184 111 220]
IV 1,3-diCHOb 384 0.1055 60 486 0.0820 120 8x [245 102 52]
V 1-F 343 0.0326 60 625 0.0323 120 8x [120 232 243]
VI 2-CHO 360 0.0631 60 664 0.0155 120 64x [44 255 155]
VII 4-CHO 354 0.0195 60 642 0.0355 120 64x [0 194 226]
VIII 5-CHO 383 0.0901 60 571 0.0450 120 8x [119 143 189]
IX 6-CHO 351 0.0575 60 652 0.0363 120 128x [0 200 160]
a   b
The colors were generated using the specified multiplier, peak width, and ρ = 1.3. Solvent is CH2Cl2; all others in hexane.

Table 3. The Experimental Color and ZINDO//B3LYP/6-31G(d) Output for XI with Two Different Dihedral Angles of the Acid Moiety

Species λ2/nm f2 λ1/nm f1 Multiplier sRGBa


XI 0° 363 0.0426 505 1.035 0.8x [255 130 165]
XI 90° 372 0.0013 517 1.066 0.8x [255 96 198]
Experiment 374 0.234b 549 1.000b 0.8x [244 71 246]
a
The shading of the cells corresponds to the sRGB values in each row computed using a w = 60 nm, ρ = 0.9 and a multiplier of 0.8x.
b
Relative absorbance values in a basic aqueous solution.

To further illustrate the benefits of this color prediction absorption transition from the ground state to the second singlet
tool, Table 2 shows the sRGB values for each of the azulene excited state (2←0) was chosen to be 60 nm, and the width of
derivatives using the transition wavelengths and absorptivities the (1←0) transition was chosen to be 120 nm. A half-width
reported in ref 11. The molar absorptivities were divided by 105 ratio of ρ = 1.3 was used for all peaks. The resulting color defini-
to approximate the magnitude of the oscillator strength. Using tions are much more specific than the simple words green, pale
the experimental spectra in ref 11 as a guide, the width of the green, blue, violet, and so forth.
Phenolphthalein Color Predictions
O O The same model chemistry (ZINDO//B3LYP/6-31G(d))
was applied to the protonated (X) and deprotonated (XI) forms
O O of phenolphthalein (Scheme I). The fully protonated form
ź2 Há
was correctly predicted to be colorless while the fully deproto-
nated form with the broken lactone bridge was predicted to be
O strongly colored, matching the experimental results of Tamura
et al. (23).
HO OH
O The structural details of XI were quite interesting. When
X XI the lactone bridge was broken and the acid moiety was planar
colorless pink to the ring, the theoretical color was a red hue, not the familiar
Scheme I. The protonated and deprotonated forms of phenol­ pink color of phenolphthalein (Table 3). As the acid moiety
phthalein. dihedral angle with the ring was rotated to 90°, the theoretical

336 Journal of Chemical Education  •  Vol. 86  No. 3  March 2009  •  www.JCE.DivCHED.org  •  © Division of Chemical Education 
In the Classroom

