2.1.

Materials

All chemicals [${\mathrm{Zn}}_4\mathrm{O}{(\mathrm{TFA})}_6$ precursor, CAS 1299489-47-6; methacrylic acid (MAA), CAS 79-41-4] were purchased from Sigma-Aldrich and were used without further purification. All the used solvents were reagent grade.

2.2.

Synthesis of Photoresist

MAA (12.0 eq, 0.11 g) and ${\mathrm{Zn}}_4\mathrm{O}{(\mathrm{TFA})}_6$ (1.0 eq, 0.1 g) were dissolved in chloroform (10 to 15 ml) and left stirring for 2.5 h at 40°C. The solvent was then evaporated and the oily residue was precipitated by washing with toluene (5 to 10 ml). The process was repeated for five to six times to remove excess MAA in the reaction mixture and obtain a solid white compound (Fig. 1).

Fig. 1

Scheme of the Zn(MA)(TFA) synthesis by ligand exchange reaction. In the three-dimensional representations, gray tetrahedra represent the coordination geometry of the ${\mathrm{Zn}}^{2+}$ atoms (gray spheres in the center), red spheres represent oxygen atoms, green spheres represent fluorine atoms, and gray sticks are chemical bonds. Lewis formula below gives details on the ligands’ structure.

${}^1\mathrm{H}$ NMR (300 MHz, $\mathrm{DMSO}\text{-}{\mathrm{d}}_6$) $\delta$: 1.83 (3H, $─{\mathrm{CH}}_3$), 5.34 (1H, $═{\mathrm{CH}}_2$), 5.83 (1H, $═{\mathrm{CH}}_2$) ppm

${}^{13}\mathrm{C}$ NMR (300 MHz, $\mathrm{DMSO}\text{-}{\mathrm{d}}_6$) $\delta$: 19.36 ($─{\mathrm{CH}}_3$), 115.04 (s, $-{\mathrm{CF}}_3$), 122.02 ($═{\mathrm{CH}}_2$), 139.93 ($─\mathrm{C}═$), 158 (s, $─\mathrm{COO}$, TFA), 173.00 ($─\mathrm{COO}$, MA) ppm

${}^{19}\mathrm{F}$ NMR (300 MHz, $\mathrm{DMSO}\text{-}{\mathrm{d}}_6$) $\delta$: $-73.94$ ($─{\mathrm{CF}}_3$) ppm

FTIR: 523 ($w$, ${\nu }_{\mathrm{as}}$ Zn-O-Zn), 628 ($s$, COO angle bending), 1205 to 1155 ($m$, $\nu$ C-F), 1238 ($m$, coupled rocking $═{\mathrm{CH}}_2$ and $\upsilon$ $\mathrm{C}\mathrm{─}\mathrm{C}$), 1300 to 1500 ($s$, coupled ${\mathrm{CH}}_x$ deformations and ${\nu }_s$ COO), 1543 ($s$, ${\nu }_{\mathrm{as}}$ COO, bonded acid MA), 1606 ($m$, COO H-bonded MA), 1653 ($s$, ${\nu }_{\mathrm{as}}$ $\mathrm{C}═\mathrm{C}$), 1687 ($s$, ${\nu }_{\mathrm{as}}$ COO bonded TFA), 1735 ($w$, ${\nu }_{\mathrm{as}}$ COOH nonbonded TFA), 2929 to 2987 ($w$, $\nu$ $─{\mathrm{CH}}_3$), 3022 and 3105 ($s$, $\nu$ $═{\mathrm{CH}}_2$), 3200 to 3600 ($b$, ${\nu }_s$ $\mathrm{O}─\mathrm{H}$) ${\mathrm{cm}}^{-1}$.

2.3.

Characterization of the Bulk Material

NMR was recorded using a Bruker AV-400 NMR spectrometer using deuterated dimethylsulfoxide ($\mathrm{DMSO}\text{-}{\mathrm{d}}_6$) as a solvent. FTIR was performed using a Bruker ALPHA FTIR spectrometer. Mass spectra were collected on an AccuTOF LC, JMS-T100LP mass spectrometer (JEOL). Thermogravimetric analysis (TGA) was performed using a NETZSCH thermogravimetric analyzer in an ${\mathrm{Al}}_2{\mathrm{O}}_3$ crucible and heating was performed from 35°C to 700°C at $10\mathrm{K}/\min$ in an ${\mathrm{N}}_2/{\mathrm{O}}_2$ atmosphere. Elemental analysis was performed by Mikroanalytisches Labor Kolbe, Germany for elements Zn, carbon (C), and fluorine (F) in the synthesized compound Zn(MA)(TFA).

