Role of Ethanolamine on the Stability of a Sol-Gel ZnO Ink

: This work presents a detailed structural and chemical characterization of the system formed by zinc acetate dihydrate (ZAD) and ethanolamine (EA) with methoxyethanol (ME), in order to describe its stability. The origin of the mixture degradation during storage at room conditions is analyzed. Theoretical calculations of the frontier orbitals of the system ZAD plus EA under interaction with ME and CO 2 are used to deduce energy levels and stability of the different molecules appearing in the ZAD–EA–ME system. The models were tested as potential energy minimum and their photo-absorption spectra were simulated. The dimerization process leading from the simplest configuration to the most abundant one is also theoretically analyzed and used to describe the effect of the nuclearity raise on the mixture stability. Our results explain the experimental observations and provide a better understanding of the role played by EA in the formation of ZnO and, consequently, allow optimizing the technological processes to prepare these films.


INTRODUCTION
An optimal sol-gel precursor design is a key point for the emerging flexible and printed electronics development. Nowadays, sol-gel-derived ZnO is one of the most attractive oxides, due to its high electrical conductivity 1 , good ultraviolet absorption behaviour, strong room-temperature luminescence 2 , piezoelectricity, excellent environmental stability, chemical sensing 3 , compatibility with large-scale processing, low cost, abundance, biocompatibility and easy fabrication. In most reports published so far 4 , it is obtained by using zinc acetate dihydrate (ZAD) as salt and ethanolamine (EA) as stabilizer. The increasing interest in the development of sol-gel ZnO inks is reflected by recent specific studies on the effect of solvents 5,6 , precursor nature and concentration 7,8 , EA:ZAD molar ratios 9,10 , sol ageing 11 , annealing duration 12,13 , annealing temperature 14,15 , and even specific process parameters such as withdrawal speed in deep coating 16 on the properties of the resulting ZnO.
Stabilization of the Zn(II) ion is crucial in sol-gel processes to obtain high quality ZnO films.
Most of the inks reported to date involve the use of aminoalcohols as stabilizers. However, rigid aminoalcohols such as H 2 N-C 6 H 4 -2R with R = OH or CH 2 OH are not adequate 17 because of the fast degradation of the formed precursors even at room temperature, which precludes their use for ZnO preparation. Indeed, it is known that aminoalcohols present high photosensitivity: radicals based on CH 3 -OCH 2 are formed after UV-irradiation 18 and photooxidative cyclization of aminoalcohols produce 1,3-oxazines 19 . In consequence, the photosensitivity observed in the ZnO precursors containing aminoalcohols is a major issue in order to avoid degradation during ink storage.
EA is by far the preferred aminoalcohol to stabilize the Zn(II), but it does not escape from the previous observation. It is known that EA facilitates Zn(II) chelation and promotes the formation of ZnO due to the amine functional group that increases the solution pH 20 .
Consequently, a higher amount of EA leads to higher chelation rates of Zn 2+ ions along the (002) minimum-surface-energy plane, resulting in higher c-axis orientation with less porosity 10 .
In particular, processes such as hydrolysis, polymerization and complexation have been identified to notably influence the structure and rate of the crystallization 21 . Nevertheless, the presence of ethanolamine on the top of ZnO thin films significantly decreases the optical performance of ZnO devices 20 and, in general, amines cause a dramatic impact on the optoelectronic properties and morphology of the ZnO films by degrading sol transparency and stability 22 . Therefore, the technology to obtain ZnO thin films from sol-gel process is not definitively controlled yet and requires further investigation. This paper presents a detailed analysis of the origin of the ink (ZAD+EA+ME) degradation during storage at room conditions. After an introduction to the materials and methods, the experimental results are correlated with Density Functional Theory (DFT), which is known to be a useful and powerful tool to understand the evolution of many organic molecules, condensed substances and even metal organic complexes. A final calculation allows describing in detail the equilibrium between the two most important forms, since, as already shown 23 , the reaction of EA with ZAD produces tetramers ([Zn] 4 , Figure 1) that are not stable in solution. Our results provide a better understanding of the ink degradation and, consequently, allow optimizing the technological processes to prepare precursors, in order to obtain good ZnO films 24 .