visible absorption exhibited a bathochromic shift from 504 using electron density difference maps (12, 17) that describe
to 517 nm—opening a window in the blue part of the visible the total change in the electron cloud by subtracting the total
spectrum. This gave the familiar pink–purple color of basic electron density of the ground state from the total electron
phenolphthalein. density of the excited state. The electron density difference is
then mapped upon the ground-state electron density surface
Substituent Effects to provide a colored isosurface. These images were generated
for the azulene transitions (Figure  4) and show the electron
The molecular orbitals for azulene were generated us- density decreasing on the 5-membered ring and increasing on
ing GaussView (12) and are shown in Figure 4. The affected the 6-membered ring.
molecular orbital energies of I–IV are plotted relative to the These calculations are well suited for use in a physical
LUMO level for each species. The LUMO was chosen as the chemistry laboratory environment. The geometry optimization
energy reference since it interacts the least with substitution on steps for the azulenes were finished in approximately 1 hour
atoms 1 and 3. each on a range of PCs with processors in the 2 to 3 GHz range
As discussed in refs 9–11, the electron-donating fluorine with 1 to 2 GB of RAM. The phenolphthalein calculations were
substituents at positions 1 and 3 of azulene caused a bathochro- performed on a Dell Precision Server with 2 quad-core CPUs
mic shift of the HOMO to LUMO transition by increasing the and 32 GB of RAM. Only 4 processors were reserved for the
energy of the HOMO and LUMO + 1 relative to the energy geometry optimization and frequency calculation, and the CPU
of the LUMO. This change in the molecular orbital energies times were 3 hours and 7 hours, respectively. The ZINDO cal-
opened a window in the green portion of the spectrum yielding
the green hue of II.
The electron-withdrawing groups on atoms 1 and 3 (spe-
cies III and IV) caused a hypsochromic shift of the HOMO to
LUMO transition by reducing the energies of the HOMO and
LUMO + 1 relative to the LUMO. The HOMO to LUMO  + 1
difference remained constant because the conjugation affected
these levels similarly. Disubstitution (IV) reduces these orbital
energies slightly more than III resulting in more absorption in
the blue region of the spectrum. Thus, the pink hue of III is
changed to the red hue of IV.
The molecular orbitals of phenolphthalein are shown
in Figure 5. The HOMO energy is essentially unchanged by
rotation of the acid moiety. The LUMO energy is reduced by
rotating the electron-rich oxygen atoms out of the area occupied
by the LUMO wave function. The resulting bathochromic shift
reproduces the pink color well. The ZINDO calculations predict
the XI-90° structure to be 3.5 kcal/mol higher in energy than
the XI-0° structure in the gas phase. This energy change is small Figure 4. The molecular orbitals of azulene, electron density
enough to be overcome by dissolution especially given the fact difference maps for the azulene transitions, and the molecular orbital
that the rotation of the acid moiety opens up the possibility of energies of species I through IV plotted relative to the LUMO for each
hydrogen bonding to both carboxylic acid oxygen atoms. species. The electron density difference maps indicate a decrease
(red) in electron density on the 5-membered ring and an increase
(blue) in electron density on the 7-membered ring.
Discussion
The phenylazo dyes and nitroaromatics were not included
because their simulations are complicated by the need to model
thermally excited vibrational states. The phenylazo dyes in par-
ticular exhibit planar structures in the ground state that create
symmetry-forbidden n-to-π* transitions. The vibrational modes
that bend the phenylazo dihedral are thermally excited at room
temperature since their vibrational frequencies are well under
200 cm–1—the quantity of thermal energy available at room
temperature. A procedure for including the color of vibrationally
excited geometries would lengthen this article considerably and
thus is available separately from the corresponding author.
The configuration interaction results from the ZINDO
calculation (Table 1) clearly match the ground-state molecular
orbital energies and images for the azulene derivatives. However,
this is not always the case. Assigning a spectroscopic transition
to two ground-state molecular orbitals is not valid when there Figure 5. The HOMO and LUMO of deprotonated phenolphthalein
is considerable mixing of the orbitals involved in a spectroscopic (XI) with two acid moiety dihedral angles (0° and 90° on the left and
transition. In these situations, the transition is better described right, respectively).

© Division of Chemical Education  •  www.JCE.DivCHED.org  •  Vol. 86  No. 3  March 2009  •  Journal of Chemical Education 337
In the Classroom