2.4.

Thin Film Preparation

Zn(MA)(TFA) resist solution 2% (w/v) was prepared in chloroform (${\mathrm{CHCl}}_3$), and propylene glycol methyl ether acetate (PGMEA), 9:1 v/v followed by filtration using a $0.22\text{-}\mu \mathrm{m}$ PTFE filter after sonication for 4 min. All thin films for analysis and lithography were prepared by spin-coating the resist solution at 2100 rpm, $3000\mathrm{rpm}/\mathrm{s}$ for 30 s. Samples for UV–vis absorption spectroscopy were spin coated on quartz substrates of $525\mu \mathrm{m}$ thickness, and samples for FTIR spectroscopy were spin coated on a double-side polished Si-wafer of $200\mu \mathrm{m}$ thickness. The thickness of the resist thin films spin coated on silicon and quartz substrates was in the range of 25 to 35 nm, as measured by atomic force microscopy (AFM). Postapplication baking was applied $90°\mathrm{C}/30\mathrm{s}$ to remove excess of residual solvent.

Sample preparation for the EUV absorption coefficient measurement was done using silicon nitride (${\mathrm{SiN}}_x$) membranes purchased from Norcada Inc. as substrates (100 nm thickness, membrane window of $7.5\times 7.5{\mathrm{mm}}^2$). A thin film of resist was deposited on the membrane by spin coating following the same parameters as for Si substrates for lithography experiments. The thickness measured by ellipsometry was $43.3\pm 1.4\mathrm{n}\mathrm{m}$.

2.5.

Characterization of Thin Films

UV–vis absorption spectroscopy was performed using a Shimadzu UV2600 spectrophotometer, and FTIR spectroscopy of the thin films was performed in transmission mode under vacuum in a Bruker Vertex 80v spectrometer. Thickness of the thin film spin coated on ${\mathrm{SiN}}_{\mathrm{x}}$ was measured by using J.A. Woollam- VB-400 VASE Ellipsometer. The spectral range used was 250 to 1000 nm. The optical constants of the resist material were first measured independently on a sample of known thickness deposited on a Si substrate.

2.6.

EUV Exposure and Postexposure Analysis

Open frame exposures were performed for a wide range of doses by exposing $1.7\times 1.7{\mathrm{mm}}^2$ (pinhole $70\mu \mathrm{m}$) areas to EUV light at 13.5 nm. These exposures were performed at the SLS XIL-II beamline in the Paul Scherrer Institute (PSI), Switzerland.²⁹ The detail of the experimental set-up at XIL-II beamline for absorption coefficient measurement has been described in previous studies.^30,31 A pinhole of $30\text{-}\mu \mathrm{m}$ diameter and square open frame mask of $0.5\times 0.5{\mathrm{mm}}^2$ were used in this measurement. The absorption of EUV light from ${\mathrm{SiN}}_x$ membrane was calibrated by measuring the photocurrent passing through a clean ${\mathrm{SiN}}_x$ membrane as a reference photocurrent. The transmittance of the resist materials is given by the ratio of measured photocurrent of the resist and the reference photocurrent.

Propionic acid, acetylacetone (acac), and acetic acid diluted in ${\mathrm{CHCl}}_3$ were used as developers. Thin film thickness was measured by AFM, using a Bruker Dimensions Icon. For patterning line-space ($L/S$) patterns, a transmission mask was used. High-resolved SEM images were recorded using FEI Verios 460 system.

3.1.

