Electronic Structure of the Nucleobases - American Chemical Society

Institute of Metal Physics, Russian Academy of Sciences Ural DiVision, 620219 Yekaterinburg GSP-170, Russia. ReceiVed: August 16, 2004; In Final Form:...
0 downloads 0 Views 613KB Size
J. Phys. Chem. B 2005, 109, 7749-7757

7749

Electronic Structure of the Nucleobases J. MacNaughton* and A. Moewes Department of Physics and Engineering Physics, UniVersity of Saskatchewan, 116 Science Place, Saskatoon, Saskatchewan S7N 5E2, Canada

E. Z. Kurmaev Institute of Metal Physics, Russian Academy of Sciences Ural DiVision, 620219 Yekaterinburg GSP-170, Russia ReceiVed: August 16, 2004; In Final Form: February 11, 2005

We present a comparison between experimental and calculated soft X-ray spectra of DNA’s nucleobases, adenine (A), guanine (G), cytosine (C), and thymine (T) using X-ray absorption spectroscopy (XAS) and soft X-ray emission spectroscopy (XES). Spectra of the 1s thresholds of carbon, nitrogen, and oxygen give a complete picture of the occupied and unoccupied partial density of states of the nucleobases. A combination of both Hartree-Fock and density functional theory calculations are used in the comparison to experimental results. Most experimental results agree well with our theoretical calculations for the XAS and XES of all bases. All spectral features are assigned. A comparison of the experimental highest occupied molecular orbitallowest unoccupied molecular orbital energy gaps is made to the diverse values predicted in the literature.

1. Introduction The desire to build devices on the nanoscale has led to an interest in molecular electronics research. DNA may be a suitable candidate for nanowires because of its ability to selfassemble, its molecular recognition abilities, and the fact that it is widely available.1 Currently, the main limiting factor seems to be controversy regarding the efficiency of DNA as an electrical conductor. Heavily debated in the literature, contradictory results have been reported indicating insulating,2-5 semiconducting,6-8 highly conductive,9 or even superconducting10 behavior for the molecule. The complexities involved with interactions of DNA and its surrounding environment as well as the large number of atoms involved in the molecular system create significant challenges in designing experiments and the theoretical models required to gain understanding about its electronic structure. The physical structure of the DNA molecule consists of a right-handed helical stack of complementary pairs of nitrogenous bases supported by a sugar (deoxyribose) and phosphate backbone. The four nucleobases are aromatic ring-type structures and are adenine (A), guanine (G), cytosine (C), and thymine (T). The molecular structures of the bases are displayed in Figure 1. The atoms in the structures are numbered for identification purposes in the theoretical results. A systematic approach to understanding the complicated electronic structure of DNA is to first examine the electronic structures of its main components. Several recent theoretical attempts have been made to study electronic properties of isolated nucleobases and homonucleotide base stacks11-17 but do not include predictions of spectroscopic results. While the majority of these studies utilize the density functional theory (DFT) methodology, Hartree-Fock and AM1 techniques were also investigated. The calculated highest occupied molecular orbital (HOMO)-lowest unoccupied molecular orbital (LUMO) * Author to whom correspondence should be addressed. Phone: (306) 966-6380. Fax: (306) 966-6400. E-mail: [email protected].

Figure 1. Molecular structures of the four DNA bases, (a) adenine, (b) guanine, (c) cytosine, and (d) thymine.

gaps for these types of systems is found to vary (see, for example, refs 11 and 13). In this study, the experimental soft X-ray absorption spectra (XAS) and X-ray emission spectra (XES) for all C, N, and O are compared to calculated spectra for the four nucleobases of DNA. Some previous experimental work has been done primarily with energy loss spectroscopy18 and X-ray absorption of the nucleobases.19-21 This is the first comprehensive study including detailed experimental and theoretical results for both absorption and emission spectroscopy for all three edges (C, N, and O). The main transitions in the XAS and XES spectra are identified. Our experimental values of the HOMO-LUMO energy gaps are related to our calculated values as well as compared to the very different values given in the literature.

10.1021/jp0463058 CCC: $30.25 © 2005 American Chemical Society Published on Web 04/02/2005

7750 J. Phys. Chem. B, Vol. 109, No. 16, 2005

MacNaughton et al. TABLE 1: Nucleobase Geometry Optimized by StoBe bond lengtha (Å) CsC CdC CsN large ring CsN small ring CdN large ring CdN small ring CsNH2 CsH (ring) NsH (ring) NsH (amino) CdO CsCH3 CsH (methyl) a

adenine

guanine

cytosine

thymine

1.418 1.405 1.350 1.387 1.351 1.323 1.358 1.094 1.020 1.015

1.442 1.405 1.396 1.386 1.325 1.318 1.369 1.091 1.021 1.014 1.231

1.442 1.365 1.392

1.468 1.360 1.397

1.331 1.365 1.093 1.020 1.016 1.232

1.094 1.022 1.233 1.502 1.103

Bond length is averaged when there are multiple occurrences.

the ionization potentials for the samples. Good agreement with theoretical predictions is another indicator that the measured spectra do not have significant contributions from carbon contamination. 3. Calculations

Figure 2. Experimental and theoretical carbon 1s soft X-ray absorption (XAS) spectra for (a) adenine, (b) guanine, (c) cytosine, and (d) thymine. Experimental spectra are displayed with black lines, and calculated spectra are shown with gray lines. The offset used for the y-scale is 35%.

