2.1.1.

Self-developing metal halide resists

In 1978, Broers et al. demonstrated holes of 5 nm in diameter in a 250-nm-thick NaCl film using a focused electron probe in an STEM equipped with a ${\mathrm{LaB}}_6$ filament.²¹ The formation of holes was attributed to sublimation of material, and it required an exposure dose of $0.1\mathrm{C}/{\mathrm{m}}^2$. In 1981, using an intense 0.5 nm probe in a VG HB5 STEM, Isaacson and Muray studied the drilling mechanism in thinner NaCl crystalline films evaporated on carbon support grids.²² They were able to drill holes of 2 nm diameter and fabricate troughs that were 1.5 nm wide. Although impressive, these were not permanent structures due to the attack of NaCl by atmospheric water vapor. However, this study stimulated research in testing many metal fluorides as high-resolution electron beam resists.

In an 80-nm-thick film of ${\mathrm{AlF}}_3$, Muray et al. found that an electron dose of $\sim 10\mathrm{C}/{\mathrm{cm}}^2$ was required to remove aluminum and fluorine from the irradiated areas.²³ Using ${\mathrm{AlF}}_3$ film as an etch mask against reactive ion etching in a ${\mathrm{CHF}}_3$ plasma on a 50-nm-thick silicon nitride membrane and subsequent removal of ${\mathrm{AlF}}_3$ with an aqueous solution of HCl, patterns with dimensions of $8\times 10\mathrm{nm}$ were obtained. Since the resist required no further development after the exposure, it was termed as a “self-developing resist.” Further studies on ${\mathrm{AlF}}_3$ resist demonstrated both positive- and negative-tone behavior.²⁴ Interestingly, it was found that in the irradiated areas, metallic aluminum was formed at doses as low as $\sim 2\mathrm{C}/{\mathrm{cm}}^2$. Aluminum wires about 20 nm wide were produced by subsequently dissolving the resist in water.

Different metal halides demonstrate different damage mechanisms when irradiated with an intense electron probe [Figs. 1(a) and 1(b)]. Kratschmer and Isaacson investigated a host of metal fluorides such as ${\mathrm{AlF}}_3$, ${\mathrm{FeF}}_2$, ${\mathrm{BaF}}_2$, ${\mathrm{SrF}}_2$, ${\mathrm{LaF}}_3$, ${\mathrm{CrF}}_2$, and CsF for both positive- and negative-tone behavior.²⁷ Independently, using an STEM, Scherer and Craighead also showed that ${\mathrm{BaF}}_2$ and ${\mathrm{SrF}}_2$ films could indeed act as negative-tone resists.²⁸ Using in situ electron energy loss analysis, they observed that both ${\mathrm{BaF}}_2$ and ${\mathrm{SrF}}_2$ show a decline in the fluorine K-edge signal during the irradiation, with a simultaneous increase in the oxygen K-edge intensity—the source of oxygen being the vacuum of the microscope chamber. Feature sizes of about 100 nm were achieved on GaAs, with grain size of the resist film acting as a limit factor to achieve higher resolution. Scherer et al. attempted to optimize resolution of the negative fluoride resists by grain size control, which was achieved by alloying ${\mathrm{SrF}}_2$ film with 8% ${\mathrm{AlF}}_3$.²⁹ Other group II metal fluorides such as ${\mathrm{CaF}}_2$ and ${\mathrm{MgF}}_2$ were used as positive resists and developed in water or other suitable agents to chemically strip radiolysis-induced products. Mankiewich et al. achieved 30 nm resolution with ${\mathrm{CaF}}_2$ inorganic resist.³⁰ When exposed inside an STEM (pressure $\sim 5\times {10}^{-7}\mathrm{Torr}$), radiolysis of ${\mathrm{CaF}}_2$ led to the formation of CaO as the end product. They suggested that electron beam-induced radiolysis of ${\mathrm{CaF}}_2$ leads to the formation of Ca metal first, which rapidly oxidizes to CaO by residual oxygen inside the chamber. This chemical conversion required a total dose of $10\mathrm{C}/{\mathrm{cm}}^2$. Water was used as a developer because CaO is about hundred times more soluble in water than ${\mathrm{CaF}}_2$. Zanetti et al. continued the work on ${\mathrm{CaF}}_2$ as an etch mask.³¹ Instead of a simple thermal evaporation, they used molecular beam epitaxy to deposit ${\mathrm{CaF}}_2$ as a resist on GaAs. Epitaxial ${\mathrm{CaF}}_2$ films were attained due to a good lattice match between them. Electron energy loss spectroscopy (EELS) studies on crystalline and amorphous ${\mathrm{FeF}}_3$ by Saifullah et al. showed a decline in the fluorine K-edge signal during electron irradiation, with the F:Fe ratio settling to a value of $\sim 1.3$ [Figs. 1(c) and 1(d)].²⁵

Fig. 1

Drilling curves are transmitted current versus time curves that indicate the mass loss behavior of metal halides and oxides during the exposure to an electron beam. (a) Type I: Here, the gas bubbles form in the irradiated volume, coalesce together, and they “pop” to form a hole after exposure to sufficient electron irradiation. Metal cations are displaced from the irradiated volume to surround the hole. Drilling time shows a nearly linear behavior with accelerating voltage and is relatively temperature and thickness independent. Examples include many amorphous halides and oxides including ${\mathrm{AlF}}_3$, ${\mathrm{FeF}}_3$, and ${\mathrm{Al}}_2{\mathrm{O}}_3$. (b) Type II: Here continuous mass loss is observed during drilling. Material may be lost from both surfaces of the specimen. Drilling time shows highly nonlinear accelerating voltage dependence, but mass loss is accelerated at higher temperatures. Examples are mainly crystalline halides and oxides including ${\mathrm{CaF}}_2$, ${\mathrm{TiO}}_2$, ${\mathrm{FeF}}_3$, NaCl, MgO, and sodium $\beta$-alumina. (c) Bright-field and (d) annular dark field images of $\sim 4\mathrm{nm}$ lines and $\sim 8$ to 10 nm wide holes machined in crystalline ${\mathrm{FeF}}_3$ inside a VG 501HB STEM. Notice that the metal pushed to the walls of the holes provides a darker contrast in bright-field and vice versa (from Ref. 25). (e) Encyclopaedia Britannica on a pinhead. A portion of Encyclopaedia written in amorphous ${\mathrm{AlF}}_3$; each hole corresponds to 4 nm in diameter (from Ref. 26). The figure reproduces a part of page 706 of Encyclopaedia Britannica referring to Chadwick (Micropaedia, Vol. II, Bibai to Coleman. 15th edition, Helen Hemingway Benton: Chicago and London, 1979).

