CuPcs and ZnPcs, with various degrees of fluorination (CuPc, F₁₆-CuPc, ZnPc, F₄-ZnPc, F₁₆-ZnPc) (Fig. 1), were deposited as 400 Å thin-films by PVD at 25 °C with a rate of 0.2 Å/s on OTS treated Si/SiO₂ substrates with and without prepatterned gold electrodes. Pre-exposure GIWAXS and XRD spectra (Figure S1) revealed films with a high intensity (100) plane at q = ~0.50 Å⁻¹ and a single XRD peak with 2θ between 6–7°, suggesting a preferential orientation of the Pc ring to the substrate surface of ~75°; typical of α-crystal orientations with co-facial herringbone stacking^21,39. Pre-exposure output and transfer curves (Figure S2) demonstrated OTFT performance in agreement with other sources^7,27,40,41, with complete peripheral fluorination (F₁₆-) of both CuPc and ZnPc conferring n-type OTFT behavior.

Pc thin-films were exposed to 4 ppm THC vapor over a period of 90 seconds and the percent change in peak mobility (%Δµ), voltage threshold shift (ΔV_T), defect density (ΔN), hysteresis (%ΔHys), on/off current ratio (ΔOn/Off), and peak XRD intensity (%ΔXRD) was determined for each material and displayed in Fig. 1a. As previously established¹⁸, analytes, such as THC, when introduced to Pc thin films, in addition to the effects of intermolecular interactions, can also alter physical characteristics of the thin film, such as crystal packing and morphology, which can cause changes in OTFT performance. Here, with the exception of devices made with F₄-ZnPc, the sign, and magnitude, of the change in XRD peak height correlates to the magnitude of the change in µ and N. By XRD and GIWAXS, the absence of any observed new crystal morphologies suggests that, without the formation of new polymorphs, mobility decreases are related to the degree of change in α-crystallinity of the bulk film. Limited changes in hysteresis and negligible OTFT bias stress effects suggests that exposure of OTFTs to THC vapor results in irreversible and relatively stable physical thin-film changes²⁹.

In our previous works^14,18, we found that THC, as a π-conjugated molecule with electron-donating properties⁴², interacts with the central metal of MPcs, and that these interactions are deferential to the species of the central metal. The molecular size, electronegativity, and valence structure of the central metal affects analyte interactions, which, in addition to electrochemical effects, can also result in physical effects, such as distortions to the Pc ring, and induce physical charge traps³⁸. Here, CuPc and F₁₆-CuPc-based devices both show + ΔV_T with exposure to THC, while the ZnPc, F₄-ZnPc, and F₁₆-ZnPc-based devices show -ΔV_T, suggesting that the copper-THC interaction induces electron trapping while the zinc-THC interaction creates deep, but short life-time, hole traps^29,43. The peripherally fluorinated Pcs demonstrate smaller changes in XRD peak intensity, ΔN, and ΔOn/Off, suggesting limitations to both metal-analyte interactions and film restructuring as a result of their increased size or peripheral electronegativity. With the exception of hysteresis, an overall comparison of the transfer curves and XRD spectra demonstrates that ZnPc devices undergo the greatest changes across all metrics with exposure to THC vapor. This further illustrates the critical impacts of analyte-Pc interactions and thin-film restructuring on over all device performance, highlighting the role of material selection in optimizing sensing responses.

With the greatest changes observed for ZnPc devices, we sought to further examine the effects of thin-film structures by preparing 400 Å α-ZnPc (2θ = 6.92°) OTFTs by altering the PVD rate or deposition substrate temperature, and altering the substrate surface by pre-depositing a monolayer of p-sexiphenyl (p-6P) as a patterning agent to afford varying degrees of α-crystallinity^44,45. The fastest deposition rate (1 Å/s) at the lowest deposition substrate temperature (25 °C) resulted in small uniform grains by AFM, low peak intensity by XRD, and poor transfer characteristics (Fig. 2). Decreasing deposition rate, increasing deposition substrate temperature, or deposition on p-6P as a patterning agent, led to larger grain sizes, increased XRD peak intensity, and generally improved OTFT transfer characteristics (Table S1). A small peak, characteristic of β-ZnPc (2θ = 9.32°), was observed (Figure S3) for the depositions carried out at 180 °C, suggesting that some α- to β- phase transition was beginning to occur, however the relative intensity to the α- peak (2θ = 6.92°) was less than 2%.