2. Experimental Section The soft X-ray spectroscopic measurements were performed at beamline 8.0.1 at the Advanced Light Source synchrotron located at the Lawrence Berkeley National Laboratory. The X-ray absorption spectra were measured in total fluorescence yield (TFY) mode for the nucleobases. The resolving power E/∆E for the absorption spectra is about 5000 for the carbon, nitrogen, and oxygen edges. For the fluorescence spectra, the emitted radiation is partially collected in a Rowland circle-type spectrometer with spherical gratings and recorded with an areasensitive multichannel detector. The details of this endstation are described elsewhere.22 Total experimental resolution in the KR X-ray emission region is 0.4 eV full width at half-maximum (FWHM) for carbon, 0.75 eV FWHM for nitrogen, and 1.2 eV FWHM for oxygen. All absorption and emission spectra are normalized to the number of photons falling on the sample, monitored by a highly transparent gold mesh in front of the sample. Samples of the four nucleobases were purchased from Sigma and measured in powder form. Energy calibration was completed by using measurements of reference samples and by shifting the energy using literature values. For these calibrations, highly oriented pyrolytic graphite23 was used for the carbon edge, hexagonal boron nitride (h-BN)24 was used for the nitrogen edge, and TiO2 25 was used for the oxygen edge. The carbon edge is often affected by contamination of the optical beamline components. When the experimental data were obtained, the Au mesh had been refreshed by evaporating a fresh layer of Au on it. To ensure features in the spectra are real and not introduced by the normalization process, the mesh current (Io) was examined. The XAS spectra before normalization contain the same features as the final spectra displayed in Figure 2. The main contribution that normalization to Io has for the spectra displayed is a slight change in slope in the features after

Results from two different calculation programs are used to model the experimental spectra. The program GSCF326 was used for X-ray absorption calculations and involves an ab initio selfconsistent field (SCF) calculation of the core-excited and coreionized states with explicit consideration of the core-hole.27 The improved virtual orbital method involving the relaxed Hartree-Fock potential was used to obtain the core-excited states.28 The transitions to the unoccupied molecular orbitals were determined and correspond especially well to the transitions to the π* orbitals in the experimental XAS spectra for the four nitrogenous bases. A separate calculation was performed for each atom of interest with nonequivalent symmetry, and the results are summed into a final spectrum. Huzinaga-type basis sets29 were used for the calculations, with the contraction schemes (411121/3111) with a polarization function being used for heavy atoms with the core-hole, (621/41) being used for heavy atoms without the core-hole and (41) being used for the hydrogen atoms. Gaussian line widths used in the production of the simulated spectra from the calculated oscillator strengths were 0.7 eV FWHM up to the ionization potential and 4.0 eV FWHM beyond. These line widths were chosen to correspond to the experimentally observed results. The second calculation program used to calculate spectra is StoBe30 and is software designed to analyze electronic structure of molecules, with a focus on inner-shell spectroscopies. This approach uses a linear combination of Gaussian-type orbitals approach to form self-consistent solutions of the Kohn-Sham DFT equations. All calculations using StoBe used the gradientcorrected Becke (BE88)31 exchange functional and the Perdew (PD86)32 correlation functional. The orbital basis set used for the nitrogen, carbon, and oxygen atoms was a triple-ζ valence plus polarization and had the following form (7111/411/1*). Hydrogen atoms were represented by the double-ζ valence plus polarization scheme (311/1). The StoBe program calculated the X-ray absorption spectra using a combination of the transition potential method and a double basis set technique incorporated into density functional theory.33 To more accurately define the core-hole state, the non-core-excited atoms were represented by effective core potentials.34 For better representation of relaxation effects, the atom with the core-hole was described by the IGLO-III basis

Electronic Structure of Nucleobases

J. Phys. Chem. B, Vol. 109, No. 16, 2005 7751

TABLE 2: Assignment of Spectral Features for the C 1s Absorption Data A

1

2

3

4

a

G

transitiona

oscillator strength (× 10-3)

C2 f 7a′′ C2 f 8a′′ C3 f 7a′′ C5 f 7a′′ C1 f 7a′′ C2 f 9a′′ C2 f 26a′ C4 f 7a′′ C4 f 8a′′ C5 f 8a′′ C1 f 8a′′ C2 f 27a′ C3 f 8a′′ C5 f 9a′′ C5 f 26a′ C1 f 9a′′ C1 f 26a′ C2 f 28a′ C2 f 29a′ C2 f 30a′ C3 f 26a′ C3 f 9a′′ C4 f 9a′′ C4 f 26a′ C5 f 27a′

0.611 3.876 9.357 9.526 9.106 1.313 0.062 3.463 6.043 0.847 0.743 0.151 0.003 0.771 0.070 1.203 0.048 0.024 0.003 0.084 1.243 1.117 0.784 0.208 0.179

1

2

3

4

C

transitiona

oscillator strength (× 10-3)

C2 f 8a′′ C2 f 9a′′ C2 f 29a′ C3 f 8a′′ C2 f 30a′ C2 f 10a′′ C3 f 9a′′ C4 f 8a′′ C1 f 8a′′ C2 f 32a′ C2 f 33a′ C3 f 30a′ C3 f 10a′′ C4 f 29a′ C4 f 30a′ C5 f 8a′′ C1 f 9a′′ C1 f 29a′ C3 f 31a′ C3 f 32a′ C4 f 10a′′ C4 f 31a′ C5 f 29a′ C5 f 9a′′

3.046 0.202 0.150 9.119 0.145 1.830 0.020 8.511 11.35 0.050 0.021 0.993 0.794 0.022 0.207 12.37 0.046 0.207 0.194 0.527 1.215 0.127 0.023 0.015

1 2 3 4

5

T

transitiona

oscillator strength (× 10-3)

C1 f 6a′′ C2 f 6a′′ C1 f 22a′ C1 f 23a′ C3 f 6a′′ C1 f 25a′ C1 f 26a′ C1 f 8a′′ C2 f 22a′ C2 f 7a′′ C3 f 7a′′ C4 f 6a′′ C2 f 24a′ C3 f 22a′ C3 f 23a′ C4 f 7a′′

4.996 10.80 1.705 0.160 8.418 1.042 0.318 0.334 2.356 0.026 4.053 11.06 0.327 0.170 0.082 2.798

1 2 3

4

5

transitiona

oscillator strength (× 10-3)

C1 f 7a′′ C2 f 7a′′ C1 f 24a′ C1 f 25a′ C3 f 7a′′ C5 f 24a′ C5 f 8a′′ C1 f 9a′′ C1 f 27a′ C2 f 25a′ C3 f 8a′′ C4 f 7a′′ C5 f 26a′ C5 f 9a′′ C2 f 9a′′ C2 f 28a′ C3 f 24a′ C4 f 8a′′ C5 f 10a′′

