UPS, XPS, NEXAFS and Computational Investigation of Acrylamide Monomer

Abstract

Acrylamide is a small conjugated organic compound widely used in industrial processes and agriculture, generally in the form of a polymer. It can also be formed from food and tobacco as a result of Maillard reaction from reducing sugars and asparagine during heat treatment. Due to its toxicity and possible carcinogenicity, there is a risk in its release into the environment or human intake. In order to provide molecular and energetic information, we use synchrotron radiation to record the UV and X-ray photoelectron and photoabsorption spectra of acrylamide. The data are rationalized with the support of density functional theory and ab initio calculations, providing precise assignment of the observed features.

1. Introduction

Acrylamide, or prop-2-enamide in IUPAC notation, is a small organic compound widely used in industrial and laboratory processes. The monomer is incorporated into grout and soil-stabilizer products, is used in the synthesis of dyes, and serves as a precursor of a series of homo- and co-polymers whose properties strongly depend on the non-ionic, anionic, or cationic features.

Polyacrylamides are exploited in a wide range of applications [1,2,3,4]. They are used as flocculant or coagulant agents in drinking water and wastewater treatment and in mining processes, as binder and retention supports for fibres and pigments in paper production, for enhanced oil recovery, as drift control agents in commercial herbicide mixtures, as gel supports for chromatography and electrophoresis, in thickeners, in permanent-press fabrics, for contact lenses, in food-packaging adhesives, paper, and paperboard, to wash or peel fruits and vegetables, for water retention in diapers, as stabilizers and binder in cosmetic products, as media for hydroponic crops, as granules in arable lands to reduce the need of irrigation. Although polyacrylamides are quite stable, a small release of the monomer (present as residual of the polymerization process or formed by degradation) in the environment cannot be excluded [4,5]. This constitutes a potential hazard to health since acrylamide is a peripheral nerve toxin to man and to animals and causes birth defects and cancer in animals [6].

Moreover, acrylamide is present in the smoke of cigarettes [7], and it is formed from food components during heat treatment ($T\ge 393$ K, 120 °C) as a result of the Maillard reaction between the aminoacid asparagine (COOH−CHNH₂−CH₂−CONH₂) and reducing sugars [8]. Although it is not clear exactly what risk acrylamide poses to humans, in 2016, the Food and Drug Administration developed a Guidance for Industry (Docket Number: FDA-2013D-0715) that outlines strategies to help growers, manufacturers, and food service operators reduce acrylamide in the food supply. Similarly, in the European Union, the Commission Regulation (EU) 2017/2158 establishes mitigation measures and benchmark levels for the reduction of the presence of acrylamide in food.

Many spectroscopic studies are reported for acrylamide, but to the best of our knowledge, only a few of them concern the molecule in the isolated phase. The studies cover the rotational (20–60.5 GHz [9], 75–480 GHz [10]), far infrared (50–665 cm${}^{-1}$ [11]) and photoelectron (7–27 eV [12]) spectral ranges. The scope of the present work is to investigate the behaviour of acrylamide by means of synchrotron-light techniques and quantum mechanical calculations. Here, we report new data on the inner shell photoionization and photoabsorption processes and also reinvestigate the valence band photoelectron spectrum of acrylamide. This study will enhance our understanding of the fundamental properties of this molecule which is in the spotlight because of its versatility and miscellaneous applications but also because of the global ecological risk posed by it.

2. Materials and Methods Details

2.1. Experimental Methods

Photoionization and photoabsorption spectra of acrylamide were recorded at the Gas Phase beamline [13] at the Elettra synchrotron light source (Trieste, Italy) using an electron energy analyser described previously [14]. Acrylamide purchased from Alfa Aesar (purity 98%) was introduced into the experimental chamber without further purification. The sample appears as white crystalline flakes at room temperature, with melting point $m.p.$ = 357 K and boiling point $b.p.$ = 398 K/25 mmHg. It was sublimated using a custom-built, resistively heated furnace, based on a stainless steel crucible, a Thermocoax^® heating element, and a type K thermocouple. The temperature was increased up to 308 K (35 °C), reaching pressure of about p = 0.256(2) mPa. The vapour pressure estimated at the same temperature by graphical interpolation of the data reported by Carpenter and Davis [15] is 2.25(10) Pa. During the whole experiment, the pressure in the ionization region remained constant and no evidence of contamination or thermal decomposition was found. The photoionization spectra were recorded using a VG-Scienta SES-200 photoelectron analyzer [16] mounted at the magic angle (54.7${}^{°}$). The use of this angle allows us to record spectra in which the intensities of all peaks is proportional to the angle-integrated intensity, that is, angular effects are eliminated. Thus, the intensity of the photoemission bands is proportional to the angle-integrated intensity, and angular effects are eliminated. The photoabsorption spectra were acquired in total ion yield mode, by means of channel electron multiplier placed near the ionization region.