Exposure of the least crystalline films to 4 ppm of THC vapor rendered the OTFTs inoperable. Films were instead exposed to 400 ppb THC vapor over a period of 90 seconds, inducing the greatest changes in the least crystalline films, and manifesting as -%Δµ, -ΔV_T, an increase in hysteresis, decreased On/Off, and + ΔN (Table S1). Increased thin-film crystallinity reduced OTFT sensing responses, with the most crystalline films demonstrating limited changes in charge transport characteristics and negligible changes in XRD intensity. Measured by AFM, the surface area of each film was inversely proportional to the degree of alpha crystallinity, where the very high crystallinity films had the least surface area, while the low crystallinity films had the greatest surface area (very high = 6.43, high = 6.52, medium = 6.61, low = 6.81 µm of surface area in a 2.5×2.5 µm section). This suggests that smaller grains in the low crystallinity films affords increased surface area for ZnPc-THC interactions and intermolecular forces which facilitate greater changes by XRD and thus greater OTFT sensing responses⁴⁶. Previously, we demonstrated a relationship between different Pcs, their analyte-induced physical film changes, and the measured OTFT sensing responses¹⁸. Here, these findings demonstrate that this relationship between physical thin-film changes and an OTFT sensor response can be applied to a single material, highlighting the importance of thin-film nanostructures for analytes-sensor interactions.

To assay the effects of thickness on the most sensitive thin-film conditions, we prepared OTFTs with 200 and 800 Å, low-crystallinity α-ZnPc films (deposition rate 1 Å/s at 25 °C) and exposed them to 40 ppb THC vapor over a period of 90 seconds (Figure S4). Pre-exposure, OTFTs of both thicknesses demonstrated similar transfer characteristics to low-crystallinity α-ZnPc 400 Å OTFTs, indicating that film thickness between 200–800 Å does not significantly influence the electrical performance of these low-crystallinity OTFTs. Compared to the 200 Å films, XRD spectra shows an expected, 59% more intense α- peak (2θ = 6.92°) for the 800 Å film, as the additional thickness provides additional diffraction signal⁴⁶. The 2D scattering pattern determined by GIWAXS corroborates an α- phase orientation with a highly ordered (100) plane at q = 0.49 Å⁻¹ and a preferential Pc orientation to the surface of ~74°. Exposure to THC vapor showed a similar + ΔV_T, increase in hysteresis, -ΔN, and a decrease in XRD peak intensity for OTFTs of both thicknesses. The OTFTs with the thinnest, 200 Å films, in agreement with other works^47,48, demonstrated a larger decrease in on-current, greater ΔN, and greater -%Δµ, suggesting greater relative changes in film crystallinity, while changes in V_T and hysteresis suggest Pc-THC interactions which are independent of film thickness.

The initially screened thin-film conditions of 400 Å α-ZnPc OTFTs, deposited at a rate of 0.2 Å/s and 25 °C, demonstrate a negligible response when exposed to 40 ppb THC vapor (Figure S5). Comparing the response of the initially screened thin-film conditions to the sensitized response of the thinnest low crystallinity films, we demonstrate a 100x increase in post-exposure sensitivity to THC vapor. Thus, altered surface morphology and semiconductor thickness has a significant effect on OTFT sensitivity.

