Spectroscopic analysis of polymer and monolayer MoS₂ interfaces for photodetection applications Free

Two-dimensional transition-metal dichalcogenides (2D TMDs) have been extensively studied for photodetection applications due to their strong optoelectronic behavior in the visible range and tunable bandgap.¹ 2D TMD field effect transistors (FETs) have the advantage of large ON/OFF ratio and small form factor compared to other transistor based photodetectors.² However, the optoelectronic properties of 2D TMDs can deteriorate as a result of surface defects and exposure to air and water.^3,4 Researchers have used a variety of materials to combat this degradation and passivate the surface, most commonly hexagonal boron nitride.^5–7 Polymers are ideal for passivating 2D TMDs in photodetection applications as most are transparent in the visible range, and improvement in photodetection by polymer encapsulation has been previously shown.⁸ 

While the benefits of passivation are well understood, the relative passivation effect of different polymers has not been well defined despite significant study on the mechanisms of passivation.⁹ Researchers in recent years have used polymers to perform significant modulation of the optoelectronic properties of hybrid devices via doping, passivation, dielectric screening, and capacitive effects.^10,11 These schemes illustrate that polymers can contribute multiple mechanisms to the polymer/2D TMD interface. In particular, the ferroelectric polymer polyvinylidene difluoride-trifluoroethylene (PVDF-TrFE) has been used to improve the bandwidth and responsivity of photodetectors in its poled state, but the other effects of the fluoropolymer's properties, including fluorinated structure, high dielectric constant, and even pyroelectricity, on the MoS₂ interface have not been studied in detail.^12,13 As the properties of 2D TMDs can be dramatically influenced by the surface condition, a careful assessment of the effects at the interface, including passivation, are needed to advance photodetector performance. In this work, the 2D TMD and polymer interface is assessed spectroscopically for three common polymers to evaluate the efficacy of each polymer in passivating the MoS₂ surface for photodetection applications. Evaluating this interface with spectroscopic methods, especially with dynamic pump probe measurements like time resolved THz spectroscopy, mimics the processes occurring during photodetection operation and, thus, provides a clearer understanding of the fundamental physics at play. Spectroscopic methods can provide significant qualitative and quantitative information about the chemical and electronic properties of the TMD monolayer. Terahertz spectroscopy has previously shown the ability to probe the conductivity and mobility of semiconductors without the need for contact.¹⁴ As it can be difficult to establish good electrical contact with TMDs, this method is useful to remove any effects from the contact interface on carrier dynamics and electronic characterization.

For this study, a monolayer TMD was chosen rather than multilayer, even though it would provide less signal for many spectroscopic experiments, to simplify the carrier dynamics in the TMD so that the small changes anticipated from passivation could more easily be observed. In multilayer TMDs, the interlayer exciton overwhelms any other photogenerated carriers.¹⁵ These interlayer excitons can be significantly longer lived than other carriers, further obfuscating the carrier dynamics. This study focuses on MoS₂ rather than another TMD because it is more air stable than other TMDs as some of the spectroscopic techniques required prolonged air exposure. To mitigate any degradation of the monolayer due to air exposure, the samples were stored in a nitrogen glovebox when not in use. A representative group of three commonly available polymers were chosen to understand the relative passivation effects: parylene N, polymethyl methacrylate (PMMA), and PVDF-TrFE. Parylene is an extremely inert polymer used frequently in organic electronics and biomedical devices for encapsulation that has previously been shown to passivate graphene.^16–18 Additionally, parylene is specifically engineered to act as a chemical barrier for water and other solvents, which is ideal to prevent degradation of the 2D TMD. The N formulation of parylene (Pa-N) was chosen for this study as it has almost no functional groups. PMMA is a standard electron beam photoresist frequently used to encapsulate 2D TMD devices for passivation and to prevent degradation.¹⁹ PVDF-TrFE is an electroactive polymer that has been shown to be piezoelectric, pyroelectric, and ferroelectric, and as a result, is widely applied for flexible sensors.²⁰ PMMA and PVDF-TrFE have functional groups containing oxygen and fluorine that would be expected to better fix sulfur vacancies or provide doping than Pa-N. Pa-N also has the lowest dielectric constant of the three so should not provide much dielectric screening. This group of polymers then should provide reasonable insight into the mechanisms at the 2D TMD/polymer interface for photodetection applications.

The CVD-grown monolayer MoS₂ was purchased from 2D semiconductors and then approximately 300 nm of each polymer was deposited on top. PVDF-TrFE and PMMA were deposited by spin coating and Pa-N was deposited via physical vapor deposition. The PVDF-TrFE film was not poled so it does not contribute an electric field due to remnant polarization.

The samples were characterized before and after coating with Raman spectroscopy by using a 514 nm laser input. The resulting spectra after coating are shown in Fig. 1(a) and the change in peak positions as a result of coating is shown in Fig. 1(b). Raman spectroscopy was performed both before and after coating to account for any sample variation. No significant change in the location of Raman peaks was observed for samples coated with PVDF-TrFE, but for PMMA and Pa-N, a significant redshift was observed in the $A 1 g$ peak after coating. Previous work has shown that this shift can be attributed to n doping.^21–25 PMMA and Pa-N show a larger change than PVDF-TrFE, which reflects a higher doping level. This result suggests that previously shown device improvements as a result of PMMA or Pa-N encapsulation may have a doping contribution rather than a passivating one. This doping could be a result of the oxygen in PMMA's functional groups or stress on the film as a result of polymer deposition, but further investigation must be done to understand the exact mechanism.