2.2. Computational Methods

Geometry optimizations [17] and subsequent vibrational frequency calculations in the harmonic approximation to determine the nature of the stationary points were run both at the ab initio MP2 and density-functional theory (DFT) B3LYP and B2PLYP levels of calculation using the aug-cc-pVTZ basis set. The time-dependent (TD [18]) DFT approach was used in the case of electronic excited states. The ab initio geometries were used for the estimation of the vertical ionization energies by means of the Symmetry Adapted Cluster/Configuration Interaction method (SAC-CI [19]) with the cc-pVTZ basis set of the Electron Propagator Theory (EPT [20,21]) with the aug-cc-pVTZ basis, by applying the Outer Valence Green’s Function (OVGF [22,23,24]) and the Partial Third Order (P3 [25]) self-energy approximations. The SAC-CI/cc-pVTZ approach was also used for the estimation of the core shell excitation energies. All the above quantum mechanical calculations were performed using the gaussian16™ (Gaussian is a registered trademark of Gaussian, Inc., 340 Quinnipiac St. Bldg. 40, Wallingford, CT, USA) software package (G16, Rev. A.03).

The vibronic structure associated with each ionization in the low energy region of the photoelectron spectrum was simulated including the Franck–Condon activities [26], namely by computing the Franck–Condon factors ($F{C}_k$) for each vibrational mode k with frequency ${\nu }_k$. These were determined by the evaluation of the Huang–Rhys factors $S_k$ [27] as:

$$S_k=\frac{1}{2}·B_k^2$$

(1)

where $B_k$ is the dimensionless displacement parameter defined, assuming the harmonic approximation, and neglecting Duschinsky rotation [28], as:

$$B_k={\left(\frac{2\pi {\nu }_k}{\hslash }\right)}^{1/2}·\left(X_j-X_i\right)·M^{1/2}·Q_k\left(j\right)$$

(2)

where $X_i$ (or $X_j$) is the $3N$ dimensional vector of the equilibrium Cartesian coordinates of the i (or j) electronic state (here i is the neutral and j is the cationic state), M is the $3N\times 3N$ diagonal matrix of atomic masses, and $Q_k\left(j\right)$ is the $3N$ dimensional vector describing the kth normal coordinate of the j (cationic) state in terms of mass weighted Cartesian coordinates. While the approach is less rigorous than that discussed in recent work [29,30,31], it is justified by the minor changes computed for the active vibrational modes upon ionization. For each normal mode k, the Franck–Condon factor $FC$ for a transition from a vibrational level m (of the neutral molecule) to the vibrational level n (of the cation) is [27]:

$$F{C}_k{(m,n)}^2=e^{-S_k}·S_k^{n-m}·\frac{m!}{n!}·{\left[L_m^{n-m}\left(S_k\right)\right]}^2$$

(3)

where L is a Laguerre polynomial. The intensity $I_k(m,n)$ of the m to n transition for the normal mode k is given by the $FC$ factor, weighted for the population of the vibrational state m:

$$I_k(m,n)=F{C}_k{(m,n)}^2·\frac{1}{Z}·exp\left(-\frac{m\hslash {\omega }_k}{k_{B}T}\right)$$

(4)

with $k_B$ being the Boltzmann constant, T is the temperature, and Z is the partition function. The total intensity of the multimode vibrational transition, including all the active normal modes, is the simple product of the monodimensional intensities [27]. Photoelectron spectra were simulated at T = 0 K and a Gaussian broadening function with $FWHM$ = 80 and 440 cm${}^{-1}$ was convoluted with each computed intensity.