Materials and procedures
The ZnO precursor was prepared by stabilizing zinc acetate dihydrate (Zn(CH 3 COO) 2 ·2H 2 O, ZAD, from Panreac) with ethanolamine (H 2 NCH 2 CH 2 OH, EA, from Acrös Organics). Then, 2methoxyethanol (CH 3 O(CH 2 ) 2 OH, ME, from Aldrich) was added as a solvent to prepare the ink, following the method reported in previous work 23

Quantum mechanical simulations
To elucidate the effect of light, CO 2 and ME on the stability of the reagents and the precursor, the energy levels and stability of the different molecules appearing in the ZAD-EA-ME system were deduced from quantum mechanical simulations.

Stability of EA with ME and ZAD under different conditions
Several experiments were carried out in order to compare the stability of the freshly prepared ink (EA+ZAD+ME) with those of the free ethanolamine (EA) and EA in 2methoxyethanol (EA+ME). In a first step, the samples were kept under illumination but avoiding the contact with air. After 4 weeks, no modification was apparent in any of the samples ( Figure 4a).
Next, the samples were exposed to the action of air by uncovering the flasks. Figs. 4b to 4e show the evolution of the appearance of the samples after 2, 4, 6 and 8 weeks in contact with air under illumination. A slight change in the aspect of EA and EA+ME+ZAD was detected during the first 3 weeks, while EA+ME remained colourless and transparent ( Figure 4b). This observation indicates that this ink is clearly more stable than those containing more rigid aminoalcohols 17 already mentioned in the Introduction because, in the presence of ZAD and ME, these produce black suspensions in a few minutes (if R = OH) or hours (if R = CH 2 OH) at room conditions.
At the fourth week, the ink became cloudy and precipitated, while EA and EA+ME turned yellowish and orange, respectively, suggesting that some chemical transformation was taking place. As shown in Figures 4c to 4e, the colour change is more apparent in EA+ME, thus indicating that the solvent ME plays a role in this process and perhaps determines the nature of the products. Although some degradation by light may have occurred before the flasks were uncovered, degradation was only visible when the samples were kept in contact with air.
To assess how important was light excitation in these degradation processes, in a third experiment fresh samples were kept uncovered in the dark. After 4 weeks, the appearance of EA and EA+ME was similar to that observed in the previous experiment, while the complete ink showed less precipitate. Consequently, the presence of light mainly affects the sample containing Zn.
In summary, these experiments demonstrate that the degradation of all three samples is mainly due to reaction with some species contained in air, with a minor influence of light excitation in the ink degradation. The following experiments are thus focused on identifying the reacting species through the degradation products.  A 1 H-NMR experiment was performed for the 8-weeks degraded EA+ME sample (at the centre in Figure 4e), to find out the degradation products. MeOD was used as solvent for NMR, owing to the low solubility of the mixture. During specimen preparation, the deep orange colour of EA+ME fainted giving a pale-yellow solution. This could be ascribed a) to the different polarity of the solvent (MeOD, µ D = 3.09 D) and ME (µ D = 2.14 D), or b) to their ability to establish intermolecular hydrogen-bond contacts with the species in solution.
The H-NMR spectrum ( Figure S1, supporting information) suggests the coexistence of several species in solution, as several singlets were detected in the range 6.8 -8.5 ppm. Their positions are compatible with the presence of protons bound to N atoms attached to -CO or -COO arrays, as the carbamic acids, zwitterions and carbamates (see Scheme 1) recently proposed by Kortunov 33 when amines and aminoalcohols capture CO 2 . The nature of the final products and their relative abundance are described to depend on a wide variety factors such as temperature, amine or aminoalcohol concentration, solvent nature (specially its polarity and its tendency to establish hydrogen bonds), and CO 2 partial. In turn, these complexes have been reported to react with -OH groups and form several lactams and lactims with different colouration 34,35 . and the pH (and thus the presence of ME and MeOD) seems to influence the tautomer ratio between them 36-37 .
Hence, these observations suggest that ME promotes EA degradation even at room temperature, while the precursor formed when ZAD is added to the EA+ME mixture 23  To confirm this hypothesis, the degraded samples were also measured by EGA. The peak patterns obtained in this case were similar to the non-degraded ones, but the m/z = 76 EGA signal appeared to be much more intense. As shown in Figure 5b, the degraded EA+ME presents an m/z = 76 peak which is one order of magnitude more intense than the one in Figure 5a (as indicated by the multiplication factors in both graphs), suggesting a more relevant presence of carbamates (as no ME was added) as compared to deprotonated EA (m/z = 60) and the other fragments. Consequently, after degradation there is less EA while one of the degradation products (m/z = 76, scheme 1) has clearly increased. Mass spectrometry of EGA can thus provide not only qualitative but also quantitative clues of the ink degradation at room temperature by means of the m/z = 76 vs. m/z = 60 ratio.