Spatially resolved photoluminescence (PL) spectra were collected by using a hyperspectral PL imaging microscope, where a full spectrum is collected for every pixel in an image. Spectra were collected over a 70  $μ m 2$ area with a Photon etc., hyperspectral microscope using an incident 532 nm 26 mW excitation beam. In addition to areas with intense PL, there were visible grain boundaries and low-intensity spots on the MoS₂ films that we attribute to local variations in the precursor compounds used during the CVD growth.²⁶ In order to characterize just the MoS₂ film, we first divided the polymer sample spectra by the uncoated substrate control spectra to remove background noise, then masked the pixels with the lowest tenth percentile of intensity at the A exciton peak. The total collected area of 70  $μ m 2$ consisted of 512 × 512 pixels, then removing the mask consisting of the lowest tenth percentile intensity pixels resulted in a final pixel count of $2.36 × 10 5$⁠, each with a full spectrum of PL intensities. The spectra were normalized based on the maximum of the A exciton peak at 670 nm. Pixels were binned and averaged to produce a distribution of PL spectra over the whole image area which is shown in Fig. 2(a).

The A exciton peak near 670 nm was not significantly changed as a result of PVDF-TrFE coating. For PMMA and Pa-N coating, the peak is redshifted as shown in the box plot in Fig. 2(b), which is consistent with the doping observed in the Raman spectra. Because the photoluminescence spectra is normalized to the A exciton peak, the B exciton at 625 nm is more prevalent for some samples than others. Additionally, the full width half maximum increases for the PMMA coating as shown in the box plot in Fig. 2(c), which is the opposite effect than what is usually expected for successful passivation.²⁷ These results suggest then that doping rather than passivation is the prominent mechanism at the interface.

The real conductivity measured for all samples corresponds to those obtained in previous work for monolayer MoS₂.^30,31 Averaged values over the frequency range with associated standard deviation are shown in Table I.

The real conductivity of the samples coated with PVDF-TrFE and PMMA was within error of the uncoated sample. The Pa-N and PMMA coated samples show a higher real conductivity than the other measured samples, which corresponds to the n doping effect observed by Raman and photoluminescence spectroscopy. The difference in conductivity between the Pa-N and PMMA samples could be accounted for by variations in the conductivity between the MoS₂ samples before coating as defects in the film can have a large impact on the conductivity.

In TRTS spectra for each sample, we observed a sharp dip in THz transmission immediately following photoexcitation as shown in Fig. 4. In uncoated, Pa-N coated, and PMMA-coated samples, THz transmission recovered nearly to pre-excitation levels after around 5 ps, which is consistent with other reported values under similar conditions.^32,33 Secondary signals near 3 ps delay arise from substrate back reflections. The timing and amplitude of these reflections are dependent on step-scan delay position and substrate thickness which varies from sample to sample. These spurious signals are observed but do not contain any relevant information about carrier conductivity or population. The PVDF-TrFE sample retained a higher THz absorption out to a delay of 20 ps, which is indicative of the continued presence of charge carriers in the MoS₂ film.

The polymer coated samples all show more initial charge generation than the uncoated sample. This finding can be in part attributed to the doping effects shown in Raman and PL spectroscopy. The PVDF-TrFE coated sample has a significantly longer tail, which indicates a longer charge lifetime. There are several possible mechanisms that could be contributing to this result. The additional n doping from the Pa-N and PMMA coating causes an increased carrier concentration, which can encourage the formation of lower mobility trions.^30,34,35 The fluorine content in the PVDF-TrFE could provide superior passivation of sulfur vacancies in the MoS₂ or draw out excess charge to limit recombination and trion formation.

This work assesses the relative passivation effects of common polymers on monolayer MoS₂ to identify what materials can be used to improve photodetection in 2D TMD FETs. While the previous work cites the electric field arising from the remnant polarization of poled PVDF-TrFE is responsible for improved photodetection in 2D TMD FETs, this work suggests that the material properties of PVDF-TrFE contribute to the noted improvement in the carrier lifetimes of photogenerated charge.¹² These longer lifetimes will have a significant positive impact on the performance of an FET photodetector. While more work should be done to understand the exact mechanism at the PVDF-TrFE/2D TMD interface, this finding indicates that PVDF-TrFE and other fluoropolymers could be used to significantly improve photodetection in hybrid 2D TMD photodetectors. Future work will investigate if the findings in this work persist when these polymer coatings are added to multilayer TMDs, TMDs other than MoS₂, and TMD FETs.

D.M.H. gratefully acknowledges the support from the NIST Summer Undergraduate Research Fellowship. C.K.M. gratefully acknowledges the funding from the National Research Council Research Associateship Program. The authors gratefully acknowledge Adam Biacchi for assistance and training on the use of the Raman spectrometer and hyperspectral imaging microscope.

The authors have no conflicts to disclose.

Christine K. McGinn: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Supervision (equal); Writing – original draft (lead); Writing – review & editing (equal). Daniel Michael Harrington: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Software (lead); Writing – original draft (supporting). Edwin J. Heilweil: Data curation (equal); Investigation (equal); Methodology (equal); Writing – review & editing (equal). Christina A. Hacker: Conceptualization (equal); Resources (lead); Supervision (lead); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.