3. Results and Discussion

3.1. Conformational Analysis

Acrylamide is the smallest conjugated amide. It is formed by the vynil and amide planar frames connected by a rotatable bond: CH₂=CH−CONH₂. Its conformation is usually described by the relative orientation of the carbon–carbon double bond and the carbonyl, through the CCCO torsional dihedral angle ($\tau$). We explored the conformational space of acrylamide by running a series of quantum mechanical calculations where $\tau$ was constrained to fixed values ($\tau$ = 0–360${}^{°}$, $\Delta \tau$ = 10${}^{°}$), whereas all the other parameters were freely optimized. The results achieved at three levels of calculation (B3LYP/aug-cc-pVTZ, B2PLYP/aug-cc-pVTZ and MP2/aug-cc-pVTZ) are qualitatively similar, as can be seen in Figure 1.

The potential energy surface (PES) of acrylamide in its neutral ground state ($S_0$) evidences the presence of two non-equivalent minima, see Figure 2. The global minimum is planar with the double bonds in $syn$ orientation ($\tau =0^{°}$). The other local minimum has an $anti$ arrangement, which is typical of amides in peptides and protein, and lies at 5.6/5.2/4.0 kJ mol${}^{-1}$, while the $syn$ to $anti$ barrier is about 19/19/18 kJ mol${}^{-1}$ (B3LYP/B2PLYP/MP2).

Taking into account the vibrational zero-point energy correction, this energy difference increases, becoming 6.2/6.2/5.6 kJ mol${}^{-1}$ (B3LYP/B2PLYP/MP2). Because of the steric hindrance between the CH${}_2$ and NH${}_2$ groups, the $anti$ geometry is slightly non planar as indicated by the presence of two equivalent minima at $\tau \simeq \pm {156}^{°}$ in the curve. Since the interconversion barrier between the equivalent conformations is quite low (0.75/1.08/1.61 kJ mol${}^{-1}$ (B3LYP/B2PLYP/MP2) at $\tau$ = 180${}^{°}$) due to tunnel effect, the ground vibrational state is expected to be observed in the form of a doublet. These results are in agreement with the rotational spectroscopy data of Marstokk et al. [9] and Kolesniková et al. [10]; indeed, the authors (i) assigned the transition lines of both the $syn$ and $anti$ conformers, (ii) estimated the conformers energy difference to be 6.5(6) kJ mol${}^{-1}$ [9], and (iii) determined the doubling splitting energy of the anti conformer as 13.844586(2) cm${}^{-1}$ [10]. In the crystal phase only, the $syn$ species has been identified [32]. According to the Boltzmann distribution, and taking into account the effect of the degeneracy of $anti$ species, the relative population of the two conformers at the experimental working temperature (T = 308 K) is $N_{anti}/N_{syn}$ = 0.16(4).

We used the same approaches to calculate the PES of the acrylamide cation in its ground state ($D_0$), where the single occupied molecular orbital ($MO$) is a $\sigma$-type non bonding orbital centered on the oxygen (${\sigma }_n$). The DFT results plotted in Figure 1 show that, for the cation, the $syn$ and $anti$ arrangements correspond to maxima while the minima correspond to two non-planar equivalent species with $\tau$ = 64/67${}^{°}$ (B3LYP/B2PLYP). Unfortunately, we did not succeed in determining the whole path also at the MP2/aug-cc-pVTZ level, but we successfully optimized the geometries of the minimum and the $syn$ arrangement (red squares in Figure 1). As in the case of DFT, the planar $syn$ structure displays an imaginary frequency, albeit considerably smaller (i120 cm${}^{-1}$) than that determined at the B3LYP level (i436 cm${}^{-1}$), indicating a much shallower PES along the torsion axis. Thus, although the ab initio result predicts a non-planar minimum similar to DFT, such a minimum corresponds to a remarkably decreased torsion angle ($\tau$ = 34${}^{°}$) accompanied by a drastically reduced stabilization compared to DFT. More precisely, DFT finds the $syn$ form 24/18 kJ mol${}^{-1}$ (B3LYP/B2PLYP) above the minimum, whereas the ab initio energy difference is only 3.8 kJ mol${}^{-1}$. The DFT and ab initio landscapes are remarkably different, with the small energy difference predicted by MP2 calculations clearly suggesting a reduced relevance of twisting. This is further demonstrated by the almost unchanged bond lengths of the planar and twisted structures (

Table 1. Energy and structure parameters of acrylamide calculated using the aug-cc-pVTZ basis set.