Time-Dependent Density Functional Theory, TD-DFT, calculations
These experimental results support the proposed reaction mechanism (Scheme 1). However, it is still not clear what prevents EA and EA+ME to degrade more under illumination, while the complete ink does. In order to understand the influence of external factors -such as exposition to light or air-on the mixtures stability and their tendency to react, computational studies based on TD-DFT and Molecular Orbital (MO) calculations were undertaken.
For comparison purposes, an initial study of EA both in vacuum and in ME was carried out.
The results, summarized in Figure 6 and Table 1, revealed that: i) the HOMO-LUMO gap energy is high (> 8.5 eV) in both cases; ii) the first absorption bands involve mainly the electronic HOMO → LUMO transition; and iii) in ME, these bands shift to higher energies and the weight of the HOMO --> LUMO transition increases as well as its oscillator strength, indicating a higher tendency of this transition to occur.
Moreover, the contribution of the nitrogen 2p z orbital in the HOMO of EA+ME is higher than for the isolated aminoalcohol (see Figure 6). This indicates that it becomes more nucleophilic, and therefore more prone to react with CO 2 . This is consistent with the faster degradation of EA+ME during storage in air.   (Figure 7 and Table 2) the LUMO is mainly a π* orbital of the bidentate AcOligand, while it is basically Zn(II) centred when the calculations are performed for vacuum or in ME. Moreover, the contribution of the atoms of the bridging ligands and EA's N is also modified for [Zn] 2 +CO 2 . Thus, carbon dioxide will react not only with the excited state but it will act as an assistant during photoexcitation, leading to a destabilization of the precursor. Also, in contrast with the presence of ME, the increase of nuclearity and the presence of CO 2 makes increase the wavelength towards the visible range and its transition probability.
Finally, the combined presence of both ME and CO 2 does not change significantly the excitation wavelengths and probability weights of the monomer. On the contrary, for [Zn] 2 , TD-DFT calculations suggest that ME slightly reduces these two parameters independently of the presence of CO 2 . Therefore, the dimer is found to be sensitive to ME and CO 2 , but, when both are present, ME seems to dominate.
In summary, these TD-DFT calculations have proved that: i) ME and CO 2 induce significant and relevant changes on the properties and reactivity of both [Zn] 1 and [Zn] 2 , and ii) the dimer is expected to be more reactive and photosensitive than the monomer.  alone and with ME, CO 2 and both.
In view of these findings, the only thing necessary to finish describing the ink degradation is the equilibrium ratio between monomers and dimers. Consequenlly, in order to get further information on the relative stability of the two species, the study of the formation of [Zn] 2 from two isolated [Zn] 1 units was envisaged.