Determination of the Organic Shell Composition

The Zn oxocluster [Zn(MA)(TFA)] was synthesized by ligand exchange method from the commercially available oxo[hexa(trifluoroacetato)]tetrazinc ${\mathrm{Zn}}_4\mathrm{O}{(\mathrm{TFA})}_6$ (Fig. 1). The starting material comprises a core of four Zn atoms bridged by one O atom (${\mu }_4\text{-}\mathrm{O}$) and six TFA ligands that bridge two Zn atoms through the carboxylate group.^32,33 Since this reaction proceeds in equilibrium due to competitive binding of the two types of carboxylate ligands, MAA is added in excess to favor the shift of the equilibrium to the right side of the chemical equation and have a high abundance of MA ligands in the shell of the synthesized product Zn(MA)(TFA).

NMR and FTIR spectroscopic analysis performed on the synthesized bulk oxocluster evidenced the presence of both TFA and MA ligands (see Sec. 2.2). To confirm whether the tetrameric oxocluster is preserved during ligand exchange, mass spectrometry experiments were performed. This is an ideal technique since Zn metal has different naturally occurring isotopes, which provides the mass spectrum with a unique characteristic isotopic pattern distribution arising from the four Zn atoms in the oxocluster. The spectra obtained using different solvents for the vaporization step are shown in Fig. 2.

Fig. 2

Experimental (black) and simulated (blue) mass spectra of Zn(MA)(TFA) oxocluster from (a) acetonitrile solution and acetonitrile-${\mathrm{d}}_6$ solution (inset) and (b) methanol solution.

The isotopic distribution observed in the mass spectra was in concordance with the expected one for a tetrameric Zn oxocluster (${\mathrm{Zn}}_4\mathrm{O}$). Two peak envelopes in the mass regions $m/z$ 778 to 790 and $m/z$ 806 to 818 were detected, which matched the mass of a ${\mathrm{Zn}}_4\mathrm{O}$ cluster with five MA ligands and the ${\mathrm{Zn}}_4\mathrm{O}$ cluster with four MA and one TFA ligands complexed with two acetonitrile molecules, ${[\mathrm{Zn}{(\mathrm{MA})}_5+2{\mathrm{CH}}_3\mathrm{CN}]}^{+}$ and ${[\mathrm{Zn}{(\mathrm{MA})}_4(\mathrm{TFA})+2{\mathrm{CH}}_3\mathrm{CN}]}^{+}$, respectively [Fig. 2(a)]. Such species would result from the loss of a trifluoroacetate or a methacrylate ligand in the Zn(MA)(TFA) material, as shown in Fig. 2. The complexation of acetonitrile to the cluster was confirmed by recording mass in deuterated acetonitrile-${\mathrm{d}}_3$, which shifted the envelope of peaks with the specific isotopic distribution by $+6$ $m/z$ units [inset in Fig. 2(a)]. Furthermore, when the spectrum was recorded using methanol as the carrying solvent, the molecular peak was observed at 702 $m/z$, corresponding to ${[\mathrm{Zn}{(\mathrm{MA})}_5]}^{+}$ [Fig. 2(b)]. These assigned peaks in the mass spectra indicate that in the synthesized Zn(MA)(TFA), MA is the most abundant ligand in the cluster. It can therefore be concluded that upon ligand exchange reaction most of the TFA ligands in the shell of the precursor cluster were substituted by MA ligands while keeping the Zn oxo core unaltered.

The composition of the organic shell (TFA/MA ratio) was further defined by elemental analysis. The results (experimental % Zn: 23.86, % C: 37.24, and % F: 7.29) were assigned to the chemical formula $[{\mathrm{Zn}}_4{\mathrm{O}}_{18}{\mathrm{C}}_{31}{\mathrm{H}}_{37.5}{\mathrm{F}}_{4.5}]$ (theoretical % Zn 25.03, % C 35.61, and % F 8.19), which would correspond to an average of 7 MA and 1.5 TFA ligands per tetrameric cluster. This assignment indicates the presence of an excess of nonbonded carboxylic acids in the bulk material, in agreement with FTIR. The occlusion of carboxylic acids in the crystals formed by metal oxoclusters is a common phenomenon.^32,34,35

The residue detected experimentally in TGA was 32% (Fig. 3), in good agreement to the calculated ZnO residue that results from the combustion of compound with molecular formula ${\mathrm{Zn}}_4{\mathrm{O}}_{18}{\mathrm{C}}_{31}{\mathrm{H}}_{37.5}{\mathrm{F}}_{4.5}$ (31%).

