A Comparative Study of the Semiconductor Behavior of Organic Thin Films: TCNQ-Doped Cobalt Phthalocyanine and Cobalt Octaethylporphyrin

The structure formed by cobalt phthalocyanine (CoPc) and cobalt octaethylporphyrin (CoOEP) with electron-acceptor tetracyano-π-quinodimethane (TCNQ), was studied by Density Functional Theory (DFT) methods. According to theoretical calculations, both cobalt systems can establish dispersion forces related to TCNQ and also in both cases the link between them is built by means of hydrogen bonds. Based on the results of these DFT calculations, we developed experimental work: the organic semiconductors were doped, and the thermal evaporation technique was used to prepare semiconductor thin films of such compounds. The structure of the films was studied by FTIR and Raman spectroscopy. The optical properties of the CoPc-TCNQ and CoOEP-TCNQ films were investigated by means of UV-Vis measurements. The results obtained were used to estimate the type of transitions and the optical bandgap. The results were compared to the previously calculated theoretical bandgap. The CoOEP-TCNQ film presented the smallest theoretical and experimental bandgap. Finally, the electrical properties of the organic semiconductors were evaluated from a PET (polyethylene terephthalate)/indium tin oxide (ITO)/cobalt macrocycle-TCNQ/silver (Ag) device we prepared. The CoOEP-TCNQ-based device showed an ohmic behavior. The device manufactured from CoPc-TCNQ also showed an ohmic behavior at low voltages, but significantly changed to SCLC (space-charge limited conductivity) at high voltage values.


Introduction
The demand for materials that are cheap, flexible and versatile has been growing in the ever-evolving electronics industry, which has led to a significant investment of resources in the development of organic molecular thin film technology, for use in optoelectronic devices [1][2][3][4][5]. Thin film technology is the basis of an astounding development in solid state electronics [6]. Film thickness, lattice dimensions, purity, surface roughness and the imperfection level of the layer represent significant parameters for defining the type, mechanism and stability of electrical transport [7]. Phthalocyanines (Pcs) and porphyrins have been incorporated into organic thin film transistors (OTFTs), organic photovoltaics (OPVs), and organic light emitting diodes (OLEDs). Pcs and porphyrins are a promising

DFT Results and Discussion
Representations of the transition metal complexes considered in this study are shown in Figure 1. Theoretical calculations were carried out in order to establish the possibility of formation of charge transfer complexes between these molecules and TCNQ. Both cobalt systems, with phtalocyanine and porphyrin, can establish dispersion forces related to TCNQ; additionally, in both cases the link between them is built by means of hydrogen bonds, which arise from the interaction of terminal nitrogen atoms of TCNQ and peripheral hydrogen atoms from the organic rings. There are different bond lengths and positions for such interactions, since the transition metal complexes have two different kinds of peripheral hydrogen atoms: those which are joined at the end of the aromatic rings and those which saturate the methyl groups; this last category is found only in the porphyrin case. The hydrogen bond interactions can only be established on the periphery of the rings; therefore, two kinds of interactions may exist: one with hydrogen atoms with an aliphatic source and one with hydrogen atoms with an aromatic source. The description of both kinds of structures is given below in Figure 1.

DFT Results and Discussion
Representations of the transition metal complexes considered in this study are shown in Figure  1. Theoretical calculations were carried out in order to establish the possibility of formation of charge transfer complexes between these molecules and TCNQ. Both cobalt systems, with phtalocyanine and porphyrin, can establish dispersion forces related to TCNQ; additionally, in both cases the link between them is built by means of hydrogen bonds, which arise from the interaction of terminal nitrogen atoms of TCNQ and peripheral hydrogen atoms from the organic rings. There are different bond lengths and positions for such interactions, since the transition metal complexes have two different kinds of peripheral hydrogen atoms: those which are joined at the end of the aromatic rings and those which saturate the methyl groups; this last category is found only in the porphyrin case. The hydrogen bond interactions can only be established on the periphery of the rings; therefore, two kinds of interactions may exist: one with hydrogen atoms with an aliphatic source and one with hydrogen atoms with an aromatic source. The description of both kinds of structures is given below