4.667 9.358 0.031 0.215 10.01 1.178 0.068 2.092 0.682 0.020 2.772 12.54 3.049 3.550 0.363 0.683 0.206 2.479 1.813

Transitions are from the core level in the labeled atom (labels from Figure 1) to the given unoccupied state.

set.35 The StoBe calculations for the XAS were completed using the Cs symmetry group. This allows the main features of the absorption to be labeled with the specific transitions that are occurring. These theoretical spectra have been broadened in the same way as the GSCF3 results for comparison purposes. All nonresonant X-ray emission spectra for the nucleobases were calculated using the StoBe code. The valence-core level transitions in these calculations are based on the calculated ground-state Kohn-Sham orbitals.36 Similar to the XAS calculations, the StoBe XES calculations were completed using the Cs symmetry group. This allows the main features of the emission to be labeled. The theoretical discrete emission rates for each constituent were Gaussian-broadened with a constant line width of 0.7 eV FWHM for the N and O edges and a line width of 1.0 eV FWHM for the C edge. The broadened results from the individual atoms were summed into a final spectrum for each element that is compared to the experimental results. Information regarding geometry for these planar molecules, optimized by the StoBe program, is found in Table 1. For all cases, the theoretical spectra have been shifted in energy to align with the experimental data. For the XAS data, the calculated spectra were shifted to line up with the first low-energy π* feature. In the case of the XES, the theoretical data were moved to align with the high-energy edge. The minimum has been set to zero for all spectra. The labeling for the XAS and XES features corresponds to the two separate absorption and emission calculations done with the StoBe program. In the absorption calculation, the core-hole effects are included and therefore have a shifting effect on the orbital energies. This results in a slight overlap in the orbital naming scheme between the two techniques. (For example, in the case of the carbon edge for guanine, the orbital 31a′ shows up as an unoccupied state and occupied state depending on if you look at the labeling for the XAS or XES). This is not considered a large concern since the symmetry of the orbitals is still accurate (a′ or a′′), the oscillator

Figure 3. Experimental and theoretical nonresonant carbon KR emission spectra for (a) adenine, (b) guanine, (c) cytosine, and (d) thymine. Experimental spectra are displayed with black lines, and calculated spectra are shown with gray lines.

strengths are precise and comparisons can still be made successfully within the XAS or the XES results. 4. Results and Discussion Figure 2 shows the carbon 1s XAS spectra for two doublering purines, (a) adenine and (b) guanine, and the two single-

7752 J. Phys. Chem. B, Vol. 109, No. 16, 2005

MacNaughton et al.

ring pyrimidines, (c) cytosine and (d) thymine. These experimental results agree with previously published electron energy loss spectra for adenine and thymine.18 Calculations were completed using both the GSCF3 and the StoBe program. The agreement between theoretical and experimental data is good in the region of the lower-energy features that result from the excitation of the carbon 1s core electron to the π* unoccupied orbitals. The broader feature, located at higher energy, is a result of several excitations to σ* unoccupied states. By calculating each carbon site individually using GSCF3, theoretical results indicate that for the relatively simple pyrimidine molecules each sharp peak is a result of a π* feature that corresponds to a specific nonequivalent carbon site. These features in the GSCF3 spectra are numbered C1 through C4 to correspond to the carbon sites labeled in Figure 1. The nonequivalent sites are not as obvious in the spectra of the purines in Figure 2, parts a and b, because of peak overlap. These results and comparisons to resonant inelastic X-ray emission spectra have been shown previously.19 The results from the StoBe calculations allow greater detail in labeling the features corresponding to bound transitions in the spectra. The numbered features in the StoBe spectra are described in detail in Table 2. Each arrow is labeled with the transitions that are occurring in the region (0.3 eV from the arrows energy location. In more common terms, the a′ orbital has σ-like symmetry, and the a′′ orbital has π-like symmetry. Although the spacings of the peaks in the GSCF3 calculations seem to correspond better to experiment than the StoBe results, the details for the transitions that are occurring agree in both theoretical methods. The results in Table 2 for cytosine and thymine confirm the prediction from the GSCF3

Figure 4. Experimental and theoretical nitrogen 1s XAS spectra for (a) adenine, (b) guanine, (c) cytosine, and (d) thymine. Experimental spectra are displayed with black lines, and calculated spectra are shown with gray lines. The offset used for the y-scale is 35%.

TABLE 3: Assignment of Spectral Features for the C Kr Emission Data A

1 2 3

4

5

6

G

transition

oscillator strength

14a′ f 6a′ 15a′ f 7a′ 15a′ f 9a′ 17a′ f 6a′ 18a′ f 7a′ 19a′ f 10a′ 20a′ f 7a′ 21a′ f 7a′ 20a′ f 8a′ 21a′ f 8a′ 22a′ f 10a′ 22a′ f 6a′ 23a′ f 7a′ 23a′ f 8a′ 23a′ f 9a′ 24a′ f 10a′ 25a′ f 10a′ 25a′ f 6a′ 2a′′ f 7a′ 26a′ f 8a′ 26a′ f 9a′ 5a′′ f 6a′ 28a′ f 7a′ 5a′′ f 8a′ 28a′ f 8a′ 5a′′ f 9a′ 6a′′ f 10a′

0.154 0.168 0.181 0.100 0.185 0.192 0.194 0.197 0.144 0.161 0.209 0.417 0.269 0.181 0.325 0.379 0.510 0.408 0.520 0.113 0.830 0.252 0.108 0.358 0.128 0.183 0.321

1

2 3

4

5

6

7 8

C

transition

oscillator strength

15a′ f 7a′ 16a′ f 7a′ 16a′ f 8a′ 17a′ f 10a′ 19a′ f 8a′ 20a′ f 9a′ 22a′ f 11a′ 19a′ f 7a′ 20a′ f 8a′ 21a′ f 9a′ 22a′ f 9a′ 23a′ f 7a′ 24a′ f 7a′ 25a′ f 8a′ 1a′′ f 9a′ 26a′ f 9a′ 26a′ f 10a′ 27a′ f 11a′ 2a′′ f 11a′ 1a′′ f 7a′ 27a′ f 8a′ 27a′ f 9a′ 28a′ f 10a′ 29a′ f 11a′ 3a′′ f 7a′ 3a′′ f 8a′ 29a′ f 8a′ 4a′′ f 9a′ 4a′′ f 10a′ 5a′′ f 11a′ 31a′ f 9a′ 32a′ f 11a′ 7a′′ f 11a′ 7a′′ f 9a′ 7a′′ f 10a′