Fig. 3

TGA of Zn(MA)(TFA) powder sample (bulk material).

3.2.

Stability Studies

Chemical bonds between carboxylate and metal cations are rather labile, i.e., the metal–ligand bond can dissociate at lower energy, especially in the presence of chemical species that can compete in the coordination of the metal, such as water.³⁶ Therefore, it is crucial to study the stability of the newly synthesized Zn-cluster both in the bulk material and when deposited as a thin film for the actual nanolithographic application. For this purpose, the chemical composition of the bulk material and the thin films was monitored under different atmospheres by spectroscopic means.

The stability of the bulk material was studied by FTIR spectroscopy (Fig. 4). A significant decrease in the intensity of O-H band (broad band 3750 to $2800{\mathrm{cm}}^{-1}$) and of the peak assigned to the ${\upsilon }_{\mathrm{as}}$ of nonbonded TFA ($1725{\mathrm{cm}}^{-1}$) was observed in the normalized FTIR spectra after 2 months storage in low moisture conditions. These indicate the loss of free trifluoroacetic acid over time. The broadening of the COO stretching band at $1605{\mathrm{cm}}^{-1}$ also suggests changes in the environment of carboxylates over this period of time. Such changes together with the decrease of the peaks in the 500 to $750{\mathrm{cm}}^{-1}$ region, where bending modes of the COO are expected, suggest that some bonded ligands might be lost and/or change their coordination geometry. This might result from partial hydrolysis and/or rearrangement of the organic shell. Yet, the presence of the characteristic Zn-O-Zn stretching band of the tetrameric oxo core³⁷ indicates that the degradation was only partial.

Fig. 4

(a) Normalized FTIR spectra of freshly synthesized oxocluster Zn(MA)(TFA) and after 2 months and (b) zoom of spectral region from 1780 to $400{\mathrm{cm}}^{-1}$.

In addition to the stability of the bulk material, it was crucial to investigate the molecular structure of the oxoclusters when deposited as a thin film and the effect of different atmospheres present in the lithographic process. The stability of the thin film was monitored at ambient conditions in air, nitrogen atmosphere (glovebox), and high vacuum ($<{10}^{-6}\mathrm{mbar}$). The latter is highly relevant since EUV exposure is performed under high vacuum ($<{10}^{-6}\mathrm{mbar}$), which may act as a driving force for ligands loss and undesired aggregation of the inorganic clusters.^28,38 After exposing the thin films to different conditions, the thin films were analyzed by UV–vis absorption and FTIR spectroscopy.

First, the composition of the thin film right after deposition by spin-coating was investigated (Fig. 5). The FTIR absorption spectrum of Zn(MA)(TFA) deposited as a thin film shows that the extra carboxylic acids initially present in the bulk crystalline material are lost during spin-coating since the bands at $1725{\mathrm{cm}}^{-1}$ (assigned to the COOH asymmetric stretching of the nonbonded TFA) and at $1606{\mathrm{cm}}^{-1}$ (tentatively assigned COOH of MAA bonded to the Zn cluster in a weak manner³⁹) were not observed in the thin film FTIR spectrum. The presence of bonded MA was evidenced by the peak at $1543{\mathrm{cm}}^{-1}$ (COO bidentate asymmetric stretching), the envelope at 1500 to $1300{\mathrm{cm}}^{-1}$ (various chelating and bridging COO stretching modes coupled to asymmetric and symmetric ${\mathrm{CH}}_x$ deformations modes), and the vibrational modes corresponding to ${\nu }_s$ $\mathrm{C}═\mathrm{C}$ at $1653{\mathrm{cm}}^{-1}$ (observed as a shoulder) and the peak at $1238{\mathrm{cm}}^{-1}$ corresponding to coupled rocking $═{\mathrm{CH}}_2$ and $\upsilon$ $\mathrm{C}─\mathrm{C}$.³⁸ The presence of coordinated TFA was indicated by the intense peak $1675{\mathrm{cm}}^{-1}$ (COO asymmetric stretching’s of bonded TFA, which shifts compared to the powder sample, $1687{\mathrm{cm}}^{-1}$) and the characteristic C-F stretching’s at 1155 and $1205{\mathrm{cm}}^{-1}$.