in Figure 1. Firstly, there are two possible complexes between the cobalt porphyrin compound and TCNQ via hydrogen bonds, as is indicated above. They arise because the porphyrin ring has two types of terminal hydrogen atoms on its periphery. The first group corresponds to the saturation sphere of the methyl groups and the other to the sole hydrogen atom on the aromatic ring between the corner rings. Calculations indicate that both hydrogen bonds can be formed. Figure 2 shows the structure of both kinds of porphyrin complexes, specifically showing the possible hydrogen bonds. The bond lengths in the two cases vary, but both values are short enough to allow this kind of interaction. The hydrogen bond between the nitrogen atom from TCNQ and the aliphatic hydrogen atoms from the porphyrin is very short, 2.13 Å, in contrast to that with aromatic terminal hydrogen atoms of the same molecule, which is 2.94 Å. However, the Grimme correction energy difference between the two kinds of hydrogen bond is not particularly large; the first one is 10.9 kcal/mol and the second is 8.9 kcal/mol. A statistical Boltzman energy distribution analysis shows that the population of both conformations is 57%, favoring the shorter interaction structure. Firstly, there are two possible complexes between the cobalt porphyrin compound and TCNQ via hydrogen bonds, as is indicated above. They arise because the porphyrin ring has two types of terminal hydrogen atoms on its periphery. The first group corresponds to the saturation sphere of the methyl groups and the other to the sole hydrogen atom on the aromatic ring between the corner rings. Calculations indicate that both hydrogen bonds can be formed. Figure 2 shows the structure of both kinds of porphyrin complexes, specifically showing the possible hydrogen bonds. The bond lengths in the two cases vary, but both values are short enough to allow this kind of interaction. The hydrogen bond between the nitrogen atom from TCNQ and the aliphatic hydrogen atoms from the porphyrin is very short, 2.13 Å, in contrast to that with aromatic terminal hydrogen atoms of the same molecule, which is 2.94 Å. However, the Grimme correction energy difference between the two kinds of hydrogen bond is not particularly large; the first one is 10.9 kcal/mol and the second is 8.9 kcal/mol. A statistical Boltzman energy distribution analysis shows that the population of both conformations is 57%, favoring the shorter interaction structure. The frontier molecular orbital analysis is in agreement with the behavior predicted for the charge transfer complexes, and Figures 3 and 4 show the shape of the HOMO (Highest Occupied Molecular Orbital, π) and LUMO (Lowest Unoccupied Molecular Orbital, π*) for both conformations; however, notably in the case of the HOMO, there is an accidental degeneration, and in the case of the LUMO, the LUMO+1 orbital is close enough to the LUMO to suggest a similar accidental degeneration. As is expected for the HOMO in both cases, the strong probability of electron presence is found in the zone of the peripheral framework of the porphyrin moiety. Similarly, the LUMO shows the presence of empty regions mainly in the TCNQ zone, although there is also participation of the porphyrin ring's central part, i.e., the cobalt atom coordination sphere. The energy gap for this species is 0.89 eV. The average value for an energy gap should be 0.9 eV, a result which suggests strong semiconductor behavior. The scheme for the second conformation is practically the same, and the band gap is 0.91 eV, meaning that the behavior should be similar in both cases, in spite of the different positions and lengths of the hydrogen bonds. Based on the above, the calculated energy value for the HOMO is of −8.245 eV, while the value for the LUMO is −7.265 eV.  Hydrogen bond complexes between cobalt octaethylporphyrin (CoOEP) and tetracyano-π-quinodimethane (TCNQ): (a) hydrogen atoms with an aliphatic source and (b) hydrogen atoms with an aromatic source.
The frontier molecular orbital analysis is in agreement with the behavior predicted for the charge transfer complexes, and Figures 3 and 4 show the shape of the HOMO (Highest Occupied Molecular Orbital, π) and LUMO (Lowest Unoccupied Molecular Orbital, π*) for both conformations; however, notably in the case of the HOMO, there is an accidental degeneration, and in the case of the LUMO, the LUMO+1 orbital is close enough to the LUMO to suggest a similar accidental degeneration. As is expected for the HOMO in both cases, the strong probability of electron presence is found in the zone of the peripheral framework of the porphyrin moiety. Similarly, the LUMO shows the presence of empty regions mainly in the TCNQ zone, although there is also participation of the porphyrin ring's central part, i.e., the cobalt atom coordination sphere. The energy