The return of FEG-SEM instead of STEM to perform sub-10 nm electron beam lithography marked an important milestone in self-developing metal halide resists. Taking a cue from the earlier work by Langheinrich and Beneking who demonstrated $\sim 3$ to 4 nm resolution in a STEM using ${\mathrm{AlF}}_3$-doped LiF inorganic resist;³² in 1995, Fujita et al. using the same resist, demonstrated 5-nm linewidth patterns with 60 nm periodicity on a silicon substrate using a 30-kV electron beam inside an FEG-SEM.³³ Electron stimulated desorption studies suggested the formation of a metal-rich layer by preferential desorption of fluorine. This process suppresses further desorption of fluorine. Self-developing reaction in this resist is achieved by the surface diffusion of the residual metal film.³⁴

Despite their very high resolution, metal halides as electron beam resists suffered from couple of disadvantages—their high hygroscopicity and very high dose requirement to either drill a hole or expose the resist. These drawbacks were somewhat alleviated using metal oxides as high-resolution resists, on which we will focus our attention.

2.1.2.

Self-developing metal oxide resists

Using a highly intense focused electron beam in a VG HB5 STEM, Mochel et al. were able to drill 2-nm holes and cut 2-nm lines through a 200-nm-thick crystalline sample of sodium $\beta$-alumina in a few seconds.³⁵ This was a permanent hole drilling as opposed to what was observed in many metal halides where the holes got “healed” when exposed to the ambient. Furthermore, a stable partially formed hole could be observed if the electron irradiation was stopped before the hole was completely drilled. The hole drilling phenomenon was followed by monitoring the transmitted current versus time. It was observed that the initial constant mass loss was followed by a decrease in the mass loss rate, ending with the formation of a hole [Fig. 1(b)]. The hole drilling process in crystalline sodium $\beta$-alumina was further studied by Berger et al. who used EELS and energy filtered imaging, together with transmitted current versus time curves.³⁶ Using the information obtained from these techniques, they suggested that electron irradiation resulted in the removal of material atom plane by atom plane from both the surfaces during hole drilling. The incident beam also led to ionization, which resulted in desorption of oxygen, with aluminum migrating to the sides of the hole. Devenish et al. used a standard thermionic source in a conventional TEM to drill holes in sodium $\beta$-alumina.³⁷ Salisbury et al. coined the acronym SCRIBE (subnanometer cutting and ruling by an intense beam of electrons) to differentiate the process of hole drilling or “machining” in inorganic materials from conventional electron beam lithography using organic resists.³⁸

Among metal oxides, amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ was perhaps the most popular material that was studied using SCRIBE. Studies by Mochel et al.³⁵ and Berger et al.³⁶ demonstrated that hole drilling at a nanoscale in an anodized amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ takes place differently from what was observed in crystalline sodium $\beta$-alumina. The transmitted current versus time showed a quick increase in current followed by a plateau region that increases slowly until the current jumps to a final value [Fig 1(a)]. In other words, amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ undergoes a small but quick mass loss initially followed by a pause during which no mass loss occurs. Finally, an abrupt transition to a hole takes place. Interestingly, it was observed that if the beam was switched off in the plateau region, then the partially drilled hole healed up quickly with no observable sign of the incomplete hole. Using a conventional tungsten thermionic source in a conventional TEM, therefore at a lower current density, Devenish et al. attempted to drill amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$.³⁷ It was observed that at room temperature, the ${\mathrm{Al}}_2{\mathrm{O}}_3$ film suffered some damage, but no holes could be drilled. However, at liquid nitrogen temperature, holes were easily drilled. A possible explanation for this observation is that at low temperatures, gases such as water vapor and oxygen in the TEM column may be condensed or adsorbed on amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$. The electron beam ionizes the adsorbed species, which then participate as reactive ions to boost the hole drilling. Just like what was noticed earlier for anodized amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$, some partially drilled holes were also observed to heal up in amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$. It was suggested that the drilling process was a balance between outward electron-beam-induced migration and inward back diffusion of aluminum back, which tends to fill up the hole. Lowering the temperature should decrease the rate of back diffusion of aluminum and hence increase the drilling rate.