Fig. 5

Normalized FTIR absorption spectra of Zn(MA)(TFA) as powder and deposited as thin film.

The stability of Zn(MA)(TFA) oxocluster thin films can readily be monitored by UV–vis spectroscopy. The $\pi \text{-}\pi *$ electronic transition of the terminal methylene of the MA ligand provides a characteristic signal at $\sim 198\mathrm{nm}$, which can be used to monitor changes in the thin film. The UV–vis spectrum of the freshly spin coated thin film was compared to the spectra for same thin film after certain intervals in air atmosphere [Fig. 6(a)]. The spin coated thin film was stable for at least 4.5 h while bleaching was clearly observed after 24 h. Similarly, the stability was monitored after applying high vacuum ($<{10}^{-6}\mathrm{mbar}$) for 1 h, resulting in a slight decrease in the absorption band and suggesting that vacuum did not induce a considerable loss of ligands. Another measurement was performed after leaving the latter sample for 3.5 h at room temperature, which showed negligible change in the absorption band during this period of time [Fig. 6(b)].

Fig. 6

Monitoring stability of thin film of Zn(MA)(TFA) photoresist by (a), (b) UV–vis absorption spectra and (c), (d) FTIR absorption spectra.

In order to further investigate specific changes in the organic ligands, FTIR of the thin films was recorded. For this purpose, spectra of the freshly prepared sample and of samples after 24 h in different atmospheres were recorded. By comparing the normalized absorption spectra [Fig. 6(c)], a concomitant slight decrease in the intensities of the peaks at $1421{\mathrm{cm}}^{-1}$ (for ${\mathrm{CH}}_2$ deformations combined with COO symmetric stretching’s) and of the small shoulder at $1238{\mathrm{cm}}^{-1}$ (due to rocking of vinylic methylene group $═{\mathrm{CH}}_2$ and $\mathrm{C}─\mathrm{C}$ stretching of neighboring groups)³⁸ is observed for the sample kept in ${\mathrm{N}}_2$. Yet, a more dramatic decrease of these bands and broadening of the peaks at 1675 and $1543{\mathrm{cm}}^{-1}$ was detected for the sample kept at room conditions, accompanied by a relative increase in the broad band at 3100 to $3600{\mathrm{cm}}^{-1}$, typical for O-H stretching [Fig. 6(d)].

These experiments suggest that moisture in the air can lead to a partial hydrolysis of Zn(MA)(TFA) cluster. Also, polymerization of the terminal double bond could also be favored by the natural light available at room conditions and in ${\mathrm{N}}_2$ atmosphere. FTIR spectra of the sample before and after high vacuum (1 h) showed more modest changes. The ratio between the peak associated to the $═{\mathrm{CH}}_2$ rocking in MA ($1238{\mathrm{cm}}^{-1}$) and the peaks assigned to C-F stretching modes in TFA ($1205-1155{\mathrm{cm}}^{-1}$) decreased slightly. Also the intensity of the envelope in the $1500-1300{\mathrm{cm}}^{-1}$ related to MA was lower. In parallel, a slight decrease in the absorption band at 198 nm was observed after 1 h of vacuum. These spectroscopic signs could indicate a small degree of MA ligand loss or cross-linking of the terminal double bonds in the MA ligands. [Fig. 6(d)]. Subsequent monitoring of the same sample after 1 h at room conditions was in agreement with the previous UV–vis spectroscopy studies and indicated that no significant hydrolysis occurred within this time frame.

In light of the spectroscopic changes, the storage of the thin films in different atmospheres for long periods of time could have an important effect on the patterning capabilities of the resist. Such studies were out of the scope of this preliminary work and here we focused on working in time scales that guaranteed the integrity of the material. Nevertheless, it should be noticed that evolution of the thin film in vacuum could have an effect on the lithographic performance if there are idle times before exposure. Such effect could be affecting similar inorganic resists and might be subject of our studies in the future.

3.3.