0.300 0.230 0.152 0.165 0.156 0.155 0.230 0.150 0.172 0.178 0.162 0.295 0.468 0.110 0.156 0.279 0.469 0.460 0.133 0.530 0.234 0.517 0.778 0.142 0.120 0.665 0.624 0.316 0.285 0.296 0.144 0.193 0.429 0.232 0.441

1 2

3

4

5 6

7

T

transition

oscillator strength

12a′ f 6a′ 14a′ f 5a′ 14a′ f 6a′ 14a′ f 7a′ 16a′ f 8a′ 17a′ f 8a′ 19a′ f 6a′ 19a′ f 7a′ 1a′′ f 7a′ 20a′ f 8a′ 21a′ f 8a′ 19a′ f 5a′ 1a′′ f 5a′ 1a′′ f 6a′ 20a′ f 6a′ 21a′ f 6a′ 2a′′ f 6a′ 21a′ f 7a′ 3a′′ f 8a′ 2a′′ f 5a′ 22a′ f 5a′ 3a′′ f 7a′ 3a′′ f 5a′ 23a′ f 8a′ 4a′′ f 8a′ 5a′′ f 8a′ 23a′ f 6a′ 5a′′ f 7a′

0.335 0.196 0.228 0.130 0.129 0.136 0.213 0.131 0.174 0.768 0.603 0.281 0.362 0.291 0.152 0.300 0.535 0.753 0.201 0.171 0.561 0.518 0.381 0.151 0.276 0.450 0.159 0.144

1 2 3 4 5

6

7

transition

oscillator strength

12a′ f 5a′ 13a′ f 6a′ 16a′ f 7a′ 16a′ f 8a′ 15a′ f 5a′ 17a′ f 6a′ 18a′ f 8a′ 17a′ f 5a′ 18a′ f 7a′ 1a′′ f 5a′ 1a′′ f 6a′ 21a′ f 6a′ 21a′ f 7a′ 22a′ f 7a′ 2a′′ f 8a′ 24a′ f 8a′ 2a′′ f 9a′ 24a′ f 9a′ 25a′ f 9a′ 22a′ f 5a′ 2a′′ f 5a′ 23a′ f 6a′ 3a′′ f 6a′ 25a′ f 7a′ 4a′′ f 7a′ 4a′′ f 5a′ 27a′ f 8a′ 6a′′ f 8a′

0.264 0.196 0.100 0.176 0.185 0.190 0.340 0.258 0.421 0.499 0.194 0.105 0.523 0.103 0.173 0.629 0.562 0.757 0.757 0.449 0.208 0.380 0.448 0.165 0.431 0.193 0.145 0.654

Electronic Structure of Nucleobases

J. Phys. Chem. B, Vol. 109, No. 16, 2005 7753

TABLE 4: Assignment of Spectral Features for the N 1s Absorption Data A

1

2

3

4

a

G

transitiona

oscillator strength (× 10-3)

N2 f 7a′′ N4 f 7a′′ N5 f 7a′′ N5 f 8a′′ N1 f 7a′′ N2 f 8a′′ N2 f 26a′ N2 f 9a′′ N4 f 26a′ N4 f 9a′′ N4 f 27a′ N5 f 9a′′ N5 f 27a′ N1 f 8a′′ N1 f 26a′ N2 f 27a′ N3 f 7a′′ N4 f 28a′ N4 f 29a′ N4 f 30a′ N5 f 28a′ N5 f 29a′ N5 f 30a′ N1 f 27a′ N1 f 9a′′ N1 f 28a′ N2 f 29a′ N2 f 30a′ N3 f 26a′ N3 f 9a′′

5.890 4.318 1.689 4.581 1.427 1.423 0.151 1.508 0.017 2.443 0.189 0.211 0.054 0.262 1.371 0.179 3.501 0.054 0.006 0.050 0.017 0.193 0.075 0.321 0.392 4.702 0.047 0.004 3.167 1.043

1 2

3

4

5

C

transitiona

oscillator strength (× 10-3)

N2 f 8a′′ N2 f 9a′′ N2 f 29a′ N4 f 8a′′ N4 f 29a′ N4 f 9a′′ N1 f 29a′ N1 f 8a′′ N2 f 10a′′ N2 f 31a′ N3 f 8a′′ N4 f 10a′′ N4 f 31a′ N4 f 32a′ N5 f 29a′ N1 f 30a′ N1 f 31a′ N3 f 29a′ N3 f 30a′ N5 f 9a′′ N5 f 30a′ N1 f 10a′′ N1 f 32a′ N1 f 33a′ N3 f 32a′ N3 f 33a′ N5 f 32a′ N5 f 10a′′ N5 f 33a′

7.231 0.162 0.074 1.509 0.134 0.887 1.742 3.020 1.813 0.006 4.204 2.514 0.602 0.057 2.124 1.767 0.459 0.114 3.356 0.017 4.636 0.016 0.125 0.317 0.526 0.131 1.064 0.006 0.496

1 2 3 4 5

6

T

transitiona

oscillator strength (× 10-3)

N2 f 6a′′ N2 f 7a′′ N2 f 23a′ N2 f 24a′ N1 f 6a′′ N2 f 25a′ N3 f 6a′′ N1 f 23a′ N1 f 7a′′ N2 f 8a′′ N3 f 7a′′ N3 f 23a′ N1 f 24a′ N1 f 25a′ N3 f 24a′

3.415 0.001 0.049 0.779 1.146 0.572 2.056 2.047 0.801 3.273 1.361 1.547 1.171 3.763 0.994

1 2 3 4 5

transitiona

oscillator strength (× 10-3)