		$\mathit{\tau }$/${}^{°}$	${\mathit{E}}_{\mathit{e}}$/a.u.	$\mathbf{\Delta }{\mathit{E}}_{\mathit{e}}$/eV	C=C/Å	C-C/Å	C-N/Å	C=O/Å	CCC/${}^{°}$	CCO/${}^{°}$	CCN/${}^{°}$
B3LYP	$S_0$$syn$	0	−247.400743	0	1.327	1.493	1.366	1.219	121.1	114.5	123.4
	$S_0$$anti$	157	−247.398607	0.06	1.328	1.491	1.367	1.220	126.2	117.2	120.6
	$D_0$	64	−247.058750	9.31	1.350	1.487	1.307	1.246	120.5	126.3	106.1
	$D_0$$syn$ (TS)	0	−247.049500	9.56	1.335	1.465	1.311	1.280	123.7	122.2	118.1
TD-B3LYP	$D_1$$syn$	0	−247.045674	9.66	1.403	1.508	1.329	1.22	118.3	116.8	118.0
B2PLYP	$S_0$$syn$	0	−247.204462	0	1.330	1.491	1.365	1.221	120.7	114.5	123.3
	$S_0$$anti$	156	−247.202496	0.05	1.332	1.489	1.368	1.222	125.7	116.9	120.8
	$D_0$	67	−246.857625	9.44	1.356	1.486	1.308	1.240	119.7	127.4	103.5
	$D_0$$syn$ (TS)	0	−246.850666	9.63	1.337	1.459	1.309	1.287	122.8	118.4	122.6
MP2	$S_0$$syn$	0	−246.852848	0	1.334	1.490	1.364	1.224	120.3	114.3	123.3
	$S_0$$anti$	154	−246.851325	0.04	1.337	1.487	1.369	1.225	124.9	116.9	121.1
	$D_0$	34	−246.485231	10.00	1.343	1.457	1.304	1.286	120.1	124.7	114.3
	$D_0$$syn$ (TS)	0	−246.483791	10.04	1.326	1.453	1.305	1.297	121.3	118.8	123.1

), in contrast to the remarkable C=O bond length shortening occurring in the twisted DFT structures. Furthermore, the small energy difference predicted between the planar and twisted structures at the ab initio level is within the error of the method. As will be shown below, the simulation of vibronic structure strongly supports a planar or negligibly twisted $D_0$ structure.

Finally, we used TD-B3LYP/aug-cc-pTZVP to calculate the structure of the first excited state of the acrylamide cation ($D_1$), where the singly occupied $MO$ is a $\pi$-type orbital composed of the non bonding out-of-plane p orbitals of nitrogen and oxygen (${\pi }_n$). The optimized geometry is found to be planar as in the case of the neutral species. The energy and structure parameters related to the calculated stationary points are summarized in Table 1.

3.2. UPS

The photoionization spectrum of acrylamide has been recorded in the 6.5–26.5 eV spectral range using a 30.4 nm He(II) source by [12]. Here, we extend the investigation up to 37 eV, using 98 eV energy photons. The measured spectrum is shown in Figure 3.

According to the Koopmans’ theorem [34], a rough correlation can be found between the observed ionization energies and the calculated energy of the occupied $MOs$ of the most stable conformer, whose shapes and energies are given in Figure 4 and in

Table 2. Theoretical molecular orbital energy ($MOE$, eV), vertical binding energy ($BE$, eV) and pole strength ($PS$) values of $syn$-acrylamide compared to the experimental peaks’ ionization energies.