Sensitivity Toward EUV Light and Lithographic Performance

%Transmittance ($T_x$) of the resist was measured experimentally by using synchrotron EUV light at the Paul Scherrer Institute. The linear absorption coefficient, $\alpha$, for Zn(MA)(TFA) was calculated after determining the thin film thickness by ellipsometry as per Beer–Lambert law as

An $\alpha$ value of $12.4\pm 0.4\mu {\mathrm{m}}^{-1}$ was found for a thin film of Zn(MA)(TFA). This is close to the reported values for tin-based resists, where the $\alpha$ ranges from 15 and $19\mu {\mathrm{m}}^{-1}$, whereas for organic photoresists is typically $4.8\mu {\mathrm{m}}^{-1}$.^17,30,31

As a reference value, the theoretical linear absorptivity was calculated for this material. To do so, we considered that the material was consisting of Zn-tetrameric clusters with 5 MA and 1 TFA ligands (molecular formula ${\mathrm{C}}_{22}{\mathrm{H}}_{25}{\mathrm{O}}_{13}{\mathrm{Zn}}_4{\mathrm{F}}_3$), and we approximated its density ($2.4\mathrm{g}/{\mathrm{cm}}^3$) assuming that the molecular packing was the same as for an analogous tetrameric cluster consisting of six acetate ligands, ${\mathrm{Zn}}_4{(\mathrm{OAc})}_6$ ($1.9\mathrm{g}/{\mathrm{cm}}^3$)⁴⁰ and correcting for the different molecular weights (${\rho }_1/{\mathrm{MW}}_1={\rho }_2/{\mathrm{MW}}_2$). The obtained $\alpha$ value using these assumptions was $14.6\mu {\mathrm{m}}^{-1}$.

In addition to photon absorption and the chemistry triggered by EUV photons, the interaction of the developer with the unexposed and exposed resist largely defines the contrast ($\gamma$) of the lithographic process. Therefore, choosing the right developer is crucial to obtain good $\gamma$ values and understanding the molecular structure of thin films can greatly assist in the choice. Our first attempt was to use ${\mathrm{CHCl}}_3$ as a developer, which is also used as solvent for cluster synthesis and thin film deposition. However, this solvent could not redissolve the thin film resulting from the spin-coating of the material on the silicon substrate. This is further evidence that upon thin film formation, some changes occur in the material compared to its structure when it is isolated as a crystal (most likely loss of excess of nonbonded acids, as seen in FTIR, and potentially other structural rearrangements) such that the solubility in ${\mathrm{CHCl}}_3$ decreases.

Thus, contrast curves were obtained for Zn(MA)(TFA) photoresist by using different developers that could interact more strongly with the Zn-clusters than pure ${\mathrm{CHCl}}_3$ and that could supplement the loss of the nonbonded acids during the thin film deposition or even compete with the existing carboxylate ligands. Diluted solutions (0.05%) of propionic acid, acetylacetate (acac), and acetic acid in ${\mathrm{CHCl}}_3$ were tested. The resists behaved as a negative tone resist in all cases and the three contrast curves in Fig. 7 show low onset values, $D_0$, revealing the high sensitivity of Zn(MA)(TFA) toward EUV photons. Yet, different profiles in the contrast curves were obtained for each developer. The dose to retain most of the film thickness, $D_{100}$, varied significantly as a function of the developer. This behavior results in different contrast values ($\gamma$) for each developer, which is typically defined by the slope of the contrast curve and here is approximated with a linear fitting of the slope (dashed line in Fig. 7).

Fig. 7

Contrast curves of EUV exposed Zn(MA)(TFA). (a) Developed with a diluted solution of different organic chelating agents without PEB; (b) after PEB at 100°C and 180°C. Dashed lines in the graphs represent the linear fitting of the curve slope.

We attribute the differences among developers to distinct interactions between the chelating agents and the cluster. For instance, acetic acid and propionic acid can coordinate with the Zn cation in the cluster through a different binding mode compared to acac due to the different geometry of the binding sites. Diluted acetic acid seems to assist in the dissolution of the material better than diluted propionic acid, but it also interacted with the material on the exposed area, causing its partial dissolution. Thus, the contrast for dilute acetic acid as developer was quite low (${\gamma }_{\text{acetic}}=0.45$), whereas for propionic acid was considerably higher (${\gamma }_{\text{propionic}}=2.95$), although more points at lower doses were needed to give a more accurate contrast value (see below). Further, acac, which is a relatively strong chelating agent, did not yield a good contrast (${\gamma }_{\mathrm{acac}}=0.63$) and gave signs of dissolution of the exposed part, as in the case of acetic acid. Among the tested developers, diluted propionic acid was thus considered a better developer rendering higher $\gamma$ and hence was used for all the further lithography experiments.