gap for this species is 0.89 eV. The average value for an energy gap should be 0.9 eV, a result which suggests strong semiconductor behavior. The scheme for the second conformation is practically the same, and the band gap is 0.91 eV, meaning that the behavior should be similar in both cases, in spite of the different positions and lengths of the hydrogen bonds. Based on the above, the calculated energy value for the HOMO is of −8.245 eV, while the value for the LUMO is −7.265 eV. The frontier molecular orbital analysis is in agreement with the behavior predicted for the charge transfer complexes, and Figures 3 and 4 show the shape of the HOMO (Highest Occupied Molecular Orbital, π) and LUMO (Lowest Unoccupied Molecular Orbital, π*) for both conformations; however, notably in the case of the HOMO, there is an accidental degeneration, and in the case of the LUMO, the LUMO+1 orbital is close enough to the LUMO to suggest a similar accidental degeneration. As is expected for the HOMO in both cases, the strong probability of electron presence is found in the zone of the peripheral framework of the porphyrin moiety. Similarly, the LUMO shows the presence of empty regions mainly in the TCNQ zone, although there is also participation of the porphyrin ring's central part, i.e., the cobalt atom coordination sphere. The energy gap for this species is 0.89 eV. The average value for an energy gap should be 0.9 eV, a result which suggests strong semiconductor behavior. The scheme for the second conformation is practically the same, and the band gap is 0.91 eV, meaning that the behavior should be similar in both cases, in spite of the different positions and lengths of the hydrogen bonds. Based on the above, the calculated energy value for the HOMO is of −8.245 eV, while the value for the LUMO is −7.265 eV.   In the case of the phtalocyanine complex, the TCNQ moiety settles into a horizontal position, in order to generate two interactions between itself and the lateral aromatic rings, resulting in two hydrogen bonds with an average length of 2.43 Å and a total energy of 18.5 kcal/mol. The frontier molecular orbital analysis shows that the HOMO probability is focused on the phthalocyanine moiety, whereas the LUMO is mainly found in the TCNQ zone; these probability regions are presented in Figure 5. In this case, the values calculated for the HOMO and the LUMO are −5.333 and −3.075 eV, respectively, and the band gap is 2.35 eV, which indicates a moderate semiconductor behavior. These DFT calculations are indicative of a better semiconductor behavior in the CoOEP-TCNQ system, with respect to CoPc-TCNQ. The above may be due to the greater number of interactions through hydrogen bonds and the shorter bond length of 2.13 Å between the nitrogen atom from TCNQ and the aliphatic hydrogen atoms from the porphyrin.  In the case of the phtalocyanine complex, the TCNQ moiety settles into a horizontal position, in order to generate two interactions between itself and the lateral aromatic rings, resulting in two hydrogen bonds with an average length of 2.43 Å and a total energy of 18.5 kcal/mol. The frontier molecular orbital analysis shows that the HOMO probability is focused on the phthalocyanine moiety, whereas the LUMO is mainly found in the TCNQ zone; these probability regions are presented in Figure 5. In this case, the values calculated for the HOMO and the LUMO are −5.333 and −3.075 eV, respectively, and the band gap is 2.35 eV, which indicates a moderate semiconductor behavior. These DFT calculations are indicative of a better semiconductor behavior in the CoOEP-TCNQ system, with respect to CoPc-TCNQ. The above may be due to the greater number of interactions through hydrogen bonds and the shorter bond length of 2.13 Å between the nitrogen atom from TCNQ and the aliphatic hydrogen atoms from the porphyrin. In the case of the phtalocyanine complex, the TCNQ moiety settles into a horizontal position, in order to generate two interactions between itself and the lateral aromatic rings, resulting in two hydrogen bonds with an average length of 2.43 Å and a total energy of 18.5 kcal/mol. The frontier molecular orbital analysis shows that the HOMO probability is focused on the phthalocyanine moiety, whereas the LUMO is mainly found in the TCNQ zone; these probability regions are presented in Figure 5. In this case, the values calculated for the HOMO and the LUMO are −5.333 and −3.075 eV, respectively, and the band gap is 2.35 eV, which indicates a moderate semiconductor behavior. These DFT calculations are indicative of a better semiconductor behavior in the CoOEP-TCNQ system, with respect to CoPc-TCNQ. The above may be due to the greater number of interactions through hydrogen bonds and the shorter bond length of 2.13 Å between the nitrogen atom from TCNQ and the aliphatic hydrogen atoms from the porphyrin.