N1 f 7a′′ N2 f 7a′′ N1 f 8a′′ N2 f 8a′′ N1 f 27a′ N2 f 27a′ N1 f 28a′ N2 f 28a′ N1 f 29a′ N1 f 9a′′ N2 f 29a′ N2 f 9a′′

2.317 1.375 0.271 1.630 0.634 2.394 2.421 0.905 1.224 2.039 0.234 1.178

Transitions are from the core level in the labeled atom (labels from Figure 1) to the given unoccupied state.

results that the main transitions occurring in each of the main four prepeaks corresponds to a specific nonequivalent carbon site. The sites that these features are resulting from are in agreement for the two techniques. Nonresonant carbon KR X-ray emission spectra of the bases are displayed in Figure 3. The excitation energy used for measuring these spectra was at 320 eV, well above the 1s carbon absorption edge. The calculations were completed with the StoBe program. Small differences exist in the spectra, resulting from differences in the occupied density of states for the four molecules. Experimental and theoretical agreement is quite good. The carbon emission is one large feature resulting from a multitude of transitions occurring from occupied states with p-symmetry to the open K-shells in the various carbon atoms. The main transitions and oscillator strengths associated with the numbered features in Figure 3 are identified in Table 3 for the carbon edge. Figure 4 displays the nitrogen 1s X-ray absorption spectra for the four bases. For the nitrogen edge, both StoBe and GSCF3 were used to calculate the spectra. Although many of the main features are represented in the theory, some variations in peak intensity and location occur between experimental and calculated results in both theoretical sets of spectra. A possible source of deviation is that the calculations were performed for gas-phase molecules and measurements are for powdered samples. The two spectra with the largest variations seem to be guanine and cytosine, which likely are results of the solid-state structures. The crystal structure of guanine monohydrate (a comparable structure to guanine) forms a layered graphite-type structure hydrogen bonded together with N-H‚‚‚N-type bonds.37 In the case of cytosine, the structure is held together with N-H‚‚‚O

Figure 5. Experimental and theoretical nonresonant nitrogen KR emission spectra for (a) adenine, (b) guanine, (c) cytosine, and (d) thymine. Experimental spectra are displayed with black lines, and calculated spectra are shown with gray lines.

and N-H‚‚‚N bonds.38 In both cases, the nitrogen atoms are heavily involved in the hydrogen bond network in the crystal

7754 J. Phys. Chem. B, Vol. 109, No. 16, 2005

MacNaughton et al.

TABLE 5: Assignment of Spectral Features for the N Kr Emission Data A

1 2

3 4

5

6

7

G

transition

oscillator strength

16a′ f 1a′ 18a′ f 4a′ 18a′ f 5a′ 19a′ f 1a′ 21a′ f 2a′ 21a′ f 3a′ 22a′ f 5a′ 21a′ f 1a′ 23a′ f 2a′ 24a′ f 4a′ 22a′ f 1a′ 24a′ f 2a′ 25a′ f 3a′ 26a′ f 4a′ 2a′′ f 5a′ 1a′′ f 1a′ 2a′′ f 2a′ 3a′′ f 4a′ 3a′′ f 5a′ 4a′′ f 2a′ 27a′ f 3a′ 4a′′ f 3a′ 28a′ f 4a′ 28a′ f 5a′ 4a′′ f 1a′ 29a′ f 3a′ 29a′ f 4a′ 6a′′ f 4a′ 29a′ f 5a′

0.228 0.241 0.563 1.274 1.461 0.203 0.116 0.472 0.636 0.160 0.606 0.424 0.415 0.484 0.515 0.733 0.834 1.145 0.145 0.710 1.719 1.323 0.422 1.593 1.344 0.339 1.791 0.713 1.155

1

2

3

4

5

6 7

C

transition

oscillator strength

19a′ f 2a′ 19a′ f 3a′ 20a′ f 4a′ 22a′ f 5a′ 21a′ f 2a′ 22a′ f 2a′ 21a′ f 3a′ 25a′ f 6a′ 23a′ f 2a′ 23a′ f 3a′ 24a′ f 4a′ 26a′ f 5a′ 26a′ f 6a′ 26a′ f 2a′ 1a′′ f 3a′ 1a′′ f 4a′ 28a′ f 5a′ 3a′′ f 6a′ 3a′′ f 3a′ 30a′ f 5a′ 4a′′ f 6a′ 30a′ f 6a′ 4a′′ f 4a′ 31a′ f 5a′ 31a′ f 6a′ 5a′′ f 2a′ 5a′′ f 3a′ 6a′′ f 4a′ 7a′′ f 6a′

0.197 0.691 0.610 0.145 0.366 1.478 0.472 0.254 0.382 1.281 1.217 0.559 0.674 0.851 0.633 0.415 1.035 0.284 0.457 0.999 1.081 1.667 1.024 1.592 1.017 1.181 1.224 0.638 0.622

1 2

3

4 5 6

T

transition

oscillator strength

14a′ f 2a′ 15a′ f 3a′ 16a′ f 4a′ 16a′ f 2a′ 17a′ f 2a′ 18a′ f 3a′ 19a′ f 4a′ 1a′′ f 4a′ 1a′′ f 2a′ 20a′ f 2a′ 2a′′ f 3a′ 3a′′ f 4a′ 3a′′ f 2a′ 23a′ f 4a′ 4a′′ f 4a′ 4a′′ f 3a′ 4a′′ f 2a′ 24a′ f 2a′ 5a′′ f 2a′

1.030 0.768 0.271 0.574 1.542 1.193 0.725 0.391 0.796 0.464 1.182 0.212 0.392 2.584 0.934 1.835 0.617 0.250 0.687

1

2

3 4 5 6 7

transition

oscillator strength

15a′ f 3a′ 16a′ f 3a′ 16a′ f 4a′ 17a′ f 4a′ 18a′ f 3a′ 19a′ f 3a′ 20a′ f 3a′ 18a′ f 4a′ 19a′ f 4a′ 20a′ f 4a′ 1a′′ f 3a′ 1a′′ f 4a′ 3a′′ f 3a′ 3a′′ f 4a′ 4a′′ f 3a′ 4a′′ f 4a′ 26a′ f 3a′ 5a′′ f 4a′ 6a′′ f 3a′ 27a′ f 4a′