Studies by Hollenbach and Buchanan showed that radio frequency sputtered thin films of amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ in a VG HB5 STEM showed significant increase in sensitivity when compared with anodized ${\mathrm{Al}}_2{\mathrm{O}}_3$ films.³⁹ The hole formation time in a 90-nm-thick sputtered amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ could be reduced to 50 ms (rather than tens of seconds), and dose requirement for 5 nm wide holes was determined to be $2.5\times {10}^3\mathrm{C}/{\mathrm{cm}}^2$. The reduction in dose requirement was attributed to the microstructural characteristics and incorporation of a relatively larger atom of argon acting as a structural modifier in the sputtered film. Thus, the faster rate of removal of material from the irradiated area could be due to the more open structure of the amorphous material, leading to increased diffusion rates of atoms. Similar hole drilling results were also obtained for sputter-deposited amorphous films of ${\mathrm{Y}}_2{\mathrm{O}}_3$, ${\mathrm{Sc}}_2{\mathrm{O}}_3$, ${\mathrm{MgO}·\mathrm{Al}}_2{\mathrm{O}}_3$, and $3{\mathrm{Al}}_2{\mathrm{O}}_3{·2\mathrm{SiO}}_2$. Morgan et al. investigated this material further and found that electron irradiation with a high magnification raster ($>500\mathrm{k}$) causes the holes to heal up.⁴⁰ Interestingly, when the holes were drilled closer together than the critical separation, a proximity effect was observed causing the holes to become pear-shaped. On the contrary, crystalline ${\mathrm{Al}}_2{\mathrm{O}}_3$ did not show these effects.⁴¹

While studying the SCRIBE process in MgO smoke cubes, Salisbury et al. found that it is possible to “machine” holes and steps on the nanometer scale on the surface of MgO with {100} faces oriented parallel to the electron beam.³⁸ Further investigations by Turner et al.⁴² and Devenish et al.⁴³ have demonstrated the formation of atomically smooth surface faces when a finely focused beam is rastered over the surface of a MgO crystal. Energy dispersive x-ray spectroscopy during “machining” showed a constant Mg:O ratio suggesting that the material is removed in stoichiometric groups.

Berger et al. observed that hole drilling in ${\mathrm{TiO}}_2$, TiO, and ${\mathrm{Ti}}_2{\mathrm{O}}_3$ was similar to that seen in metal $\beta$-aluminas—a gradual loss of oxygen and displacement of titanium under the beam.⁴⁴ However, the holes did not penetrate through the films and the hole drilling stopped when a critical Ti:O ratio was reached. A comparative study of “machining” of amorphous and crystalline ${\mathrm{TiO}}_2$ films using a finely focused beam of electrons by Saifullah et al. showed a surprising result—the latter more amenable to hole drilling than the former.⁴⁵ Time-resolved EELS studies showed that both amorphous and crystalline ${\mathrm{TiO}}_2$ films lose oxygen continuously during the exposure to a finely focused electron beam with the final composition settling close to that of TiO.

${\mathrm{SiO}}_2$, although not exactly a metal oxide and slightly less ionic than other oxides discussed so far, is also amenable to the SCRIBE process. Using windowless energy dispersive x-ray spectroscopy and EELS studies, Chen et al.⁴⁶ and later Saifullah et al.⁴⁷ showed that electron beam evaporated self-supporting thin films of ${\mathrm{SiO}}_2$ can be directly reduced to silicon with an intense 100 kV electron probe inside a VG 501HB STEM. Columns of silicon as small as 2 nm wide were fabricated at an electron dose $>3.0\times {10}^5\mathrm{C}/{\mathrm{cm}}^2$. During the exposure, both silicon and oxygen are lost, the former more slowly than the latter. Using an intense electron probe in FEG-SEM, Fujita et al. exposed a thermally grown ${\mathrm{SiO}}_2$ on an Si (111) substrate.⁴⁸ This resulted in the formation of SiO, which was selectively thermally desorbed inside an ultrahigh vacuum chamber to produce 10-nm-wide open windows. Pattern transfer produced Si wires of 10 nm width.

Other oxides studied for electron beam-induced hole drilling were ZnO crystals,³⁸ ${\mathrm{MoO}}_3$, and ${\mathrm{YBa}}_2{\mathrm{Cu}}_3{\mathrm{O}}_7$.³⁷ Hole drilling in ZnO crystals gave rise to faceted holes, but the indentation formed equally at both entrance and exit surfaces. In ${\mathrm{YBa}}_2{\mathrm{Cu}}_3{\mathrm{O}}_7$, electron irradiation gave rise to the loss of oxygen, thus transforming it into a semiconductor. Devenish et al. suggested that Josephson junctions could be potentially produced on a nanoscale using the SCRIBE technique.³⁷ Pauza et al. used electrons to create a narrow damaged section on ${\mathrm{YBa}}_2{\mathrm{Cu}}_3{\mathrm{O}}_7$ film and showed that weak links can be written directly on this material.⁴⁹

With the rare exception of a few studies, almost all the work on metal oxides (and metal halides) as self-developing electron beam resists were carried out using free-standing thin films or nanoscale particles of oxides inside a dedicated STEM. To gain a better understanding of the electron beam exposure behavior of self-developing inorganic resists on a substrate, Saifullah et al. conducted systematic studies using sputtered ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist on a silicon substrate inside an STEM.⁵⁰ By monitoring the changes taking place in the resist as a function of electron dose using EELS spectra, they noticed that even though both aluminum and oxygen are lost from the irradiated area, a very thin film of consisting of aluminum and oxygen remains. Even a prolonged exposure did not remove this film. Amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ is resistant to reactive ion etching in fluorine-containing plasmas. Presence of small amount of amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ on the substrate can result in the lack of pattern transfer.

2.1.3.

Electron beam damage mechanisms in inorganic resists based on metal halides and metal oxides

Electron beam damage mechanisms involved during the SCRIBE process of inorganic resists based on metal halides and metal oxides are varied with different mechanisms dominant in different materials (Fig. 2). Here, we summarize them.

Fig. 2

Schematic of various mechanisms of permanent displacement of atoms under an electron beam. (a) Radiolysis and (b) knock-on displacement. $E_{\mathrm{th}}$ and $E_{\mathrm{th}*}$ are threshold energies for radiolysis and knock-on displacement, respectively. (c) Knotek–Feibelman mechanism. Here, the interatomic Auger decay of a maximal valency ionic solid (${\mathrm{TiO}}_2$) induces desorption. In Auger events, electrons are always removed from anion species resulting in highly repulsive final states and subsequent desorption.