Experimental Results and Discussion
On the basis of the DFT calculations, the preparation of the semiconductor materials CoPc-TCNQ and CoOEP-TCNQ was carried out. In this molecular doping, the mixture between the acceptor and the two donor species is carried out, in order to facilitate the electronic transfer between both types of species. TCNQ is chosen to be the host/matrix organic material and the cobalt complexes have been used as dopants. The quantity of CoPc and CoOEP in the semiconductor material was controlled in terms of their mass and in a macrocycle-TCNQ ratio of 0.3:1 [42,43]. The doping of organic semiconductors usually requires this dopant:matrix ratio, so that dopant-semiconductor and dopant-dopant intermolecular contacts are fostered [42]. In order to study the possible applications for molecular electronics, thin films of both semiconductors were deposited by the low-pressure evaporation technique. As the semiconductors are subjected to considerable thermal gradients during their evaporation and deposition, FTIR spectroscopy was employed to verify that no decomposition of these materials occurred. The FTIR spectra of the semiconductor films deposited over monocrystalline silicon were compared to those obtained for the powders as KBr pellets, and the band assignments are presented in Table 1. The FTIR spectra for the CoOEP-TCNQ and CoPc-TCNQ doped films and their precursors (TCNQ, CoOEP and CoPc) are shown in Figure 6. CoOEP (Figure 1a) is made up of cyclic tetrapyrroles; the pyrrole forms a cycle by means of bridging carbons and differs from CoPc ( Figure 1b) in that the latter contains nitrogen atoms in a meso-position. CoPc presents four isoindole units (a pyrrole ring fused to a benzene ring), joined by four nitrogen atoms. The metallooctaethylporphyrine (MOEP) characteristic bands are located around 1468, 1230, 992, 924 and 755 cm −1 for the powder [44][45][46][47][48], which are very close to the position observed for the same bands in the films. As for the CoPc-based semiconductor, the band responsible for the pyrrole in-plane stretching vibration in the phthalocyanine ring is observed at 1332 cm −1 and the bands located at 1288, 1164, 1120 and 754 cm −1 are the result of the interaction between the carbon and hydrogen atoms [49]. The bands observed around 1611 and 1476 cm -1 result from a C=C stretching mode [49][50][51]. Similarly, the C≡N stretching band of TCNQ appears around 1673 cm −1 ; the peaks around 1615, 1451 and 1207 cm −1 are related to C=C-H bending, C-CN stretching and C=C ring stretching, respectively [52]. As is apparent in Table 1, the results of KBr pellets resemble closely those obtained for thin films. The slight changes in values are ascribed to the concentration of efforts in the bonds, due to the thermal shock between the temperature of the semiconductor in gaseous state and the room temperature of the substrates. In summary, the FTIR technique indicates that CoOEP-TCNQ and CoPc-TCNQ molecules manifest chemical stability and that the evaporation technique is favorable for obtaining films of these materials. Moreover, FTIR spectroscopy confirms the presence of the expected hydrogen bonds for the CoPc-TCNQ and CoOEP-TCNQ systems. These bonds can be indirectly detected by the displacement of the strong peak around 2225 cm −1 , which is characteristic of the lower frequency absorbance of conjugated nitriles [53,54].
The Raman spectra in Figure 7a confirmed the presence of the CoOEP or CoPc and the TCNQ moieties within the films. The peaks observed in the 400-1700 cm −1 range are associated with the vibrational modes in the CoOEP and CoPc molecules. Bands of benzene ring deformation appear at 589 and 592 cm −1 , and macrocycle breathing bands are observed at 697 and 703 cm −1 , connected to the pyrrole stretching for CoOEP and CoPc, respectively [55][56][57][58]. The major changes in the Raman spectra for the different films are observed in the range from 1200 to 1550 cm −1 . This region corresponds to cobalt-nitrogen in-plane stretching and bending vibrations, as well as displacements on the C-N-C bridge bond of the macrocycle [55][56][57][58]. The position change of the pyrrole stretching band is observed in Figure 7a. The band changes from 1536 cm −1 for CoOEP to 1541 cm −1 for CoPc [55]. In a similar way to that found using IR spectroscopy, the presence of the TCNQ molecule is observed in the Raman spectra: 1257, 1443 and 1568 cm −1 signals, ascribed to C=C-H bending, C-CN stretching, C=C ring stretching, respectively [58,59]. The differences in the Raman spectra of the films originate in the macrocycle arrangement. Significantly, the presence of peaks of considerable intensity around 986, and 1363 cm −1 may be indicative of the molecular interactions between the CoOEP or CoPc and TCNQ [55][56][57][58][59].      The spectral distribution of absorbance, measured at nearly normal incidence in the 200-1000 nm wavelength range for the CoOEP-TCNQ and CoPc-TCNQ films and their precursors CoOEP and CoPc is shown in Figure 7b,c. The differences in the spectra are due to the presence of TCNQ. At longer wavelengths (λ > 800 nm) the CoOEP-TCNQ film becomes almost transparent in the nonabsorbing region [44,60], while at shorter wavelengths (λ < 800 nm), the film presents absorption (Figure 7b). The main characteristic of the absorption zone of the CoOEP complex is the presence of two bands, the B band in the UV region of the spectrum, between 300-360 nm, and the Q band in the visible region, between 560 and 715 nm. The B and Q bands are associated with π-π* transitions, arising from the π conjugated electrons of the CoOEP structure. The Q band is associated with the ππ* transitions from the HOMO to the LUMO, corresponding to transitions from orbital a1u to orbital eg (see Figures 3 and 4). The B band is also the result of π-π* electronic transitions, but from orbitals a2u and b2u to the LUMO eg. On the other hand, the two UV-Vis characteristic bands are observed in the spectrum of the CoPc-TCNQ film: the Q band between 548 and 709 nm, and the B band between The spectral distribution of absorbance, measured at nearly normal incidence in the 200-1000 nm wavelength range for the CoOEP-TCNQ and CoPc-TCNQ films and their precursors CoOEP and CoPc is shown in Figure 7b,c. The differences in the spectra are due to the presence of TCNQ. At longer wavelengths (λ > 800 nm) the CoOEP-TCNQ film becomes almost transparent in the non-absorbing region [44,60], while at shorter wavelengths (λ < 800 nm), the film presents absorption (Figure 7b). The main characteristic of the absorption zone of the CoOEP complex is the presence of two bands, the B band in the UV region of the spectrum, between 300-360 nm, and the Q band in the visible region, between 560 and 715 nm. The B and Q bands are associated with π-π* transitions, arising from the π conjugated electrons of the CoOEP structure. The Q band is associated with the π-π* transitions from the HOMO to the LUMO, corresponding to transitions from orbital a 1u to orbital e g (see Figures 3  and 4). The B band is also the result of π-π* electronic transitions, but from orbitals a 2u and b 2u to the LUMO e g . On the other hand, the two UV-Vis characteristic bands are observed in the spectrum of the CoPc-TCNQ film: the Q band between 548 and 709 nm, and the B band between 300 and 371 nm (Figure 7c). When comparing the spectrum of CoPc and that of the CoPc-TCNQ film, it is observed that the presence of TCNQ shifts the Q and B bands towards the ultraviolet region and also slightly decreases the absorption of the material.