0.882 0.419 0.420 1.302 0.890 0.901 0.702 0.560 0.654 1.177 0.650 0.652 0.552 0.504 0.803 0.159 0.231 1.925 1.125 0.378

TABLE 6: Assignment of Spectral Features for the O 1s Absorption Data cytosine

1 2 3

guanine

transitiona

oscillator strength

O1 f 6a′′ O1 f 7a′′ O1 f 25a′ O1 f 26a′ O1 f 29a′ O1 f 8a′′

0.001519 0.003455 0.000019 0.000043 0.000089 0.000524

1 2 3

thymine

transitiona

oscillator strength

O1 f 8a′′ O1 f 9a′′ O1 f 33a′ O1 f 10a′′ O1 f 34a′

0.004250 0.000026 0.000090 0.000706 0.000043

1 2 3 4 5

a

transitiona

oscillator strength

O1 f 7a′′ O2 f 7a′′ O1 f 8a′′ O2 f 8a′′ O1 f 27a′ O2 f 27a′ O1 f 9a′′ O1 f 30a′ O1 f 10a′′ O2 f 29a′ O2 f 30a′

0.003862 0.002417 0.001169 0.002569 0.000096 0.000077 0.000655 0.000235 0.000075 0.000052 0.000431

Transitions are from the core level in the labeled atom (labels from Figure 1) to the given unoccupied state.

structure and could cause intermolecular effects on the measured spectra. The experimental results agree with previously published results for the nitrogen edge XAS of the bases.20-21 The multiple prepeak features result mainly from transitions to the π* unoccupied levels in the different nitrogen sites. The transitions to the multiple σ* states dominate the broader feature at higher energy. The main transitions corresponding to the regions ((0.3 eV) under the numbered arrows in Figure 4 are labeled along with their oscillator strengths given from the StoBe program and can be found in Table 4. Figure 5 shows the nonresonant N KR X-ray emission spectra for adenine, guanine, cytosine, and thymine with the corresponding theoretical spectra calculated using StoBe. The excitation energy used for this emission data was 420 eV. All features in the experimental data are reasonably seen in the theoretical predictions. Like the carbon emission, the nitrogen XES involves many different transitions from occupied levels occurring in the multiple nitrogen atoms in the molecules. The most prominent

transitions, along with oscillator strengths corresponding to the numbered peaks in Figure 5, are shown in Table 5. The oxygen 1s X-ray absorption data are displayed in Figure 6. This study involves the three bases that contain oxygen, guanine, cytosine, and thymine. Experimental XAS data are compared to theoretical calculations performed with StoBe and GSCF3. The sharper peaks (lower energy) are mainly from the transitions to the π* states, while the broader features (higher energy) are the results of transitions to σ* states. These experimental results are in agreement with previously published experimental results for the nucleobases.21 Guanine and cytosine have one main π* feature due to the single oxygen atom in their respective structures, while thymine has a double peak π* feature, resulting from the excitation of the two oxygen sites in its molecular structure. The main transitions corresponding to the regions ((0.3 eV) under the numbered arrows in Figure 6 are labeled along with their oscillator strengths given from the StoBe program, which can be found in Table 6. For the oxygen,

Electronic Structure of Nucleobases

Figure 6. Experimental and theoretical oxygen 1s XAS spectra for (a) guanine, (b) cytosine, and (c) thymine. Experimental spectra are displayed with black lines, and calculated spectra are shown with gray lines. The offset used for the y-scale is 35%.

StoBe seems to provide better results, particularly in the case of thymine. When the two sites were calculated in GSCF3, they had significant overlap in the first peak, and the defined twopeak structure occurring in the experimental results was not properly modeled. Figure 7 includes the results from nonresonant O KR X-ray emission measurements for guanine, cytosine, and thymine. These emission spectra were measured with the excitation energy set at 560 eV. This emission is more discrete than the carbon and nitrogen edges because there are less sites contributing to the emission process. Table 7 displays the labels for the most prominent transitions occurring in the oxygen emission process, corresponding to the numbered peaks in Figure 7. By comparison of the number of main transitions in a molecule with one oxygen atom (cytosine and guanine) to a molecule with two oxygen atoms (thymine), the number of transitions involved in the emission process has roughly doubled. By combination of XES and XAS experimental data onto a common energy axis, it is possible to determine an energy gap measurement.39-41 This procedure of determining the HOMOLUMO gap is displayed in Figure 8 for the adenine molecule, using the nitrogen data carefully calibrated with known spectral peak locations of h-BN.24 By extension of a line along the

J. Phys. Chem. B, Vol. 109, No. 16, 2005 7755

Figure 7. Experimental and theoretical nonresonant oxygen KR emission spectra for (a) guanine, (b) cytosine, and (c) thymine. Experimental spectra are displayed with black lines, and calculated spectra are shown with gray lines.

TABLE 7: Assignment of Spectral Features for the O Kr Emission Data cytosine

1 2 3 4

transition

oscillator strength

21a′ f 1a′ 22a′ f 1a′ 3a′′ f 1a′ 24a′ f 1a′ 5a′′ f 1a′

0.422982 4.531453 2.117876 6.382281 2.556386

guanine transition 1 27a′ f 1a′ 2 3a′′ f 1a′ 29a′ f 1a′ 3 6a′′ f 1a′ 4 31a′ f 1a′ 32a′ f1a′ 7a′′ f1a′

oscillator strength

thymine transition

oscillator strength

0.572670 1 1a′′ f 1a′ 0.605986 1.623532 21a′′ f 2a′ 0.952401 4.450748 2 22a′ f 1a′ 3.868707 2.328636 23a′ f 2a′ 3.300495 0.759807 3a′′ f 2a′ 1.404307 5.830437 3 4a′′ f 1a′ 1.509710 1.097151 25a′ f 2a′ 0.824462 4 5a′′ f 1a′ 1.925977 26a′ f 1a′ 5.621324 5a′′ f 2a′ 2.369543 26a′ f 2a′ 0.851319 5 27a′ f 1a′ 0.990141 6a′′ f 1a′ 0.979252 27a′ f 2a′ 5.660287 6a′′ f 2a′ 0.792720

spectral slope and identification of the intersection with the slope of the noise floor, the barriers of the gap were determined. The linear extrapolation method was used to remove the “tail”