		HF a	B3LYP b	B2PLYP c	SAC-CI d		P3 e		OVGF f		Exp.
		$\mathit{MOE}$	$\mathit{MOE}$	$\mathit{MOE}$	$\mathit{BE}$	$\mathit{PS}$	$\mathit{BE}$	$\mathit{PS}$	$\mathit{BE}$	$\mathit{PS}$	$\mathit{BE}$
1	O(1s)	−558.27	−519.67	−535.80	537.99	0.79	-	-	-	-	537.22
2	N(1s)	−424.19	−390.48	−404.55	406.77	0.78	-	-	-	-	405.97
3	C1(1s)	−309.01	−279.87	−292.06	294.88	0.77	-	-	-	-	294.16
4	C3(1s)	−306.24	−277.45	−289.42	291.79	0.75	-	-	-	-	291.03
5	C2(1s)	−306.07	−277.36	−289.38	291.60	0.76	-	-	-	-	291.03
6	$\sigma$	−37.47	−28.36	−32.14	31.73	0.17	-	-	-	-	32.6
7	$\sigma$	−33.05	−24.90	−28.30	27.26	0.14	-	-	-	-	28.3
8	$\sigma$	−29.05	−21.57	−24.71	24.28	0.63	-	-	-	-	24.3
9	$\sigma$	−23.91	−17.81	−20.37	20.55	0.83	-	-	-	-	20.6
10	$\sigma$	−21.01	−15.69	−17.92	18.31	0.87	18.98	0.86	19.11	0.87	18.6
11	$\sigma$	−19.61	−14.65	−16.74	17.37	0.89	17.95	0.87	18.04	0.88	17.6
12	$\sigma$	−18.92	−13.94	−16.04	16.41	0.87	17.07	0.86	17.18	0.87	16.4/17.0 g
13	$\sigma$	−16.74	−11.84	−13.85	14.13	0.90	14.82	0.88	15.10	0.89	15.0
14	$\sigma$	−16.27	−11.76	−13.69	14.18	0.90	14.86	0.89	14.77	0.90	14.63
15	$\pi$	−15.69	−11.53	−13.28	13.87	0.89	14.33	0.86	14.53	0.87	14.07
16	$\sigma$	−14.75	−10.44	−12.23	12.76	0.91	13.48	0.89	13.58	0.90	13.17
17	$\pi$	−10.60	−7.95	−8.99	10.29	0.93	10.59	0.90	10.53	0.90	10.675
18	${\pi }_n$	−11.35	−7.69	−9.25	9.71	0.91	10.33	0.88	10.44	0.89	10.296
19	${\sigma }_n$	−11.69	−7.28	−9.09	9.19	0.90	9.86	0.88	10.43	0.89	9.756

^a MP2/aug-cc-pVTZ//HF/aug-cc-pVTZ. ^b B3LYP/aug-cc-pVTZ. ^c B2PLYP/aug-cc-pVTZ. ^d MP2/aug-ccpVTZ//SAC-CI/cc-pVTZ. ^e MP2/aug-cc-pVTZ//P3/aug-cc-pVTZ. ^f MP2/aug-cc-pVTZ//OVGF/aug-cc-pVTZ. ^g Shoulder.

, respectively.

We observe that, concerning the outer valence band, different methods provide different ordering of the $MOs$. However, using more sophisticated calculations, it is possible to obtain more reliable predictions. The vertical ionization energies obtained using the SAC-CI and EPT (P3 and OVGF) methods are compared to the $MOs$ energy values in Table 2 and are used for the assignment of the observed peaks, whose energy values are given in the same table. Notably, all these methods find that the lower ionization energy is associated with the ${\sigma }_n$ orbital (n. 19), followed by the ionization of ${\pi }_n$ (n. 18) and ${\pi }_{\mathrm{C}=\mathrm{C}}$ (n. 17). The assignment of Åsbrink et al. [12] follows the same order, involving the features lying at 10.0, 10.3 and 10.7 eV. Since the first peak observed in our ionization spectrum has a lower energy (9.8 eV) and the spectral region between 9.5 and 11.5 eV appears quite crowded, we decided to record the spectrum at higher resolution. The resulting spectrum is shown in Figure 5 with the theoretical vertical ionization energies related to the three outer valence orbitals.

The lower energy peak (labelled as A in Figure 5) lies at 9.756 eV. It has two shoulders (B and C) likely due to vibrational contributions: indeed, the energy differences are B-A = 27 meV and C-A = 52 meV, which are consistent with low frequency normal modes. Two strong peaks (D and E) are found at 9.943 and 10.116 eV. For the hypothesis of a vibrational progression starting from peak A, the involved quanta will be 187 meV (ca. 1508 cm${}^{-1}$) and 173 meV (ca. 1395 cm${}^{-1}$). A similar trend was also observed in the spectrum of formamide [35], and it was tentatively assigned to the asymmetric stretching of the N-C=O frame, an interpretation supported by recently reported vibronic structure simulations [36]. According to this interpretation, we observe that the shape of peak “D” is akin to that of “A”.