The electronic transitions in organic semiconductors can be described in terms of the band theory [44]; it is proposed that the valence band is formed by the HOMO, whereas the LUMO contributes to the conduction band [44,60,61]. In organic semiconductors the type of optical transition is usually studied, as well as the value of the optical energy gap. The absorption data conform to power-law behavior [61]: here, parameter A depends on transition probability, α is the absorption coefficient, E g opt is the optical band gap and r is a number that characterizes the transition process, where r = 1 2 and r = 2 for direct and indirect allowed transitions, respectively [61]. The dependence of (αhν) r on hν photon energy was plotted and the optical band gap was evaluated from the x-axis intercept at (αhν) 1/2 = 0 and (αhν) 2 = 0. Figure 8 shows that the CoOEP-TCNQ (Figure 8a,b) film has a greater semiconductor nature than the CoPc-TCNQ film (Figure 8c,d). Not only is the optical gap lower in the case of the film containing porphyrin, but apparently this organic semiconductor undergoes several transitions. The change in slope is an indication that more than one energy transition occurs between the energy bands of the CoOEP-TCNQ film. The first transition is the E g opt that corresponds to the onset of optical absorption and the formation of a bound electron-hole pair, or Frenkel exciton; the second transition is the fundamental energy gap (E g ) [60,61]. Another aspect to consider is that indirect transitions are the dominant transitions in thin films. This is due to the amorphous nature of the films, as a result of the deposit method used. Finally, when comparing the curves of the macrocycle-TCNQ systems with those of their corresponding precursors CoPc and CoOEP, marked differences are observed between them, as a result of the presence of TCNQ. For the case of the CoPc-TCNQ film, the optic band gap is greater than that of the CoPc macrocyclic complex. Apparently, the TCNQ does not enhance the charge transport in the phthalocyanine, and the CoPc-TCNQ films have a low semiconductor behavior. In the case of the CoOEP-TCNQ system, the electronic transitions are favored within the doped semiconductor. The optical band gap values obtained are within the range reported for cobalt porphyrin films [60]; however, in the present case, it was not necessary to perform an annealing treatment to reduce the band gap. It is the presence of the HOMO in the zone of the peripheral framework of the porphyrin moiety and the LUMO in the TCNQ region and the metal coordination sphere which favors the charge transport and the low bandgap observed. As expected, for both macrocycle-TCNQ films, the optical band gap obtained experimentally is higher than the bandgap calculated from DFT because the theoretical bandgap does not consider external effects of the semiconductor molecules (nature of the films, presence of impurities and degree of stacking of the electron donor and acceptor molecules). However, the same trend is observed in both the experimental and theoretical band gap results: the lowest band gap values are those for the film with CoOEP (1.98 and 2.1 eV for indirect and direct transitions, respectively), while the CoPc film shows a band gap of 3.8 eV for indirect transitions and 4.1 eV for direct transitions. These results are interesting due to the fact that Pcs normally have better charge transport properties compared to porphyrins [14,29]. It seems that when the cobalt(II) ion is at the center of the macrocycle, the behavior changes, probably influenced by the two types of hydrogen bonds that are generated between TCNQ and CoOEP ( Figure 2). The electrical characterization (J-V) of the devices constructed with both kinds of materials (Figure 9a,b) complement the previous results. In both graphs, it is observed that there is no difference between the measurements made on the flexible devices in dark conditions and in natural lighting. This behavior is an indication that these films do not exhibit photovoltaic properties; however, they are candidates to be used in other types of optoelectronic devices. The CoOEP-TCNQ film shows an ohmic behavior, while the CoPc-TCNQ film also has an ohmic behavior at low voltages, but changes to a space charge limited current (SCLC) mechanism at high voltages. The electric resistor (R) for ohmic regions is calculated directly from the Ohm equation. Considering a voltage of 0.2 V, which is in the linear part of both graphs of Figure 9, the resistor is of 19.8 Ω and of 2368.57 Ω for the CoPc-TCNQ and CoOEP-TCNQ devices, respectively. These values are below those reported for the widely used CuPc [62], which is indicative of the adequate charge transport capacity of these doped films. However, the resistor in the device with CoOEP-TCNQ is two orders of magnitude higher than the resistor obtained in the phthalocyanine-based