Download presentation

Presentation is loading. Please wait.

Published byMireya Marion Modified over 3 years ago

1
Analysis of the Visible Absorption Spectrum of I 2 in Inert Solvents Using a Physical Model Joel Tellinghuisen Department of Chemistry Vanderbilt University Nashville, TN 37235

2
The visible absorption spectrum of I 2 comprises 3 over- lapped electronic transitions —A X, B X, and C X. Excited-state potential curves in absorption region (gas phase, Es relative to first dissociation limit, dashed regions known from discrete spectroscopy). R eX

3
Absorption is ~20% stronger in inert solvents than in gas phase. Also slightly blue-shifted in solution. ( 10)

4
How to extract the components? Phenomenological approach: Assume each component can be described by a Gaussian-type function and least-squares fit to a sum of such functions. Widely used for both gas-phase and solution spectra; only method used for the latter to date. Physical model: Nonlinear LS spectral simulations with adjustable parameters representing unknown potential curves and transition moment functions. Used successfully on gas-phase halogen spectra since 1976,* including recent reanalysis for I 2.# * Br 2 — R.J. LeRoy, et al., JCP 65, 1485 (1976). # J. Tellinghuisen, MSS 2011 and JCP 135, 054301 (2011). Required: Spectra as a function of temperature T.

5
Problems with phenomenological method In tests on known component bands in I 2 (g) and Br 2 (g),* Common 3-parameter forms are statistically inadequate for representing single bands. 4- and 5-parameter extensions better but “flaky;” component resolutions can depend strongly on chosen band form. Worst: Among competing functional forms, the statistically best deconvolution is not the physically best. * R.I. Gray, et al., JPC A 105, 11183 (2001). Results from this approach on I 2 in n-heptane and CCl 4 :* A X band blue-shifted but otherwise little changed in shape and intensity. 20% increase in intensity probably due to enhanced C X band.

6
Illustration of deficiencies in phenomenological approach

7
So, what about the physical model? Clearly the way to go on gas-phase spectra: theoretically sound. reproduces spectra within experimental uncertainty. Use on solution spectra? cannot be considered theoretically complete. will require additional parameters to account for T- and solvent-dependent shifts of spectral peaks. Try and see!

8
Spectra simulated using standard numerical quantum methods as described in previously cited 2011 JCP article on I 2 gas-phase analysis. Spectra treated as purely bound-free — pseudocontinuum model. X potential: Morse curve having e = 212.59(8) cm 1 and e x e = 0.56 cm 1 (values for I 2 in n-heptane), and R e = 2.666 Å (gas). Excited-state potentials: U(z) = A 0 + B 0 exp( a 1 z + a 2 z 2 + …) with z = R R 0 and R 0 = 2.7 Å. [Results — no need for a 3, a 4, …] Better convergence defining adjustable parameters as U(R 0 ) (= A 0 + B 0 ) and (dU/dR) 0 (= a 1 B 0 ). T-dependent shifts: U T = B T exp( bz), with b state- but not T- dependent. Transition moment functions: |µ e (z)| = µ 0 + µ 1 z Additional: Scale parameters for T 23°C (mostly 1.000±0.001) Nonlinear LS fit model

9
Data Recorded 400-850 nm in previously cited 2001 JPCA study; resolution and sampling interval = 1 nm, Shimadzu UV-2101 PC. molar absorptivities, (l mol 1 cm 1 ), estimated from spectra spanning peak absorbances A = 0.3-1.5 as recorded for 3-5 different concentrations at each T. n-heptane: Three data sets for each of 4 Ts: 15.6°, 22.7°, 40.2°, and 50.0°C (Ts from a calibrated thermistor). CCl 4 : Two data sets, same 4 Ts for one, just 22.7° and 50.0° for the other. wavelength corrections: Following paper.

10
Results* Preliminary findings: Transition moment slopes µ 1 least-determined (true also for gas-phase analysis). For B–X, within one SE of gas-phase, so fixed at that value (0.72 D/Å). Pursued two models: (1) µ 1 similarly frozen at gas phase for other two bands, and (2) µ 1 freely fitted for these two. Latter statistically better (by ~10% in 2 ) but yielded questionably strong R-dependence for A–X, and unusual U A. Both statistically better than phenomenological model, by 14- 46% in terms of summed squared residuals, depending on data set and model. But more parameters — 32-34 vs. 27. * JPC A 116, 391 (2012).

11
Component resolution for I 2 in CCl 4, (dashed) compared with I 2 (g) (solid). For solution analysis, µ 1 values were fixed at gas- phase results. Error bars are 1- (too small to see for A X).

12
Corresponding potential curves for I 2 (CCl 4 ) (dashed) and I 2 (g) (solid). Small-R downturns in the dashed curves are considered artifacts of the fitting, as this region and that at large R are not well sampled by the data. Energies relative to X-state minimum. (at 23°C)

13
T-shift parameters for I 2 in CCl 4. Solid curves and points obtained fitting µ 1 for C X and A X; open points and dashed lines for all µ 1 frozen at gas-phase values. [All B T 0 at 22.7°C.]

14
Corresponding results for I 2 in n-heptane. (A state)

15
Component bands for I 2 in CCl 4, as obtained fitting two µ 1 values (solid) and holding them fixed at gas phase values (dashed)

16
A-state potential curves and A–X transition moment functions from the two analyses. Solid: two µ 1 values fitted. Dashed: All µ 1 fixed at gas-phase values. Ground- state potential and classical radial probability distribution shown at bottom.

17
Vibrational components in A X spectra from two different analyses. Top: µ 1 fitted. Bottom: All µ 1 frozen at gas-phase values.

18
Summary The physical model is statistically better than phenomenological and it reduces the model-dependent ambiguities in the band composition analysis for I 2 absorption in inert solvents. However, it does require more parameters; and dependence on different functional forms for potentials has not been thoroughly examined. The previous indication that most of the intensity gain in solution occurs in the C X transition is strongly supported. No obvious need for >3 transitions, nor for >2 dimensions in the potentials. Finding disparate descriptions of the A X transition from different assumptions in the model shows that the physical model is not foolproof!

Similar presentations

Presentation is loading. Please wait....

OK

Chapter 7 Statistical Data Treatment and Evaluation

Chapter 7 Statistical Data Treatment and Evaluation

© 2019 SlidePlayer.com Inc.

All rights reserved.

To make this website work, we log user data and share it with processors. To use this website, you must agree to our Privacy Policy, including cookie policy.

Ads by Google