Table 4. The Color Predictions Using the λmax Data from Ref 10 Compared to Data in Tables 1 and 2
Experimenta ZINDO//INDO/1b ZINDO//B3LYP/6-31G(d)c
Species λ2/nm λ1/nm [RGB] λ2/nm λ1/nm [RGB] λ2/nm λ1/nm [RGB]
I Azulene 341 580 [0 20 226] 352 587 [0 35 212] 386 583 [22 106 189]
II 1,3-diF 342 670 [0 235 138] 359 637 [0 135 92] 390 618 [0 212 135]
III 1-CHO 385 542 [184 111 220] 362 520 [225 114 193] 395 553 [186 31 92]
IV 1,3-diCHO 384 486 [245 102 52] 356 486 [248 99 75] 395 532 [255 67 65]
V 1-F 343 625 [120 232 243] 356 614 [115 220 240] 388 602 [0 167 160]
VI 2-CHO 360 664 [44 255 155] 352 624 [0 188 188] 385 609 [0 166 255]
VII 4-CHO 354 642 [0 194 226] 383 672 [0 255 156] 430 643 [0 246 119]
VIII 5-CHO 383 571 [119 143 189] 392 569 [122 142 168] 414 567 [204 221 91]
IX 6-CHO 351 652 [0 200 160] 366 674 [0 252 102] 396 644 [12 249 237]
a
Ref 11 in hexane with IV in CH2Cl2. Data are also reported in Table 2. bRef 10 simulated with no solvent. Data uses the same widths, absorptivities,
asymmetry parameters, and multipliers used to model the ref 11 sRGB. cThis study with no solvent. Data are also reported in Table 1.

culations required only 30 seconds for each molecule. The fully Literature reports of color should follow the systematic
functional spreadsheet was provided to the students for their definition described earlier (8), and the reporting of predicted
use as a color-predicting application. color of a molecular species should contain the following mini-
The color prediction with the INDO model used in ref 10 mum details: (i) the computational model; (ii) the line shape
was not reliable for phenolphthalein in that the greatly distorted parameters; and (iii) the sRGB values at the concentration mul-
INDO geometry exhibited extra transitions at 713 nm and tiplier where the transmission spectrum begins to saturate. For
426 nm that are not observed experimentally. The INDO model example, one could systematically report the simulated color of
did quite well for the azulene derivatives as seen in Table 4, and azulene as “[22, 106, 189], computed via ZINDO//B3LYP/6-
the method of ref 10 is superior for completing the full range of 31G(d), lognormal, ρ = 0.9, w = 60 nm, 64×”.
azulene derivative computations in one laboratory period. In summary, the color prediction spreadsheet can be used
The computational models described here reproduce the in multiple places in the chemistry curriculum. It can be used to
colors of I, II, and IV well. The difficulty in reproducing the teach spectroscopic concepts and is useful for increasing student
colors of III, and V–IX stem from an over-estimation of the engagement of the material. Additionally, we commend those
wavelength of the λ2 transition. The mismatch between simula- who developed the azulene derivative work (9–11). We sincerely
tion and experiment could be due to the effect of solvent in the hope that this spreadsheet tool will find use in their laboratories
experimental spectra, but could also be due to the limitations of as their work has improved ours.
the ZINDO simulation (19). Spectra of each species in several
solvents would elucidate any appreciable solvent effect, and Acknowledgments
calculations using time-dependent density functional theory
or configuration interaction singles would elucidate any com- The authors thank the Welch Foundation Departmental
putational deficiencies present in the transition wavelength Development Grant and the Sam Houston State University
predictions (17). Chemistry Department for funding the student research as-
sistants who contributed to this project—Stephanie Coleman,
Conclusions Casie Jupe, Karl Kuklenz, Kara Marquez, and Jamie Tolleson.

The visual ramifications of functional group changes on Literature Cited


azulene serve as a good test of the utility of molecular modeling
and the color determination method presented in this article. 1. Brown T. L.; LeMay H. E.; Bursten B. E. Chemistry: The Central
The difficult geometry optimization of phenolphthalein illus- Science, 10th ed.; Pearson Education: Upper Saddle River, NJ,
trated the necessity of higher levels of computational chemistry 2006; pp 1044–1045.
theory. In most cases, the computational methods for predicting 2. Zumdahl S. S.; Zumdahl S. A. Chemistry, 7th ed.; Houghton
color performed satisfactorily, and one can see that sometimes Mifflin Company: New York, 2007; pp 969–971.
drastic changes in λmax do not cause drastic shifts in color (spe- 3. Broadbent A. D. Color Res. Appl. 2004, 29 (4), 267–272.
cies IV). One can also see that relatively modest changes in the 4. Commission Internationale de l’Éclairage. CIE News [Online],
blue region can change the color very much (species VIII). 2006, 77, 1–16.
If the predicted experimental colors in Table 2 do not 5. ASTM E308-01, Standard Practice for Computing the Colors
match the actual colors of the azulene derivatives, then our of Objects by Using the CIE System; ASTM International: West
point is reinforced. The λmax and absorptivity were available in Conshohoken, PA.
the literature (9–11), but this information was inadequate for 6. International Color Consortium. http://www.color.org/ (accessed
specifying the color unequivocally. It was necessary to estimate Nov 2008).
peak widths and asymmetry parameters from spectra presented 7. Stokes, M.; Anderson, M.; Chandrasekar, S.; Motta, R. A. Stan-
in these articles. dard Default Color Space for the Internet: sRGB Version 1.10.