device. According to Figure 9a,b, these results are an indication that the CoOEP-based device has the behavior of a resistor, while the CoPc-based device functions like a diode. In the CoOEP-TCNQ film the current density is low, there is no excess electrical charge within the film for any applied voltage (Figure 9b,d). The film forms ohmic contacts with the ITO and Ag electrodes and there is a good As expected, for both macrocycle-TCNQ films, the optical band gap obtained experimentally is higher than the bandgap calculated from DFT because the theoretical bandgap does not consider external effects of the semiconductor molecules (nature of the films, presence of impurities and degree of stacking of the electron donor and acceptor molecules). However, the same trend is observed in both the experimental and theoretical band gap results: the lowest band gap values are those for the film with CoOEP (1.98 and 2.1 eV for indirect and direct transitions, respectively), while the CoPc film shows a band gap of 3.8 eV for indirect transitions and 4.1 eV for direct transitions. These results are interesting due to the fact that Pcs normally have better charge transport properties compared to porphyrins [14,29]. It seems that when the cobalt(II) ion is at the center of the macrocycle, the behavior changes, probably influenced by the two types of hydrogen bonds that are generated between TCNQ and CoOEP ( Figure 2). The electrical characterization (J-V) of the devices constructed with both kinds of materials (Figure 9a,b) complement the previous results. In both graphs, it is observed that there is no difference between the measurements made on the flexible devices in dark conditions and in natural lighting. This behavior is an indication that these films do not exhibit photovoltaic properties; however, they are candidates to be used in other types of optoelectronic devices. The CoOEP-TCNQ film shows an ohmic behavior, while the CoPc-TCNQ film also has an ohmic behavior at low voltages, but changes to a space charge limited current (SCLC) mechanism at high voltages. The electric resistor (R) for ohmic regions is calculated directly from the Ohm equation. Considering a voltage of 0.2 V, which is in the linear part of both graphs of Figure 9, the resistor is of 19.8 Ω and of 2368.57 Ω for the CoPc-TCNQ and CoOEP-TCNQ devices, respectively. These values are below those reported for the widely used CuPc [62], which is indicative of the adequate charge transport capacity of these doped films. However, the resistor in the device with CoOEP-TCNQ is two orders of magnitude higher than the resistor obtained in the phthalocyanine-based device. According to Figure 9a,b, these results are an indication that the CoOEP-based device has the behavior of a resistor, while the CoPc-based device functions like a diode. In the CoOEP-TCNQ film the current density is low, there is no excess electrical charge within the film for any applied voltage (Figure 9b,d). The film forms ohmic contacts with the ITO and Ag electrodes and there is a good energetic correlation between the work function of the electrode and the energetic level that participates in the charge transport. Furthermore, the polarity in the device is totally symmetrical, with no current rectification. When the polarity of the voltage applied across the electrodes is reversed, the current density has the same magnitude. On the other hand, the device with the CoPc-TCNQ film (Figure 9a,c) changes its behavior around 0.23 V; while at lower voltages (<0.23 V) the ohmic behavior keeps predominating, at higher voltages (>0.23 V) an SCLC transport governed by an exponential trap distribution is observed (Figure 9a) [63][64][65]. In CoPc-TCNQ, the electric transport properties are associated with the tunneling of charges through the barriers due to the difference between the HOMO and LUMO energetic levels of the semiconductor and the work function levels of the electrodes: Φ ITO = 4.7 eV and Φ Ag = 4.2 eV (Figure 9c). Since the contact is ohmic, in the vicinity of the electrodes, spatial charges are formed, which oppose the flow of current through the organic semiconductor CoPc-TCNQ; with a high enough applied field, a saturation in the current is produced. energetic correlation between the work function of the electrode and the energetic level that participates in the charge transport. Furthermore, the polarity in the device is totally symmetrical, with no current rectification. When the polarity of the voltage applied across the electrodes is reversed, the current density has the same magnitude. On the other hand, the device with the CoPc-TCNQ film (Figure 9a,c) changes its behavior around 0.23 V; while at lower voltages (<0.23 V) the ohmic behavior keeps predominating, at higher voltages (>0.23 V) an SCLC transport governed by an exponential trap distribution is observed (Figure 9a) [63][64][65]. In