The effect of postexposure bake (PEB) at $100°\mathrm{C}/30\mathrm{s}$ and $180°\mathrm{C}/30\mathrm{s}$ was also studied. According to the TGA of the powder samples, the first temperature should not yield any loss of species from the film, whereas at the latter temperature only some desorption of nonbonded trifluoroacetic molecules was expected in the bulk material. Yet, FTIR spectra indicate that these nonbonded acids are not present in the thin film since they are lost during deposition. However, the contrast decreased upon application of PEB at 100°C (${\gamma }_{\mathrm{PEB}100}=0.92$) as compared to the nonthermally treated material (${\gamma }_{\text{propionic}}=2.95$) and was practically lost at higher PEB temperature of 180°C [Fig. 7(b)]. Further spectroscopic analyses to identify the structural changes induced at this temperature need to be performed in order to identify the process induced by the heating but such investigations are outside the scope of the present work.

The lithography performance of Zn(MA)(TFA) was preliminary tested by patterning L/S features using EUV interference lithography. In Fig. 8, selected patterns obtained for 40 and 30 nm half-pitch (HP) with two batches of the material synthesized by following same procedure are shown.

Fig. 8

L/S patterns of (a), (b) 40 nm and (c), (d) 30 nm HP for two batches of synthesized resist.

We observed that the material shows relatively good lithographic performance yet the relation between feature size and dose in the printed patterns differed from batch to batch [Fig. 8]. In these preliminary tests, well defined lines were observed at $37\mathrm{mJ}/{\mathrm{cm}}^2$ for batch 1. However, their linewidths were below the intended 30 and 40 nm HP values. In contrast, batch 2 yielded wider lines at lower doses for the same intended HP values, thus indicating over exposure.

The reproducibility of the contrast curve for these two batches was also studied. Given that the Zn(MA)(TFA) oxocluster had shown high sensitivity in our first tests [Fig. 7(a)], new contrast curves using propionic acid as developer were recorded using smaller dose steps for the lower dose range [Fig. 9(a)]. These new experiments showed that the curves seemed to present some kind of two-step process for both batches, although the origin of this behavior is still not understood. We estimated the contrast values for the steeper part of the two-step slope [dashed lines in Fig. 9(a)].

Fig. 9

(a) Contrast curves of EUV exposed Zn(MA)(TFA) from two different batches (dashed lines in the graphs represent the linear fitting of slope). (b) FTIR spectra of the two batches (bulk powder).

As in the case of the L/S lithographic experiments, the two batches did not yield identical behavior. Both the contrast [see Fig. 9(a)] and the $D_{50}$ values (dose to retain half of the thickness) were different for each batch ($7.5\mathrm{mJ}/{\mathrm{cm}}^2$ for batch 1 and $11.3\mathrm{mJ}/{\mathrm{cm}}^2$ for batch 2). Yet, no correlation between the feature size versus dose and contrast can be concluded at present.

Although the two batches were synthesized following the same procedure and the FTIR spectra of the bulk materials look almost identical, some small discrepancies can be spotted [Fig. 9(b)]. A relevant one is the ratio between the peaks at $1653{\mathrm{cm}}^{-1}$ ($\mathrm{C}═\mathrm{C}$ in MA ligand) and $1687{\mathrm{cm}}^{-1}$ (COO in TFA), which could indicate different ratios of the two types of ligands in the two batches. This different composition of the organic shells could be a source of variation in the printability of the two batches. In addition, we suspect that the dynamic character of the cluster–ligand bonds could introduce further changes in the material during the deposition step. This could result in slight differences in the molecular structure of the thin films, such as further variations in the MA/TFA ratio or different extents of ligand loss/hydrolysis.

We would like to point here that all the mentioned doses are calculated using a tool factor determined by cross-calibration with other resist materials. Therefore, the dose values mentioned here are particularly specific to the calculated EUV-IL tool factor for the specific mask and pin-hole combination and might differ when using a different EUV exposure tool.⁴¹