Mochel et al. proposed that in crystalline sodium $\beta$-alumina, oxygen was desorbed via the Knotek–Feibelman mechanism, and aluminum migrated to the sides of the hole during “machining” with a focused beam of electrons.³⁵ This model was also reiterated by Humphreys et al. for hole drilling in ${\mathrm{Al}}_2{\mathrm{O}}_3$, where 2p, the highest occupied level of the ${\mathrm{Al}}^{3+}$ ion, is deep enough to provide energy to remove one or two electrons from ${\mathrm{O}}^{2-}$.²⁶

2.1.4.

Outlook on inorganic resists based on metal halides and metal oxides

In Sec. 2.1, we noted that electron beam lithography using inorganic resists was carried out using a dedicated STEM and membrane–film substrate. It successfully demonstrated single digit nanometer resolution as small as 2 nm. Utilizing the high resolution offered by ${\mathrm{AlF}}_3$ resist and taking a cue from the question posed by Richard Feynman “Why cannot we write the entire 24 volumes of the Encyclopaedia Britannica on the head of a pin?” in his paper titled “There’s Plenty of Room at the Bottom,”⁵⁶ Humphreys et al. demonstrated that the writing of contents of the Encyclopaedia Britannica on a pinhead is indeed possible.²⁶ They also pointed out that if the letters are represented by a code of dots and dashes in patterned ${\mathrm{AlF}}_3$ resist then the contents of the Encyclopaedia would fit on a pinhead with plenty of room to spare [Fig. 1(e)]. Another advantage of inorganic resists is that they are usually self-developing and require little or no chemical development post-exposure. However, they suffered from a steeper requirement of electron dose, especially when they were patterned on a solid substrate.⁵⁰ Moreover, the use of membrane–film substrate is not ideal for device fabrication due to its fragile nature. In contrast to using a dedicated STEM for lithography, both FEG-SEM and conventional electron beam lithography machine offer a more practical alternative for studying achievable resolution in resists and device fabrication on solid substrates. Amorphous oxides such as ${\mathrm{Fe}}_2{\mathrm{O}}_3$ and ${\mathrm{WO}}_3$ deposited on silicon substrates were investigated for their suitability as electron beam resists as well as etch masks. In the case of ${\mathrm{Fe}}_2{\mathrm{O}}_3$, it undergoes amorphous to crystalline transition under an electron beam, which results in change in its solubility in HCl.⁵⁷ On the other hand, amorphous ${\mathrm{WO}}_3$ films showed irradiation-induced chemical change and an amorphous to crystalline transition at lower and higher voltages, respectively, both affecting the change in solubility of exposed region in basic solutions.^58,59 Although the dose requirement for exposure was lowered by some orders of magnitude, there was a noticeable increase in the LER due to the amorphous to crystalline transition.

2.3.1.

Spin-coatable metal oxide resists

Stabilized metal alkoxide-based resists

Metal alkoxides are popular precursors for the development of electron beam-sensitive spin-coatable metal oxide resists. These are reactive compounds due to the presence of electronegative alkoxy groups, thus making the metal atom vulnerable to nucleophilic attack. Furthermore, alkoxides are highly unstable in an ambient atmosphere and susceptible to hydrolysis resulting in the formation of hydrated metal oxides alkoxides. This makes them unsuitable for handling in a lab atmosphere. However, their hydrolytic activity can be carefully tailored via chelation with $\beta$-diketones (e.g., acetylacetone, benzoylacetone) and $\beta$-ketoesters (e.g., methylacetoacetate, ethylacetoacetate). Both $\beta$-diketones and $\beta$-ketoesters undergo keto-enol tautomerism – the enol form is stabilized by chelation with metal alkoxides. This reduces the hydrolytic activity, most likely due to steric hindrance, and enables their handling under lab conditions. Due to high solubility of chelated metal alkoxides in organic solvents, this allows them to be spin-coated on a substrate in a conventional manner.

Aluminum tert-butoxide chelated with acetylacetone and ethylacetoacetate was the first metal alkoxide used as an electron beam-sensitive negative resist [Fig. 3(a)]. Chelation produced the gel that, when spin-coated on a substrate, gave ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist.^97,103 Its sensitivity was $\sim 8\mathrm{mC}/{\mathrm{cm}}^2$ (at 80 kV) and exhibited $>{10}^6$ times more electron beam-sensitivity than sputtered amorphous ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist films.¹⁰⁴ For the first time, the sensitivity of a metal oxide resist was demonstrated to be very close to the conventional electron beam resists such as calixarene and HSQ.^105,106 Furthermore, it was shown that the sensitivity of spin-coatable ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist was tailorable depending upon the chelating agent that was employed to stabilize the alkoxide.¹⁰⁷ The weaker the ability of a chelating agent to bind with aluminum alkoxide, the higher is the electron beam sensitivity of the corresponding ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist. Due to its small molecular size, ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist showed single digit nanoscale resolution (8 nm) in the center of a $500\mu \mathrm{m}\times 500\mu \mathrm{m}$ main field and $\sim 10\mathrm{nm}$ at its corners [Fig. 3(b)].⁹⁹ Since the lift-off step was absent, the patterned structure was mainly ${\mathrm{Al}}_2{\mathrm{O}}_3$, a functional material, which could be used as a mask to etch silicon.¹⁰³

Fig. 3

Preparation and electron beam lithography of spin-coatable metal oxide resists. (a) Both $\beta$-diketones ($\mathrm{R}={\mathrm{CH}}_3$ for acetylacetone) and $\beta$-ketoesters ($\mathrm{R}={\mathrm{OC}}_2{\mathrm{H}}_5$ for ethylacetoacetate) undergo keto-enol tautomerism and the enol form is stabilized by reaction with a metal alkoxide. As an example, aluminum tri-sec-butoxide, $\mathrm{Al}{({\mathrm{OBu}}^{\mathrm{s}})}_3$, forms a bidentate molecular complex that is air stable and electron beam-sensitive. Single digit nanometer scale electron beam lithography of (b) ${\mathrm{Al}}_2{\mathrm{O}}_3$ (from Ref. 99), (c) ${\mathrm{TiO}}_2$ (from Ref. 100), (d) ${\mathrm{ZrO}}_2$ (from Ref. 101), and (e) ${\mathrm{HfO}}_2$ using their respective resists (from Ref. 102). The resists from (c), (d) were prepared by reacting the respective alkoxides with benzoyl acetone. (f) Sub-10 nm dense features patterned using ${\mathrm{TiO}}_2$ resist (from Ref. 100).