CoPc-TCNQ, the electric transport properties are associated with the tunneling of charges through the barriers due to the difference between the HOMO and LUMO energetic levels of the semiconductor and the work function levels of the electrodes: ɸITO = 4.7eV and ɸAg = 4.2eV (Figure 9c). Since the contact is ohmic, in the vicinity of the electrodes, spatial charges are formed, which oppose the flow of current through the organic semiconductor CoPc-TCNQ; with a high enough applied field, a saturation in the current is produced. The highest current density of 4 × 10 −4 A/cm 2 is obtained in the CoPc-based device. This value is one order of magnitude higher than the current density flowing through the porphyrin device. As for the density of the CoPc precursor, some of the authors of this work have reported J~4.75 × 10 −8 A/cm 2 values for CoPc and its derivatives [66,67], so apparently TCNQ favors current flow through the device. Finally, for CoPc-TCNQ in the ohmic region (Figure 9a), J is given by the equation [63,64]: where p0 is the concentration of thermally generated holes, e the electronic charge, μ the hole mobility, V the applied voltage and d the film thickness. In the SCLC region, J is given by the expression [63,64]: where in addition to the previously defined symbols, is the effective density of states in the valence band, ℰ is the permittivity, P0 is the concentration of traps per unit of energy, is The highest current density of 4 × 10 −4 A/cm 2 is obtained in the CoPc-based device. This value is one order of magnitude higher than the current density flowing through the porphyrin device. As for the density of the CoPc precursor, some of the authors of this work have reported J~4.75 × 10 −8 A/cm 2 values for CoPc and its derivatives [66,67], so apparently TCNQ favors current flow through the device. Finally, for CoPc-TCNQ in the ohmic region (Figure 9a), J is given by the equation [63,64]: where p 0 is the concentration of thermally generated holes, e the electronic charge, µ the hole mobility, V the applied voltage and d the film thickness. In the SCLC region, J is given by the expression [63,64]: where in addition to the previously defined symbols, N v is the effective density of states in the valence band, ε r is the permittivity, P 0 is the concentration of traps per unit of energy, k is Boltzmann constant and T L is the temperature parameter which characterizes the trap distribution, and: where T is the ambient temperature. The total trap concentration N t(e) is equal to: The electrical parameters of the CoPc-TCNQ device calculated using equation (1) for the ohmic region and equation (2) for the SCLC region are shown in Table 2. The µ value is in the interval reported for other devices manufactured with phthalocyanine derivatives (10 −10 -10 −7 m 2 V −1 s −1 ) [35,43,64,65]. With respect to the results obtained for the concentration of thermally generated holes (p 0 ), they are larger than those found for NiPc and CoPc thin films and of the same order of magnitude of those reported for CuPc and ZnPc derivatives [35,63,64]. The above is considered an indication that the CoPc-TCNQ films are better semiconductors than CoPc films without dopant. As regards the concentration of traps per unit of energy (Po), the value obtained is within the range reported for some MPcs (M = Ni, Cu, Co, Pb, Zn): 8 × 10 43 to 1.15 × 10 47 J −1 m −3 . Finally, the total trap concentration (N t(e) ) calculated is also in the range of the same phthalocyanines previously mentioned (6 × 10 20 to 9.3 × 10 26 m −3 ) [35,43,[63][64][65][66][67][68][69]. Of course, these values depend on the purity of the MPc, the doping conditions, the phases present, the thermal treatment and the material of the electrodes in the devices. In the current study it is evident that, although the CoOEP-TCNQ system exhibits a lower theoretical and experimental band gap, as well as two types of interactions through hydrogen bonding, it is the device with CoPc-TCNQ which presents a better electric behavior. According to the diagram of Figure 9c, the energy position of the HOMO and LUMO orbitals respect to the work function of the electrodes gains importance in the electric charge flow.

DFT Calculations
All calculations were carried out by applying a DFT method based on a combination of Becke's gradient corrections [70] for exchange and Perdew-Wang's for correlation [71]. This is the scheme for the B3PW91 method, which is included in the Gaussian16 [72] package. The calculations were performed using the 6-31G** basis set. Frequency calculations were carried out at the same level of theory in order to confirm that the optimized structures were at a minimum of the potential surfaces. The hydrogen bonds were studied by applying the Atoms in Molecules Theory (AIM) [73] and the dispersion Grimme correction (G3), taking advantage of the version included in the DFT-D3 method [74].