To support this analysis, we modelled the vibronic structure of the $D_0\leftarrow S_0$ and $D_1\leftarrow S_0$ transitions using MP2/aug-cc-pVTZ and TD-B3LYP/aug-cc-pVTZ calculated geometries and vibrational normal coordinates to evaluate the Huang–Rhys factors governing vibronic progressions. As discussed above, the MP2 results for the $D_0$ state suggest a limited stabilization of the non-planar structure with a modest twisting and almost unchanged bond lengths. Thus, we simulated the vibronic structure taking the planar $D_0$ structure as a reference. The vibronically active vibrational frequencies and corresponding Huang–Rhys factors are listed in

Table 3. Calculated wavenumbers ${\tilde{\nu }}_k$ and Huang–Rhys factors $S_k$ for the vibrational normal modes of the $D_0\leftarrow S_0$ (MP2/aug-cc-pVTZ) and $D_1\leftarrow S_0$ (TD-B3LYP/aug-cc-pVTZ) transitions of $syn$-acrylamide.

${\mathit{D}}_0\leftarrow {\mathit{S}}_0$			${\mathit{D}}_1\leftarrow {\mathit{S}}_0$
${\tilde{\mathit{\nu }}}_{\mathit{k}}$/cm${}^{-1}$	${\tilde{\mathit{\nu }}}_{\mathit{k}}$/meV	${\mathit{S}}_{\mathit{k}}$	${\tilde{\mathit{\nu }}}_{\mathit{k}}$/cm${}^{-1}$	${\tilde{\mathit{\nu }}}_{\mathit{k}}$/meV	${\mathit{S}}_{\mathit{k}}$
260	32	0.272	183	23	0.018
456	57	0.326	235	29	0.004
550	68	0.083	305	38	0.854
873	108	0.106	467	58	0.003
1061	132	0.238	648	80	0.156
1078	134	0.038	835	104	0.088
1342	166	0.011	838	104	0.001
1444	179	1.011	1063	132	0.001
1481	184	0.028	1134	141	0.002
1584	196	0.247	1297	161	0.134
1689	209	0.046	1388	172	0.410
2200	273	0.003	1492	185	0.081

, while the calculated spectra are compared to the experimental ones in Figure 5 and Figure 6.

The simulated vibronic profiles of both the $D_0\leftarrow S_0$ (blue trace in Figure 5) and $D_1\leftarrow S_0$ (pink trace in Figure 5) transitions show a resolved structure with major vibronic peaks separated by about 200 meV, which is in agreement with the experimental observation. However, the shape and relative intensity of the vibronic bands are different for the two ionizations and comparison with the detected spectrum suggests that the two lower energy peaks must be assigned to the $D_0\leftarrow S_0$ transition. By aligning the calculated and experimental spectra, we determine the adiabatic ionization energies to the $D_0$ state as $\Delta E_{0,0}$ = 9.755 eV. This value is 0.475 eV smaller than that of formamide (10.23(6) eV [36]) and 0.021 eV greater than that of acetamide (9.734(8) [36]). According to the Huang–Rhys factors (Figure 7), the main vibrational structure is related to the skeletal stretching motions associated with the largest geometry changes upon ionization, specifically ${\nu }_{\mathrm{C}-\mathrm{C}}$ = 179 meV (1444 cm${}^{-1}$), ${\nu }_{\mathrm{C}-\mathrm{N}}$ = 196 meV (1584 cm${}^{-1}$) and ${\nu }_{\mathrm{C}=\mathrm{O}}$ = 132 meV (1061 cm${}^{-1}$). Notably, the C=O stretching displays a considerably reduced frequency as a result of the remarkable bond length elongation in the cation (compare 1.224 Å in $S_0$ with 1.29 Å in $D_0$, see Table 1). The “B” and “C” features of the lower energy peak can be associated with the ${\delta }_{\mathrm{C}\mathrm{C}\mathrm{C}}$ = 32 meV (260 cm${}^{-1}$) and ${\delta }_{\mathrm{C}\mathrm{C}\mathrm{N}}$ = 57 meV (456 cm${}^{-1}$) bending motions, respectively.