To examine the effects of THC exposure in real time, we operated low crystallinity α-ZnPc OTFTs at fixed saturation biases while they were continually exposed to 40 ppb THC vapor (Fig. 3). At operating biases of V_SD = −50 V and V_GS = −40 V, continuous THC vapor exposure caused an immediate, sharp, operating current decrease followed by a sustained decrease in operating current at a constant rate, which was mediated by thin-film thickness (Fig. 3c). Pulsed exposure demonstrated a similar effect (Fig. 3d and S6), where the onset of THC vapor induced an immediate decrease in operating current followed by a sustained, thickness mediated, decrease over time. However, after short exposures (~10 seconds), operating current recovered sharply when vapor application was stopped, though not to pre-exposure levels, and subsequent exposures reduced the magnitude of both the operating current decrease and recovery with vapor onset and offset.

The magnitude of the effects of vapor onset, continued exposure, and vapor offset were inversely related to film thickness, where the thinnest, 200 Å films, demonstrated the largest responses, consistent with OTFT sensors made with Cl-AlPc⁴⁷. Additionally, for both pulsed and continual modes of exposure, after 90 seconds, the operating current is consistent with the operating current of films of the same thickness characterized post-exposure at the same bias conditions, suggesting prolonged exposure limits the observed recovery. Both periodic exposures to forced air and continued device operation in air (Fig. 3b), yielded negligible changes in operating current.

The observed, sharp, but semi-reversible, decreases in operating current with THC vapor onset suggests that the introduction of THC immediately causes hole-trapping effects, lowering the operating current. Sustained, irreversible decreases in operating current with continued THC vapor application then suggests that over time, irreversible structural defects occur within the film, permanently lowering operating current, and limiting repeated sensor use. That both the immediate and sustained decreases in operating current are thickness dependant supports these conclusions, as both the amounts of THC relative to Pc within the film and the THC-induced structural defects will be greatest for the thinnest films. Periodic exposure of the 200 Å films to 40 ppb cigarette smoke yields small onset current decreases with smoke application, slight current recovery when smoke application stops, and a slight decrease in current over time (Fig. 3e). However, the magnitude of the operating current decrease was small and the results noisy, indicating that other vaporized organic compounds have different effects on the electrical performance of the substrates, and demonstrating the innate selectivity of ZnPc for this application. Thus, every step of sensor manufacturing impacts device performance, from material selection, deposition conditions, and thin-film thickness.

To analyze the effects of crystal polymorphism on sensing response, we prepared β-ZnPc thin-films by treating 400 Å α-ZnPc thin-films with toluene vapor for 24 h (Fig. 4) resulting in OTFTs with lower peak µ, lower V_T, lower off current, and lower N in comparison to α-ZnPc OTFTs. Additionally, large, µm-scale rectangular crystals are observable by SEM. XRD spectra of the toluene-treated films showed two sharp peaks of equal intensity at 2θ = 7.04 and 9.32°, consistent with reported β-Pc morphology³¹, and a small shoulder peak at 2θ = 6.84°. The azimuthally-integrated GIWAXS pattern of the films is in good agreement with the β-ZnPc pattern predicted by single crystal XRD (Figure S7) confirming complete transition from α- to β- phase. The 2D scattering pattern reveals a highly ordered crystalline thin film with the (100) plane at q = 0.50 and the (10-2) plane at q = 0.68 Å⁻¹ along the q_z axis suggesting multiple molecular orientations however with a preference of the Pc ring aligned ~45° to the surface.

Exposure to 400 ppb THC vapor over a period of 90 seconds resulted in a 10.3 µA increase in peak operating current, a −7.4 V in ΔV_T, and a 4.1 V increase in hysteresis. By XRD, the intensity of the peak at 2θ = 7.04° does not change, while the intensity of the peak at 2θ = 9.32° decreases by 53%. The shoulder peak at 2θ = 6.84° increases in intensity by 20-fold, and a new peak at 2θ = 7.38° appears, indicating morphological changes. Visualized by SEM, previously well-ordered, narrow, rectangular crystals become broad and sheet-like post-exposure (red arrows). By GIWAXS we observe a decrease in relative peak intensity corresponding to the (10-2) plane (q = 0.68 Å⁻¹) and peak splitting at q = ~0.50 Å⁻¹, corresponding to the (100) plane (Figure S7) in β-ZnPc thin films post-exposure to THC vapor. Additionally, the 2D scattering pattern of post-exposure films exhibit a narrowing of the partial arcs and a change in preferential ZnPc orientation to the surface from ~45° to ~74°.