The chemistry of spin-coatable ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist was used as a model to further develop resists of ${\mathrm{TiO}}_2$,¹⁰⁰ ${\mathrm{ZrO}}_2$,¹⁰¹ ${\mathrm{HfO}}_2$,¹⁰² ${\mathrm{PbTiO}}_3$, lead zirconium titanate (PZT), yttrium iron garnet (YIG), among others [Figs. 3(c)–3(f)].¹⁰⁸ Unsurprisingly, all these resists showed single digit nanometer resolution—out of which ${\mathrm{TiO}}_2$ and ${\mathrm{ZrO}}_2$ resists stand out for their low LER in sub-10 nm patterns and showed little deviation of the patterned structures from the designed widths. With further tweaking of sol–gel chemistry, resists for ${\mathrm{Al}}_2{\mathrm{O}}_3$,¹⁰⁹ ${\mathrm{TiO}}_2$,¹¹⁰ ${\mathrm{GeO}}_2$,¹¹¹ PZT¹¹² among others were developed that are amenable to not only electron beam but also photon-based lithographies.¹⁰⁹ Both ${\mathrm{Al}}_2{\mathrm{O}}_3$ and ${\mathrm{TiO}}_2$ resists used their respective alkoxides along with a phenylsilane to enable condensation of the inorganic network when exposed to an electron beam.^109,110 On the other hand, ${\mathrm{GeO}}_2$ and PZT utilized their respective alkoxides with epoxy¹¹¹ and methacrylate groups,¹¹² respectively, to achieve patterning.

Electron beam patterning of a metal oxide using its respective metal alkoxide without a chelating agent is indeed possible but its deposition on a wafer is very cumbersome. Mitchell and Hu performed electron beam lithography on the condensed titanium iso-propoxide films to obtained ${\mathrm{TiO}}_2$ patterns.^113–115 The deposition of titanium iso-propoxide required a high vacuum chamber and cooling of the wafer. Titanium iso-propoxide showed a high sensitivity to an electron beam, which is comparable to the conventional resists such as PMMA.

It was mentioned earlier that the chelation of metal alkoxides with $\beta$-diketones and $\beta$-ketoesters leads to the hydrolytic stability of spin-coatable metal oxide resists. With respect to chemical bonding, the metal atom is bound to $\beta$-diketones and $\beta$-ketoesters in a bidentate fashion. When the resist is exposed to a beam of energetic electrons, the peaks associated with $\mathrm{\nu }(\mathrm{C}═\mathrm{O})$ and $\mathrm{\nu }(\mathrm{C}═\mathrm{C})$ vibrations in the chelate ring decrease in intensity with an increasing electron dose [Fig. 4(a)]. This suggests rapid radiolysis of the chelate ring, breakdown of the organic components, and their removal via the vacuum system of the electron beam writer [Fig. 4(b)]. In the FTIR spectra, with an increasing electron dose, this leads to bleaching of the bands associated with the original chelating organic molecules of the resist. The breakdown of organic components in the resist makes it insoluble in organic solvents such as alcohols and acetone, thus giving a negative-tone behavior.¹⁰³

Fig. 4

(a) Electron beam exposure mechanism in spin-coatable ${\mathrm{Al}}_2{\mathrm{O}}_3$ resist. Exposure to a beam of energetic electrons result in the breakdown of chelate bonds that makes the resist insoluble in organic solvents such as acetone. (b) Proposed model of the exposure mechanism (from Ref. 103).

In the case of metal oxide resists where an alkoxide and a phenyl silane was used, exposure of the resist films to electrons and photons (UV or x-ray) result in degradation of the organic components and condensation of the inorganic network.^109,110 This process makes the exposed resist insoluble in organic solvents.

Metal carboxylate-based resists

Carboxylate salts of metals, especially metal naphthenates and neodecanoates (many of them popularly used as wood preservatives!), have been used as negative electron beam resists to pattern metal oxides via a post-heat treatment step. These are stable viscous liquids at room temperature. They are excellent candidates for high-resolution electron beam lithography. ${\mathrm{YBa}}_2{\mathrm{Cu}}_3{\mathrm{O}}_7$ was the earliest oxide that was patterned using a mixture of metal naphthenates.¹¹⁶ Several subsequent reports thereafter utilized submicron direct-write capabilities of metal naphthenates,¹¹⁷ octylates,¹¹⁸ and ethylhexanoates^119,120 to fabricated ferroelectric oxide patterns. Kiyohara et al. exploited the high etch resistance of the electron beam direct-write patterned metal octylates for pattern transfer into CVD diamond films.¹²¹ Saifullah et al. showed that zinc naphthenate was capable of single digit nanometer resolution (7 nm) with an aspect ratio of $\sim 9$ over the entire $500\mu \mathrm{m}\times 500\mu \mathrm{m}$ main field [Fig 5(a)].^122,123 At 100 kV, zinc naphthenate showed a sensitivity of $\sim 15\mathrm{mC}/{\mathrm{cm}}^2$ and contrast of $\sim 3.3$. These nanopatterns demonstrated very low LER ($3\sigma =2.8\mathrm{nm}$). Heat-treatment of these patterns gave $\sim 5\mathrm{nm}$ wide ZnO lines [Fig. 5(b)]. Jones et al. fabricated nanoscale field-effect transistors of ZnO using a zinc neodecanoate resist.¹²⁴ Comparing zinc naphthenate and zinc neodecanoate as resists for fabricating ZnO nanowires and FETs, Tiwale et al. demonstrated that both the precursors gave similar electronic properties.¹²⁵