Dimerization
As shown in our previous paper 23 , the tetramer is the most stable species in solid state whereas, in the ink, the dimer is found to be more abundant than the monomer. In this section, simulations will confirm the higher stability of [Zn] 2 .
Calculations based on Density Functional Theory (DFT) to elucidate the reaction path from two infinitely separated monomers were carried out at 0 K. According to these simulations, state in Figure 8). As shown in Figure 8

CONCLUSIONS
The present paper has reported on a detailed study, relying on experiments as well as on computational simulations, on the stability of an ink based on ZAD+EA for generating ZnO films. It has been shown that degradation is mainly driven by reaction of EA with atmospheric CO 2 , and that EA reactivity increases when it is dissolved in ME and, even further, in the EA+ME+ZAD ink. This has been explained by TD-DFT calculations as due to the fact that N and O atoms of deprotonated EA become more nucleophilic in EA+ME and in EA+ME+ZAD, respectively. The diminution of EA concentration in the degraded ink has been seen by EGA and NMR experiments, which also delivered information about the degradation products.
Experiments have also shown that reactivity with CO 2 is affected by illumination mainly when the sample contains Zn. In fact, the energy required for the first electronic transition decreases when EA+ZAD comes into contact with ME but CO 2 destabilizes the precursor, acting as assistant during photodegradation. Finally, calculations make clear that the dimer is less stable under photoexcitation than the monomer. This is relevant because, in solution, the monomers tend to combine into dimers. The reaction path of this process has been elucidated.
In conclusion, for the first time, degradation of a sol-gel ZnO precursor ink containing EA has been described, and the factors influencing its stability have been detailed.

Contents :
Mathematical details for calculations Table S1. Final atomic coordinates (in Å) of the optimized EA.    S1. Partial view of the 1H-NMR spectrum (400MHz) in the range 6.7 < δ < 8.7 ppm of the sample containing EA+ME after 8 weeks exposed to light and air. KGaA, Weinheim, Germany.    S1. Partial view of the 1 H-NMR spectrum (400MHz) in the range 6.7 < δ < 8.7 ppm of the sample containing EA+ME after 8 weeks exposed to light and air.

MATHEMATICAL DETAILS FOR CALCULATIONS
Time-Dependent Density Functional Theory (TD-DFT) method 40 is based on the excitation energy calculations from the eigenvalues of the matrix H (2) H (1) , where H (1) and H (2) are the Hessian's real and imaginary orbitals respectively, The main objective of the DFT method is then to find the suitable energy that minimizes the expression of EKS. From the form of the Schrödinger equation one can see that the energy functional contains three terms: the kinetic energy (T), interaction with the external potential (V ex ), the Hartree potential (V h ) and the electron-electron interaction (V ee ), and thus one may write the functional as E = T + V ex + V h + E xc .
This set of non-linear equations (the KS equations) describes the behavior of non-interacting "electrons" in an effective local potential. For the exact functional, and thus exact local potential, the "orbitals" yield the exact ground state density and exact ground state energy.
The KS approach achieves an exact correspondence of the density and ground state energy of a system consisting of non-interacting fermions and the "real" many body system described by the Schrödinger equation.
For calculations in which the energy surface is the quantity of primary interest, DFT offers a practical and potential highly accurate alternative to the wavefunction methods. In practice, the utility of the theory rests on the approximation used for the exchange-correlation functional, E xc .

S4
Thus, using the same nomenclature as in reference 40 , the virtual orbitals are labelled as a, b, c; where i, j and k are the occupied orbitals and p and q are any molecular orbitals (virtually occupied). The excitation energy derivative ΔE is thus: The calculation of the electronic energy derivative leads to the optimal algorithm as follows: Where h λ αβ and (αβ|γδ) λ are the derivatives of the one-and two-electron integrals in the atomic orbital (AO) basis, respectively, and S is the derivative of the overlap integrals. The coefficients γ and Γ are the relaxed one-and two-particle density matrices, respectively, and E xc contains the exchange-correlation functional terms.
The operator d/dλ is presented as: (5) And every component of the algorithm: 1.
being the ground-state one-particle density matrix . 2.