β-ZnPc, despite being highly crystalline, generally demonstrates poor semiconducting performance due to suboptimal charge transport distances caused by close intermolecular spacing and an increased face-on configuration⁴⁹. Thus, the observed increase in operating current as an OTFT “turn-on” response may be a result of partial conversion of β-ZnPc crystals to an α-, or an α-like, polymorph through exposure to THC vapor and suggests the interstitial incorporation of THC into the ZnPc crystallites. Low defect density (N = 5.4•10⁻¹²), extracted from the subthreshold swing, post-exposure suggests the sheet-like structures provide excellent charge transport pathways, with hysteresis possibly resulting from the large interfacial areas between the sheets^22,50. A new XRD peak at 2θ = 7.38° in the post-exposure β-ZnPc films, but not in the post-exposure α- films or the pre-exposure β- films, suggests a tertiary morphology in the THC vapor induced phase transition of β-ZnPc³¹.

Real-time characterization of the β-ZnPc OTFTs with continual 400 ppb THC vapor exposure (Figure S8) demonstrated a turn-off response when operated with V_SD of −50 V and V_GS of −20 V. Like the α-ZnPc OTFTs, with the onset of THC vapor there is an immediate decrease in operating current, however, rather than a subsequent sustained decrease with constant slope, instead there is a rapid decrease in operating current with a decaying slope. As a large V_T shift is observed for the β-ZnPc OTFTs characterized post-exposure, a decrease in operating current is expected at these bias conditions. Inconsistencies in the pre-exposure operating current and vapor onset effects for the β-ZnPc OTFTs, in comparison to the α-ZnPc OTFTs, are likely a result of uneven distribution and orientation of the β-crystals between OTFT electrodes.

With a V_GS bias of −40 V, there is again an immediate decrease in OTFT operating current with the onset of THC vapor with limited change for ~40 seconds, whereupon the operating current begins to fluctuate and trend upwards. Comparing to periodic exposure with the same bias conditions, we observe a similar fluctuating climb in operating current after ~35 seconds of exposure to THC vapor. Fluctuations in operating current suggest that significant µm-scale recrystallization events begin after ~35–40 seconds of exposure, in comparison to the immediate sustained decrease in operating current observed for the α-ZnPc OTFTs. Moreover, sharp decreases in operating current with vapor onset, and recovery with vapor offset, again suggests THC is introducing or acting as a hole-trap within the film. Additionally, where the most crystalline α-ZnPc thin-films demonstrated the least OTFT and XRD changes, highly crystalline β-ZnPc films demonstrated significant re-crystallization with exposure to THC vapor. Thus, the degree of molecular order on the surface is not necessarily indicative of either sensing response or the analyte-induced physical effects. Instead, the lowest crystallinity α-ZnPc OTFTs demonstrated the greatest electrical sensitivity to THC vapor, exhibiting the balance between thin-film morphology, recrystallization, and surface area for intermolecular interactions.

CuPc (copper(II) phthalocyanine), F₁₆-ZnPc (zinc 1,2,3,4,8,9,10,11,15,16,17,18,22,23,24,25-hexadecafluoro phthalocyanine), and tricholoro(octyl)silane were obtained from Sigma Aldrich. F₁₆-CuPc (copper(II) 1,2,3,4,8,9,10,11,15,16,17,18,22,23,24,25-hexadecafluoro phthalocyanine), and ZnPc (zinc phthalocyanine) were obtained from TCI chemicals. F₄-ZnPc was synthesized according to ref. ³⁵ and confirmed by mass spectrometry. All Pcs were purified by train sublimation prior to use. Cannabinoid standards were obtained from Toronto Research Chemicals. All solvents were HPLC grade and purchased from Fischer Scientific.