Fig. 5

(a) Electron beam lithography of zinc naphthenate resist showing sub-10 nm linewidths in the center as well as at the edges of the $500\mu \mathrm{m}\times 500\mu \mathrm{m}$ main deflection field. The aspect ratio of these lines is $\sim 9$. A Leica VB6-UHR nanowriter was used to pattern them. These lines show an LER ($3\sigma$) of about 2.8 nm. This result demonstrates that with a right combination of patterning tool and resist, it is indeed possible to fabricate sub-10 nm features over the entire wafer. (b) Patterned resist was converted to ZnO by heat-treatment at 500°C for 1 h. The resulted in the reduction of linewidth to $\sim 5\mathrm{nm}$ (from Ref. 122).

Electron beam lithography of iron naphthenate and the post-heat treatment of its patterns gave ${\mathrm{Fe}}_2{\mathrm{O}}_3$, which was used as a catalyst to grow carbon nanotubes.¹²⁶ Likewise, nickel and copper naphthenates can also be patterned and heat-treated in air or a reducing atmosphere to give either their oxides or metals, respectively. Using the latter technique, Nedelcu et al. were able to obtain 8- and 4-nm wide lines of nickel and copper, respectively.¹²⁷ They also noted the low LER of the patterned structures. With the help of controlled annealing in pure hydrogen atmosphere, Bi et al. were able to achieve conversion of patterned copper naphthenate into graphene@copper composite.¹²⁸ Using condensed copper(II) pentafluoropropionate on a substrate, Berger et al. were able to pattern CuO, which on further annealing in a forming gas (5% ${\mathrm{H}}_2$ in ${\mathrm{N}}_2$) gave Cu structures.¹²⁹ Bhuvana et al. adopted a slightly different approach where they reacted palladium acetate with hexadecylthiol in an equimolar ratio to produce palladium acetate hexadecylthiolate. This behaved as a negative resist due to cross-linking of carbon chains when exposed to an electron beam. Heating the patterned resist at $\sim 240°\mathrm{C}$ in air gave 30 nm wide lines of palladium metal.¹³⁰

It was observed that reducing the size of the carboxylate group to an acetate led to the formation of metal oxide straightaway when the metal acetate was exposed to an electron beam. Stark et al. had initially reported patterning of palladium acetate by an electron beam exposure and subsequent development in chloroform. The solubility change due to exposure was attributed to the decomposition of palladium acetate leading to densification.¹³¹ Chaker et al. sublimated zinc acetate onto a silicon substrate in vacuum. Exposing the thin film of zinc acetate to a beam of electrons led to the direct formation of 12-nm wide ZnO lines.¹³² Similar electron beam exposure behavior was also observed for organotin acetate-based resists.¹³³

Electron beam exposure study of a naphthenate resist using infrared absorption suggests cross-linking in the resist thus increasing the molecular weight and making it insoluble in organic solvents such as toluene. The result of this chemical transformation is negative-tone patterning.¹¹⁷ It is believed that other metal carboxylates such as metal neodecanoates, octylates, ethylhexanoates, and palladium acetate hexadecylthiolate also undergo similar cross-linking to show negative-tone behavior. On the other hand, resists based on lower molecular weight carboxylate group such as zinc acetate undergo radiolysis leading to the removal of volatile organics and formation of ZnO under the beam.¹³²

2.3.2.

Metal oxo-cage or oxo-cluster resists

Among the conventional electron beam resists, HSQ has been extensively employed for high-resolution patterning.^98,106 This resist consists of an oxo-cluster of Si as its building block. Despite the issues with long-term stability, it possesses uniquely beneficial properties such as sub-10 nm resolution with extremely low LER and high etch resistance for pattern transfer thus proved to be a prelude to the development of spin-coatable metal oxide inorganic resists.

Perhaps the most significant development that followed was the resists based on aqueous chemistries of Zr and Hf.¹³⁴ These resists are based on solution processable ${\mathrm{HafSO}}_x$ and ${\mathrm{ZircSO}}_x$ amorphous films that can easily be deposited using aqueous solutions into smooth films. They were prepared by reacting aqueous solutions of ${\mathrm{ZrOCl}}_2$ and ${\mathrm{HfOCl}}_2$ with ${\mathrm{H}}_2{\mathrm{O}}_2$ and ${\mathrm{H}}_2{\mathrm{SO}}_4$. These resists show sensitivities as low as $8\mu \mathrm{C}/{\mathrm{cm}}^2$ at 30 kV electron beam exposure [Fig. 6(a)]—a significant achievement for a purely inorganic non-CAR. More importantly, they also showed lower values of line width roughness (LWR). The differential solubility between exposed and unexposed areas arises from electron beam-induced decomposition and elimination of peroxide, which promotes condensation reaction and crosslinking of metal-oxo bridges. In a subsequent report, 21-nm lines LWR as low as 1.6 to 1.8 nm was reported along with 18-nm half pitch patternability [Figs. 6(b) and 6(c)].¹³⁵ These encouraging results led to commercialization of the resist platform by Inpria Corp. They continued improvement of the resist processing reporting 12-nm half pitch patterning and across the board high dry-etch resistance compared to HSQ.¹³⁸ It has been further reported that ${\mathrm{HafSO}}_x$ resist can be patterned down to 9 nm with 10-kV electron beam lithography [Fig. 6(d)].¹³⁶