338 Journal of Chemical Education  •  Vol. 86  No. 3  March 2009  •  www.JCE.DivCHED.org  •  © Division of Chemical Education 
In the Classroom

http://www.color.org/sRGB.html (accessed Nov 2008). Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu,
8. Williams, D. L.; Flaherty, T. J.; Jupe, C. L.; Coleman, S. A.; Mar- G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox,
quez, K. A.; Stanton, J. J. J. Chem. Educ. 2007, 84, 1873–1877. D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
9. Liu, R. S. H. J. Chem. Educ. 2002, 79, 183–185. Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong,
10. Patalinghug, W. C.; Chang, M.; Solis, J. J. Chem. Educ. 2007, 84, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, Gaussian Inc.:
1945–1947. Pittsburgh PA, 2003.
11. Shevyakov, S. V.; Li, H.; Muthyala, R.; Asato, A. E.; Croney, J. 18. Foresman, J. B.; Frisch Æ. Exploring Chemistry with Electronic
C.; Jameson, D. M.; Liu, R. S. H. J. Phys. Chem. A 2003, 107, Structure Methods, 2nd ed.; Gaussian, Inc.: Pittsburgh, PA, 1996;
3295–3299. pp 156–157.
12. Dennington, R., II.; Keith, T.; Millam, J.; Eppinnett, K.; Hovell, 19. Zerner M. C. Semiempirical Molecular Orbital Methods. In Re-
W. L.; Gilliland, R. GaussView, Version 3.09; Semichem Inc.: views in Computational Chemistry; Lipkowitz K. B., Boyd D. B.,
Shawnee Mission, KS, 2003. Eds.; VCH Publishing: New York, 1991; Vol. 2, pp 313–366.
13. WebMO, Version 8.0; WebMO LLC: Holland, MI, 2007. 20. Drago, R. S. Physical Methods for Chemists, 2nd ed.; Surfside
14. HyperChemProfessional, 7.51; Hypercube, Inc.: Gainesville, FL, Scientific Publishers: Gainsesville, FL, 1992; pp 120–121.
2007. 21. Atkins, P. W.; de Paula, J. Physical Chemistry, 7th ed.; W. H. Free-
15. Thompson M. A. ArgusLab, 4.0.1; Planaria Software LLC: Se- man and Company: New York, 2002; Chapter 17.
attle, WA, 2007. 22. Billo, E. J. Excel for Chemists, A Comprehensive Guide; Wiley-
16. Spartan ’06; Wavefunction Inc.: Irvine, CA, 2006. VCH: New York, 1997; pp 318–321.
17. Revision A.1; Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuse- 23. Tamura, Z.; Abe, S.; Ito, K.; Maeda, M. Anal. Sci. 1996, 12,
ria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A., Jr.; 927–930.
Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S.
S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Supporting JCE Online Material
Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; http://www.jce.divched.org/Journal/Issues/2009/Mar/abs333.html
Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.;
Abstract and keywords
Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.;
Hratchian, H. P.; Cross, J. B.; Adamo, C.; Jaramillo, J.; Gomp- Full text (PDF)
erts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Links to cited URLs and JCE articles
Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, Color figures (1, 2, 4, and 5) and tables (all)
G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dap-
prich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Supplement
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, MSExcel spreadsheet with the color-prediction macro

© Division of Chemical Education  •  www.JCE.DivCHED.org  •  Vol. 86  No. 3  March 2009  •  Journal of Chemical Education 339

You might also like