Si substrates with 230 nm thermally grown SiO₂ dielectric and prepatterned gold source-drain electrodes (W = 2000 μm, L = 10 μm), were purchased from Fraunhofer IPMS and used to fabricate bottom-gate bottom-contact (BGBC) transistors. Si substrates with 300 nm thermally grown SiO₂ were purchased from Ossila and used for fabrication of thin films for XRD and GIWAXS. Wafers were prepared as described in ref. ¹⁸ before being transferred to an Angstrom EvoVac thermal evaporator where deposition rate and thin-film thickness was controlled by quartz crystal microbalance and substrate temperature was controlled by thermocouple.

β-ZnPc films were generated by exposing 400 Å ZnPc films deposited at 25 °C at a rate of 0.2 Å/s to 50 °C toluene vapor in a custom vapor chamber for 24 h followed by a baking at 70 °C for 45 minutes in a vacuum oven.

THC was dissolved in methanol and loaded on to a steel wool frit where it was allowed to dry before being placed in a Volcano Medic vaporizer set to 210 °C temperature. An 8 L balloon was filled completely before being evacuated into a 50 mL volume vapor chamber in which the substrates were placed.

Organic thin film transistors were characterized by applying a voltage bias between gate and source electrodes with BeCu alloy probe tips and the source drain current (I_SD) was measured with a Keithley 2614B on a custom electrical probe station. Mobility was calculated using the following:

Where C_i is the dielectric capacitance, W and L are the width and length of the semiconducting channel, and V_T describes the threshold voltage at which I_DS begins to rapidly increase. To mitigate the effects of bias stress, the gate bias was pulsed at 20 millisecond intervals with an 80 millisecond delay. To saturate the devices, six transfer curves were obtained for each device with the final three averaged to yield a characteristic transfer curve. Defect density (N) was determined with the following:

Where S is the subthreshold slope, as estimated graphically from the transfer data, q is the electronic charge, k is Boltzmann’s constant, and T is temperature.

XRD characterization was performed using a Rigaku Ultima IV powder diffractometer with a Cu-Kα (λ = 1.5418 Å) source. Thin-film substrate measurements were taken with a scan range of 5° < 2θ < 11° at a rate of 0.5° min⁻¹ with a spin rate of 30 rpm.

GIWAXS experiments were performed at the Canadian Light Source (CLS) in Saskatoon, Canada using the Brockhouse Diffraction Sector (BXDS) beamline with a photon energy of 15.1 keV, and the SOLEIL Synchrotron facility in Saint-Aubin, France using the SIRIUS beamline with a photon energy of 10 keV. For data collected at CLS, GIWAXS patterns were collected using a Rayonix MX300 CCD detector (73.242 μm x 73.242 μm pixel size), placed 416 mm from the sample with an angle of incidence of θ = 0.3°. For data collected at SOLEIL, GIWAXS patterns were collected using a PILATUS3 S 1 M detector (172 μm x 172 μm pixel size), placed 330 mm from the sample with an angle of incidence of θ = 0.1°. All GIWAXS data was calibrated against a silver behenate standard and a poly(3-hexylthiophene-2,5-diyl) standard and analyzed using the GIXSGUI software package in MATLAB, where both polarization and solid-angle corrections were applied⁵¹.

SEM measurements were performed with a Tescan Vega II on thin-film Pc devices at 20 kV. AFM measurements were performed with a Bruker Dimension Icon AFM, equipped with ScanAsyst-Air tips, and images were processed with NanoScope Analysis v.1.8. Scans were performed at a rate of 0.814 Hz with multiple sites investigated.

Communications Chemistry thanks Maciej Krzywiecki and the other, anonymous, reviewers for their contribution to the peer review of this work.

All data is available upon request to the corresponding authors.