Fig. 6

Electron beam patterning of metal sulfoxide resists at 30 kV. (a) Exposure response curves for ${\mathrm{ZircSO}}_{\mathrm{x}}$ and ${\mathrm{HafSO}}_{\mathrm{x}}$ with metal-to-sulfate ratios of 1:0.5 and 1:0.55, respectively, after development in 25% w/w TMAH (from Ref. 134). (b) 18 nm LS patterns after $244\mu {\mathrm{C}/\mathrm{cm}}^2$ exposure and (c) 21 nm lines on 60 nm pitch showing LWR of 1.6 to 1.8 nm after $436\mu {\mathrm{C}/\mathrm{cm}}^2$ exposure (from Ref. 135). (d) 9 nm lines spaced at 35 nm patterned in HafSOx resist using 10 kV electron beam (from Ref. 136). (e) Molecular structures of organotin carboxylate resists and (f) their dose response curve when exposed with a 2 kV electron beam (from Ref. 133). (g) High-resolution patterns of dodecameric organotin resist patterned using $400\mu {\mathrm{C}/\mathrm{cm}}^2$ exposure using a 30-kV electron beam and developed in 2-heptanone (from Ref. 137).

New developments in the field of metal-oxo cage resists are primarily driven by their success in EUV lithography, occasionally with a comparative study involving electron beam patterning. Inpria Corp. introduced a next generation of metal oxide inorganic resists based on tin (Sn) in an attempt to improve their absorbance for EUV radiation.¹³⁹ This development further spurred a slew of organotin-based resists. While Sn dodecamer (${\mathrm{Sn}}_{12}$)-based resist chemistry attracted particular attention,¹⁴⁰ other chemistries, such as sodium-centered butyltin ($\beta \text{-}{\mathrm{NaSn}}_{13}$),¹⁴¹ butyltin oxo hydroxo cluster,¹⁴² and hexameric organotin carboxylates,¹³³ are interesting as well [Figs. 6(e) and 6(f)]. During the patterning of these resist materials, radiolysis of the Sn–C bond initiates hydrolysis and condensation reactions giving rise to tin oxide framework that is rendered less soluble in a developer. It has also been suggested that post-exposure bake may cause dehydration along with crosslinking of organic ligands, which may impact the solubility contrast. Resolution down to $\sim 8\mathrm{nm}$ linewidth has been thus demonstrated using Sn-based chemistries [Fig. 6(g)].¹³⁷ Oxo-cluster-based resists containing different metals such as Ti, Zr, Hf, Zn and the effect of ligand chemistry on EUV lithography performance is being actively investigated.^143–147 In a recent study, oxo-cluster resist based on Ni has also shown $\sim 9\mathrm{nm}$ resolution isolated features with LER as low as $\sim 1.8\mathrm{nm}$ patterned using electron beam lithography.¹⁴⁸

2.3.3.

Spin-coatable metal sulfide resists

This is a recent addition in the armory of metal-containing molecular resists for electron beam lithography to obtain directly metal sulfide under an electron beam. An earlier attempt involved electron beam lithography of palladium hexadecylthiolate, a negative-tone resist, which was converted to ${\mathrm{Pd}}_4\mathrm{S}$ via a post-thermolysis step.¹⁴⁹

We mentioned in the section “Stabilized metal alkoxide-based resists” that metal alkoxides are nucleophiles that can be stabilized via chelation. One of the other ways to stabilize a metal alkoxide, $[{\mathrm{M}}^{+}(\mathrm{OR}{)}^{-}]$, is to insert a mild electrophile such as carbon disulfide (${\mathrm{CS}}_2$). This leads to the formation of a metal xanthate or metal dithiocarbonate, ${\mathrm{M}}^{+}({\mathrm{ROCS}}_2{)}^{-}$ (where R is the alkyl group), which is just about stable at room temperature. Metal xanthates rapidly undergo decomposition when heated around 100°C, some even at a lower temperature. Furthermore, many of them freely dissolve in organic solvents such as chloroform. Their borderline stability and good solubility in organic solvents make them highly suitable candidates as electron beam resists.

Transition metal xanthates are conveniently prepared by metathesis reaction between a metal salt and potassium alkylxanthates in an aqueous medium. They usually exhibit bidentate bonding between the transition metal atom and the two sulfur atoms of xanthate group, which can be potentially cleaved using an energetic beam of electrons [Fig. 7(a)]. In an early report, mixture of metal xanthates was used as precursor for direct-nanopatterning copper indium sulfide for use in hybrid photovoltaics with EUV interference lithography.¹⁵³ The work of Saifullah et al. is the earliest demonstration of direct patterning of metal sulfide under an electron beam. They showed the suitability of zinc alkylxanthate-based spin-coatable resists for direct lithography of ZnS.¹⁵⁰ The dose sensitivity and solubility of zinc alkylxanthates increases with increasing length of the alkyl chain. The sensitivity of these resists was close to $35\mathrm{mC}/{\mathrm{cm}}^2$ at 100 kV and were capable of achieving very high resolution – ZnS lines as small as 6- and 10-nm dots with pitches as close as 22 nm were achieved [Figs. 7(b) and 7(c)]. Just like spin-coatable metal oxides, the lift-off step was noticeably absent here as well. Fabricated patterns of ZnS showed defect-induced photoluminescence related to sub-band-gap optical transitions [Figs. 7(d) and 7(e)]. Similarly, Recatala-Gomez et al. employed bismuth(III) ethylxanthate to pattern bismuth sulfide (${\mathrm{Bi}}_2{\mathrm{S}}_3$) to study its thermoelectric properties [Fig. 7(f)].¹⁵¹