The positioning of the $D_1\leftarrow S_0$ transition is more challenging. In Figure 5, we propose a tentative assignment with an adiabatic ionization energy $\Delta E_{0,0}$ = 10.31 eV, corresponding to peak H. This value is 0.033 eV smaller than that of formamide (10.64(2) eV [36]) and very close to that of acetamide (10.28(2) [36]). According to the Huang–Rhys factors (Figure 7), the most active normal modes are ${\delta }_{\mathrm{C}\mathrm{C}\mathrm{C}}$ = 38 meV (306 cm${}^{-1}$), ${\nu }_{\mathrm{C}-\mathrm{N}}$ = 172 meV (1388 cm${}^{-1}$), ${\delta }_{\mathrm{C}\mathrm{C}\mathrm{O}}$ = 80 meV (648 cm${}^{-1}$), and ${\nu }_{\mathrm{C}=\mathrm{C}}$ = 161 meV (1297 cm${}^{-1}$). Notably, different vibration modes are active in the ionization to $D_1$ compared to $D_0$, on account of the different geometry changes documented in Table 1 for the $D_1\leftarrow S_0$ and $D_0\leftarrow S_0$ transitions.

Finally, the peaks “I” and “L” can be reasonably assigned to the resolved ionization band of the C=C $\pi$-type orbital, their energy difference being 176 meV, a value very similar to that of the previously considered transitions.

3.3. XPS

The inner shell photoemission spectra of acrylamide are shown in Figure 8, Figure 9 and Figure 10, with the values predicted at the SAC-CI/cc-pVTZ level of calculation for $syn$-acrylamide.

The measured and theoretical ionization are listed in Table 2. The predictions are overestimated by about 0.7–0.8 eV with respect to the observations. As regards the amide frame, the observed binding energies are smaller (O(1s) −0.5 eV, N(1s) −0.5 eV, C1(1s) −0.3 eV) than those of formamide [38] suggesting an overall electron donating effect of the vinylic $\pi$ electron cloud. Differently, the observed acrylamide binding energies are greater (O(1s) +0.2 eV, N(1s) +0.2 eV, C1(1s) +0.4 eV) than those of N-methylacetamide [39]. In addition, in the case of 3-carbamoyl-2,2,5,5-tetramethyl-3-pyrrolin-1-oxyl (C₉H₁₅N₂O₂) that for our purposes can be described as an acrylamide frame embedded in cyclic nitroxyl radical through the C=C bond, the binding energies of acrylamide are blue shifted: O(1s) +0.32 eV, N(1s) +0.07 eV, C1(1s) +0.16 eV and C2/C3(1s) +0.33 [40]. Comparison with pyridin-2-one, where the amide frame is embedded in a conjugate ring system, shows that the O(1s) binding energy of acrylamide is +0.6 eV greater, whereas the N(1s) binding energies is −0.4 eV smaller [41].

3.4. NEXAFS

The observed K-edge photoabsorption spectra of acrylamide are given in Figure 11, Figure 12 and Figure 13.

The photoabsorption energy values of the main peaks are listed in

Table 4. Energies (E/eV), term values (T/eV) and assignment for the main features observed in K-shell spectra of acrylamide and analogous compounds. Calculated (SAC-CI/cc-pVTZ) energy values, term values, oscillator strengths (f) and shape of selected unoccupied $MOs$ of $syn$-acrylamide are also given.

	Acrylamide		Formamide	Acrolein	C9H15N2O2
	${\mathit{E}}_{\mathit{o}}/{\mathit{T}}_{\mathit{o}}$	${\mathit{E}}_{\mathit{c}}/{\mathit{T}}_{\mathit{c}}$ (${\mathit{f}}_{\mathit{c}}$, ${10}^{-2}$)	${\mathit{E}}_{\mathit{o}}/{\mathit{T}}_{\mathit{o}}$ [45]	${\mathit{E}}_{\mathit{o}}$ [46]	${\mathit{E}}_{\mathit{o}}$ [48]	Assignment
C1 $BE$	294.16	294.88	294.5	-	294.0
C2; C3 $BE$	291.03	291.79; 291.60			290.7
C (a)	287.85/6.31	288.40/6.48 (6.8)	288.1/6.4	286.10		C1 $1s\to 1\pi$*
C (b)	284.52/7.27	285.18/5.85 (6.0)		284.19		C3 $1s\to 1\pi$*
C (c)	284.27/7.33	284.97/6.06 (3.0)		284.19		C2 $1s\to 1\pi$*
N $BE$	405.97	406.77	406.5		405.9
N (a)	401.00/4.97	402.83/3.94 (1.0)				N $1s\to 1\sigma$*
N (b)	401.70/4.27	402.59/4.18 (1.5)	401.9/4.6		401.95/3.95	N $1s\to 1\pi$*
N (c)	402.87/3.10	404.01/2.76 (3.1)			403.15/2.75	N $1s\to 2\sigma$*
O $BE$	537.22	537.99	537.7	-	536.9
O (a)	531.14/6.08	532.30/5.69 (2.9)	531.5/6.2	530.59	531.3/5.6	O $1s\to 1\pi$*
O (b)	534.06/3.16	537.04/0.95 (0.6)		533.57	534.3/3.0	O $1s\to 2\pi$*
1$\pi$*		$2\pi$*	$1\sigma$*		$2\sigma$*