7756 J. Phys. Chem. B, Vol. 109, No. 16, 2005

MacNaughton et al. associated with it, but it is important that each gap was measured in the same way for each sample, making comparisons possible. Calibrating the energy involves shifting both emission and absorption data according to measurements done on a reference sample. This is an essential step to correct the experimental energy scale but is only as accurate as the known values for the peaks in the reference spectra. To summarize, these both can have a direct influence on the experimental gap measurement and are likely to add a larger error to the (0.2 eV of our method. Other methods of measuring HOMO-LUMO gaps exist (for example, using a combination of photoelectron spectroscopy and electron transmission spectroscopy in the gas phase), but this method as presented gives a reasonable estimation for the solid phase of these molecules. However, the gas-phase measurements may correspond better to some of the theoretical results.

Figure 8. Nitrogen edge XES and XAS spectra for adenine used for demonstration of experimental HOMO-LUMO gap measurement.

TABLE 8: Nucleobase HOMO-LUMO Gap Comparison

e

bases

∆Eexp (eV)

∆Ecalc (eV) (StoBe)

∆Ecalc (eV) (isolated bases)

∆Ecalc (eV) (base stacks)

adenine

4.7

3.71

12.1a 8.6b

3.73c 7.22d

thymine

5.2

3.85

12.7a 9.3b

3.12c 6.91d

guanine

2.6

3.66

12.2a 8.4b 4.8e

3.19c 6.44d 2.97e 2.5f 3.5g

cytosine

3.6

4.02

12.5a 9.3b

3.34c 6.60d

a Reference 13. b Reference 16. c Reference 11. d Reference 17. Reference 12. f Reference 14. g Reference 15.

caused by experimental (Gaussian) and lifetime (Lorentzian) broadening. The down slope of the emission spectrum was used to determine the first intersection (HOMO), and the onset slope of the π* feature in the absorption spectrum was used for the second intersection (LUMO). The energy difference between the two intersections gives the gap value. This is a relatively simple procedure and is easily reproduced; therefore, the error in completing this method is no more than 0.1 eV for each of the boundaries or (0.2 eV for the total experimental gap value. The results for all four bases using this method are displayed in Table 8, where they are compared to results from our StoBe calculations and other various theoretical results for isolated bases and homonucleotide base stacks. The experimental gap measurements differ between the nucleobases as a result of their unique electronic structures. The experimental values for the gap do not correspond directly to any of the theoretical predictions. However, the theoretical predictions do not seem to be in agreement either and differ substantially depending on the method of calculation applied. While it provides an estimate for comparison, this experimental method is not always straightforward to determine an absolute value for the gap since core-hole interactions, and calibrating the energy to a reference sample has effects on the final results. Core-hole effects can cause the onset of the absorption spectrum to shift to lower energy due to the influence of the core-hole on the final state in the absorption process. The presence of the core-hole tends to shift the emission spectrum to lower energy as well but has minimal impact. The linear extrapolation will also have error

5. Conclusions X-ray absorption spectroscopy and X-ray emission spectroscopy are used to study the electronic structures of the four DNA nucleobases, adenine, guanine, cytosine, and thymine and are compared to our calculations. Spectral features are assigned in the XAS and XES spectra according to the Cs symmetry used in the calculations. The experimental and theoretical results for the four nucleobases show each has a unique partial density of states for the carbon, nitrogen, and oxygen sites. For the XAS results, comparisons are made to both GSCF3 and StoBe calculations. Although producing quite similar results, GSCF3 seems better suited to calculating the carbon spectra while StoBe provides improved results for the oxygen edge. Both types of calculations have similar difficulties modeling some features in the nitrogen edge spectra. While there is good agreement between many of the experimental and theoretical spectra, a challenge remains with determining absolute values of the HOMO-LUMO gaps. The origin of the gap challenge is twofold; discrepancies in the values of the gap exist among theoretical results, and obtaining experimental values is complicated due to influences of energy calibration procedures and to a lesser extent due to the presence of core-hole effects. These experimental techniques have proven useful at probing the electronic density of states and in combination with the theoretical results have helped provide a description of the electronic structure of the nucleobases. Acknowledgment. Funding by the Natural Sciences and Engineering Research Council of Canada, the Research Council of the President of the Russian Federation (Grant No. NSH1026.2003.2), and the Saskatchewan Synchrotron Institute is gratefully acknowledged. A.M. is a Canada Research Chair. The work at the Advanced Light Source at the Lawrence Berkeley National Laboratory was supported by the U. S. Department of Energy (Contract No. DE-AC03-76SF00098). References and Notes (1) Robertson, N.; McGowan, C. A. Chem. Soc. ReV. 2003, 32, 96. (2) Braun, E.; Eichen, Y.; Sivan, U.; Ben-Yoseph, G. Nature 1998, 391, 775. (3) de Pablo, P. J.; Moreno-Herrero, F.; Colchero, J.; Go´mez Herrero, J.; Herrero, P.; Baro´, A. M.; Ordejo´n, P.; Soler, J. M.; Artacho, E. Phys. ReV. Lett. 2000, 85, 4992. (4) Storm, A. J.; van Noort, J.; de Vries, S.; Dekker, C. Appl. Phys. Lett. 2001, 79, 3881. (5) Tran, P.; Alavi, B.; Gruner, G. Phys. ReV. Lett. 2000, 85, 1564. (6) Porath, D.; Bezryadin, A.; de Vries, S.; Dekker, C. Nature 2000, 403, 635. (7) Yoo, K.-H.; Ha, D. H.; Lee, J.-O.; Park, J. W.; Kim, J.; Kim, J. J.; Lee, H.-Y.; Kawai, T.; Choi, H. Y. Phys. ReV. Lett. 2001, 87, 198102.