Materials and Methods
CoPc (cobalt(II) phthalocyanine: C 32 H 16 N 8 Co), CoOEP (2,3,7,8,12,13,17,18-octaethyl-21H,23Hporphine cobalt(II): C 36 H 44 CoN 4 ) and TCNQ (7,7,8,8-tetracyanoquinodimethane: C 12 H 4 N 4 ) were purchased from Sigma-Aldrich (Saint Louis, MO, USA) and required no further purification. The doped organic semiconductors were synthesized in a heated Monowave 50 reactor (Anton Paar México, S.A. de C.V., Hidalgo, Mexico), which was operated using a borosilicate glass vial with an integrated pressure (0-20 bar) and temperature sensor. A total of 1.03 g (5 mmol) of TCNQ was added to 1 g (1.7 mmol) of CoPc or 1 g (1.7 mmol) of CoOEP and dissolved in methanol or propanol, respectively. They were kept in the reactor for 25 min and then cooled and brought back to atmospheric pressure; next, they were filtered, washed and dried in vacuum. The doped semiconductors were subsequently deposited by the vacuum evaporation technique (Intercovamex, S.A. de C.V., Cuernavaca, Morelos, Mexico) onto different substrates: monocrystalline n-type silicon wafers (c-Si), quartz and indium tin oxide (In 2 O 3 ·(SnO 2 ) x ) coated polyethylene terephthalate (PET-ITO) films. Previously, all substrates, excluding PET-ITO, were cleansed by applying an ultrasonic process, using chloroform, methanol and acetone, and then dried in vacuum. Using tantalum crucibles, the doped semiconductors were heated to 580 K in order to produce their phase change, which was initially carried out in the gaseous state, so that they would finally be deposited in the form of thin films upon contact with the substrates, which were held at room temperature and a vacuum pressure of 1 × 10 −6 Torr. Due to the different melting temperatures of each doped semiconductor, the deposit speed was 0.8 Å/s. The thickness of each layer was monitored using a microbalance quartz crystal monitor, connected to a thickness sensor.
To verify the main functional groups of the organic ligands, an FTIR spectroscopy analysis was performed for the compounds as KBr pellets and for the films on silicon substrates, using a Nicolet iS5-FT spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA) at a wavelength range of 4000 to 500 cm −1 . The Raman spectra of the films on quartz substrates were recorded on a Horiba LabRAM HR Evolution spectrometer (Horiba, Ltd., Minami-ku Kyoto, Japan), using the 632.8 (2 mW) nm line of an HeNe laser for spectral excitation and a 50× objective. The absorbance and transmittance of the powders and films on quartz were obtained in the 200-1100 nm wavelength range, on a UV-Vis 300 Unicam spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). For the electrical characterization of the devices on ITO-coated PET substrates, a programmable voltage source, a sensing station with lighting and temperature controller circuit from Next Robotix (Comercializadora K Mox, S.A. de C.V., Mexico City, Mexico) and an auto-ranging Keithley 4200-SCS-PK1 pico-ammeter (Tektronix Inc., Beaverton, OR, USA) were employed. The evaluation of the electrical behavior of the flexible devices was performed both under lighting conditions and in darkness. The film sequence PET/ITO/cobalt macrocycle-TCNQ/Ag was used in the device setup. After manufacturing the devices, their electric properties were studied to carry out the current density vs. voltage analysis and to determine the type of electric behavior in the devices.

Conclusions
According to the DFT theoretical calculations, the cobalt macrocycles bind to TCNQ by hydrogen bonds. There are two possible complexes between the CoOEP and TCNQ: the first corresponds to the saturation sphere of the methyl groups and the other to the sole hydrogen atom on the aromatic ring between the corner rings. The hydrogen bond length in the first case is 2.13 Å, while in the second possible complex is 2.43 Å. In the case of the phtalocyanine complex, the TCNQ settles into a horizontal position, in order to generate two interactions between itself and the lateral aromatic rings, resulting in two hydrogen bonds with an average length of 2.43 Å. Experimentally, we synthesized the TCNQ doped semiconductors and prepared thin films by the thermal evaporation technique. The CoOEP-TCNQ system exhibits the smallest theoretical and experimental band gap (0.89 and 1.98 eV, respectively), which is an indication of its better semiconductor behavior. In additional work, flexible PET/ITO/cobalt macrocycle-TCNQ/Ag devices were manufactured, and upon evaluating their electrical characteristics (J-V), it was found that the CoOEP-TCNQ film has the behavior of a resistor, while the CoPc-TCNQ film shows an ohmic behavior at low voltages, which significantly changed to SCLC at high voltage values. The highest current density of 4 × 10 −4 A/cm 2 is obtained for the CoPc-based device. This value is one order of magnitude higher than the 4 × 10 −5 A/cm 2 flowing through the porphyrin device.