with the term values calculated as difference with the ionization energy:

$$T=BE-E$$

(5)

In the same table, we report the assigned features of related molecular systems formamide (NH₂−CHO [45]), acrolein (CH₂=CH−CHO [46]) and 3-carbamoyl-2,2,5,5-tetramethyl-3-pyrrolin-1-oxyl (C₉H₁₅N₂O₂ [47,48]), as well as the results of the theoretical simulation.

Most of the observed peaks can be assigned to transitions from the $1s$ atomic orbital to the first unoccupied $\pi$* orbital. The predictions are overestimated by about 0.6–1.2 eV with respect to the observations the shifts increasing in going from C to O. The $1s\to \pi$* absorption energies of acrylamide are smaller than formamide and larger than acrolein. As regards the other features, by comparison with acrolein, the observed peak O(b) can be assigned to the $1s\to 2\pi$* transition. We note that in this case the calculated value is considerably overestimated. Besides the $1s\to \pi$* peak, N(b), the nitrogen NEXAFS spectrum shows two other significant signals: a shoulder at lower energy, N(a), and an intense peak at higher energy, N(c). The same spectral profile has been observed in 3-carbamoyl-2,2,5,5-tetramethyl-3-pyrrolin-1-oxyl [47,48], although the shoulder is clearly overlapped with the vibrational resolved band of N₂, used as calibrant. Actually, in the spectrum of acrylamide, the N(a) signal does not show any additional vibrational structure. Thus, we infer that it pertains to acrylamide itself. Accordingly, calculations predict a lower energy, less intense $1s\to 1\sigma$* transition that can be related to N(a) and a higher energy more intense $1s\to 2\sigma$* transition that can be related to N(c). Despite comparison with other analogous molecular systems (see Table 4) support the assignment of peak N(b) to the $1s\to 1\pi$* transition, we observe that, based exclusively on the predicted energy values, the assignment of peaks N(a) and N(b) should be inverted. However, the predicted absorption intensities are in good agreement with the first proposed assignment.

4. Conclusions

We have performed an extensive core and valence level investigation of acrylamide and reported the theoretical and experimental photoelectron and photoabsorption spectra, as well as calculated along the C-C torsion and structural data of the ground and electronic excited states. Concerning the valence photoelectron spectra, most vibronic peaks have been successfully assigned on the basis of quantum-chemically computed geometry changes associated with ionization of the highest energy occupied ${\sigma }_n$ orbital and the following ${\pi }_n$-orbital. We find that ab initio and DFT results predict rather different equilibrium structures for the $D_0$ state, with the former level of theory indicating an almost negligible energy difference between planar and twisted structures, in contrast with DFT. Interestingly, the simulated vibronic structure based on the ab initio results agrees very closely with the observed spectra and is in line with recently reported vibronic structure simulations for the smaller formamide.

As regards the core photoelectron spectra, SAC-CI/cc-pVTZ calculations provide predictions slightly overestimated (about 2%) with respect to the experimental values. It is worth noting that the values of the inner shell orbitals calculated at the B2PLYP/aug-cc-pVTZ are of the same order of magnitude of the corresponding binding energies. SAC-CI/cc-pVTZ calculations were also useful to assign the core photoabsorption spectra. Thus, we find generally good agreement between theory and experiment, with few discrepancies that will be the subject of future investigations. More specifically, further investigation of the N(1s) absorption spectrum is warranted, and Resonant Auger–Meitner or angular resolved photoemission spectroscopy experiments could provide additional information for a refined assignment of the outer valence UPS spectrum.