Electronic Structure of Nucleobases (8) Rakitin, A.; Aich, P.; Papadopoulos, C.; Kobzar, Y.; Vedeneev, A. S.; Lee, J. S.; Xu, J. M. Phys. ReV. Lett. 2001, 86, 3670. (9) Fink, H.-W.; Scho¨nenberger, C. Nature 1999, 398, 407. (10) Kasumov, A. Yu.; Kociak, M.; Gue´ron, S.; Reulet, B.; Volkov, V. T.; Klinov, D. V.; Bouchiat, H. Science 2001, 291, 280. (11) Zhang, M.-L.; Miao, M. S.; Van Dogen, V. E.; Ladik, J. J.; Mintmire, J. W. J. Chem. Phys. 1999, 111, 8696. (12) Felice, R. D.; Calzorali, A.; Molinari, E.; Garbezi, A. Phys. ReV. B 2001, 65, 45104. (13) Reha, D.; Kabelac, M.; Ryjacek, F.; Sponer, J.; Sponer, J. E.; Elstner, M.; Suhai, S.; Hobza, P. J. Am. Chem. Soc. 2002, 124, 3366. (14) Calzorali, A.; Felice, R. D.; Molinari, E.; Garbesi, A. Physica E 2002, 13, 1236. (15) Calzolari, A.; Felice, R. D.; Molinari, E.; Garbezi, A. J. Phys. Chem. B 2004, 108, 2509. (16) Honorio, K. M.; Da Silva, A. B. F. Int. J. Quantum Chem. 2003, 95, 126. (17) Bogar, F.; Ladik, J. Chem. Phys. 1998, 237, 273. (18) Isaacson, Michael J. Chem. Phys. 1972, 56, 1813. (19) Moewes, A.; MacNaughton, J.; Wilks, R.; Lee, J. S.; Wettig, S. D.; Kraatz, H.-B.; Kurmaev, E. Z. J. Electron Spectrosc. Relat. Phenom. 2004, 137-140, 817. (20) Kirtley, S. M.; Mullins, O. C.; Chen, J.; van Elp, J.; George, S. J.; Chen, C. T.; O’Halloran, T.; Cramer, S. P. Biochim. Biophys. Acta 1992, 1132, 249. (21) Fujii, K.; Akamatsu, K.; Muramatsu, Y.; Yokoya, A. Nucl. Instrum. Methods Phys. Res., Sect. B 2003, 199, 249. (22) Jia, J. J.; Callcott, T. A.; Yurkas, J.; Ellis, A. W.; Himpsel, F. J.; Samant, M. G.; Sto¨hr, J.; Ederer, D. L.; Carlisle, J. A.; Hudson, E. A.; Terminello, L. J.; Shuh, D. K.; Perera, R. C. C. ReV. Sci. Instrum. 1995, 66, 1394. (23) Carlisle, J. A.; Blankenship, S. R.; Terminello, L. J.; Jia, J. J.; Callcott, T. A.; Ederer, D. L.; Perera, R. C. C.; Himpsel, F. J. J. Electron Spectrosc. Relat. Phenom. 2000, 110, 323. (24) Fomichev, V. A.; Rumsh, M. A. J. Phys. Chem. Solids 1968, 29, 1015.

J. Phys. Chem. B, Vol. 109, No. 16, 2005 7757 (25) Lusvardi, V. S.; Barteau, M. A.; Chen, J. G.; Eng, J., Jr.; Fru¨hberger, B.; Teplyakov, A. Surf. Sci. 1998, 397, 237. (26) Kosugi, N. Theor. Chim. Acta 1987, 72, 149. (27) Kosugi, N.; Kuroda, H. Chem. Phys. Lett. 1980, 74, 490. (28) Hunt, W. J.; Goddard, W. A., III Chem. Phys. Lett. 1969, 3, 414. (29) Huzinaga, S.; Andzelm, J.; Klobukowski, M.; Radzio-Andzelm, E.; Sakai, Y.; Tatewaki, H. Gaussian Basis Sets for Molecular Calculations; Elsevier: Amsterdam, 1984. (30) Hermann, K.; Pettersson, L. G. M.; Casida, M. E.; Daul, C.; Goursot, A.; Koester, A.; Proynov, E.; St-Amant, A.; Salahub, D. R.; Carravetta, V.; Duarte, H.; Godbout, N.; Guan, J.; Jamorski, C.; Leboeuf, M.; Malkin, V.; Malkina, O.; Nyberg, M.; Pedocchi, L.; Sim, F.; Triguero, L.; Vela, A. StoBe-deMon, version 1.0; StoBe Software: Berlin, 2002. (31) Becke, A. D. Phys. ReV A 1988, 38, 3098. (32) Perdew, J. P. Phys. ReV. B 1986, 33, 8822. (33) Triguero, L.; Pettersson, L. G. M.; A° gren, H. Phys. ReV. B 1998, 58, 8097. (34) Kolczewski, C.; Pu¨ttner, R.; Plashkevych, O.; A° gren, H.; Staemmler, V.; Martins, M.; Snell, G.; Schlachter, A. S.; Sant’Anna, M.; Kaindl, G.; Pettersson, L. G. M. J. Chem. Phys. 2001, 115, 6426. (35) Kutzelnigg, W.; Fleischer, U.; Schindler, M. NMR-Basic Principles and Progress; Springer-Verlag: Heidelberg, 1990. (36) Triguero, L.; Pettersson, L. G. M.; A° gren, H. J. Phys. Chem. A 1998, 102, 10599. (37) Thewalt, U.; Bugg, C. E.; Marsh, R. E. Acta Crystallogr., Sect. B 1971, 27, 2358. (38) Barker, D. L.; Marsh, R. E. Acta Crystallogr. 1964, 17, 1581. (39) van Buuren, T.; Dinh, L. N.; Chase, L. L.; Siekhaus, W. J.; Terminello, L. J. Phys. ReV. Lett. 1998, 80, 3803. (40) Luning, J.; Rockenberger, J.; Eisebitt, S.; Rubensson, J.-E.; Karl, A.; Kornowski, A.; Weller, H.; Eberhardt, W. Solid State Commun. 1999, 112, 5. (41) McGuinness, C.; Fu, D.; Downes, J. E.; Smith, K. E. J. Appl. Phys. 2003, 94, 3919.