Fig. 7

(a) Molecular structure of zinc alkylxanthate (left), bismuth ethylxanthate (middle), and molybdenum thiocubane (right), where R- indicates the alkyl group. Molybdenum thiocubanes have the formula ${\mathrm{Mo}}_4{\mathrm{S}}_4({\mathrm{ROCS}}_2{)}_6$. Here in the central cube, molybdenum and sulfur ions occupy diagonally opposite corners as shown in blue. Shown in black are four xanthate groups that are coordinated to a molybdenum ion in a chelating bidendate fashion. Two xanthate groups in red are coordinated to a molybdenum ion in a bridging bidendate manner. (b) Electron beam lithography of zinc butylxanthate gives 6 nm wide lines and (c) 10 nm dots with 22 nm pitch of ZnS (from Ref. 150). (d) Optical and (e) confocal images of patterned structures showing defect-induced photoluminescence from ZnS nanocrystals when imaged between 500 and 600 nm wavelength range ($\text{scale bar}=10\mu \mathrm{m}$). (f) Direct patterning of ${\mathrm{Bi}}_2{\mathrm{S}}_3$ (from Ref. 151), and (g) ${\mathrm{MoS}}_2$ at single digit nanometer scale using electron beam lithography (from Ref. 152).

For direct patterning of ${\mathrm{MoS}}_2$, tetranuclear molybdenum thiocubane (${\mathrm{Mo}}_4{{\mathrm{S}}_4}^{6+}$) complexes with xanthate ligands were used.¹⁵² They were prepared by refluxing molybdenum hexacarbonyl with different dialkylxanthogen disulfides in toluene. These complexes [${\mathrm{Mo}}_4{\mathrm{S}}_4{({\mathrm{ROCS}}_2)}_6$, where R is an alkyl group] show excellent solubility in organic solvents and show good film formability. Their electron beam sensitivity is very close to that of ZnS resist and are capable of single digit nanometer resolution. An electron-beam-induced radiolysis of a molybdenum thiocubane resist resulted in the fabrication of 9-nm-wide ${\mathrm{MoS}}_2$ lines and dense dots as small as 13 nm with a pitch of 33 nm [Fig. 7(g)].

To pattern metal sulfides using an electron beam, Wang et al. used a different precursor based on metal alkylthiocarbamate.¹⁵⁴ They reacted butyldithiocarbamic acid with transition metal oxides to form metal butyldithiocarbamates. These compounds were chemically transformed to metal sulfides via patterning using electrons/photons followed by a heat-treatment step. For electron beam writing, unlike the xanthates, the dose required for butyldithiocarbamates was enormously high—$7200\mathrm{mC}/{\mathrm{cm}}^2$ for ${\mathrm{Sb}}_2{\mathrm{S}}_3$, $3600\mathrm{mC}/{\mathrm{cm}}^2$ for ZnS, and $4800\mathrm{mC}/{\mathrm{cm}}^2$ for PbS. This is not surprising as metal alkyldithiocarbamates, in contrast to metal alkyldithiocarbonates (or xanthates), are quite stable compounds. More stability of the precursor translates into higher dose requirement to chemically change the material, with a requirement of post-heat-treatment step to remove residual organics. Micro-Raman and micro-FTIR spectroscopies were used to follow time-resolved electron beam damage studies in metal sulfide resists based on metal xanthates. These studies suggest that exposure to a beam of electrons lead to quick disappearance of xanthate moieties, most probably via the Chugaev elimination with the concurrent formation of metal sulfide [Fig. 8].^150–152 This phenomenon makes the exposed resist insoluble in organic solvents and gives rise to the negative-tone behavior.

Fig. 8

(a) Evolution of Raman spectra of ZnS resist with increasing electron dose. It is characterized by the appearance and increase in intensity of first- and second-order longitudinal optical phonon (LO and ${\mathrm{LO}}_2$, respectively) modes. (b) The corresponding TEM image shows the in situ formation of cubic ZnS nanocrystallites (from Ref. 150). (c) Evolution of Raman spectra of ${\mathrm{MoS}}_2$ resist with increasing electron dose showing the appearance of characteristic ${{\mathrm{E}}^1}_{2\mathrm{g}}$ and ${\mathrm{A}}_{1\mathrm{g}}$ peaks. (d) The TEM image shows lattice fringes that are spaced at 0.68 nm that corresponds to the d-spacing of (002) crystal planes in ${\mathrm{MoS}}_2$ (from Ref. 152).

2.4.1.

Metal-containing polymeric resists

Polymeric resists such as PMMA are widely used in electron beam lithography. To improve their properties and to serve a functional purpose, metallic species are added to them. Dating back to 1979, Webb and Hatzakis reported that copolymers of methyl methacrylate and metal methacrylates exhibit improvement of sensitivity and contrast for electron beam lithography when optimal metal content is incorporated into the composition.¹⁵⁵ They observed that the sensitivity of the resist improved with increase in the atomic number of the incorporated metal. In 1986, direct writing of gold nanostructures using electron beam exposure was demonstrated using gold containing organometallic polymer.¹⁵⁶

Chuang et al. used an interesting approach where they induced oxidation–reduction reaction with an electron beam (at a dose of $<2\mathrm{mC}/{\mathrm{cm}}^2$) in a lanthanum strontium manganese oxide (LSMO) gel that had polyvinyl alcohol as fuel and nitrate ions (${{\mathrm{NO}}_3}^{-}$) as an oxidant.^157,158 Lanthanum, strontium, and manganese nitrates provided ${{\mathrm{NO}}_3}^{-}$ ions in the gel. This type of autoignition process is characterized by a sharp and an intense exotherm and is a selfpropagating combustion reaction. Spin-coatable and water-soluble LSMO gel is a dual tone resist whose behavior can be modified by changing the dose. Further reports demonstrate that if Ag or Au salts are added to polyvinyl alcohol, electron beam exposure can induce acid catalyzed crosslinking within polyvinyl alcohol rendering the exposed regions insoluble in water thus exhibiting negative-tone behavior.^159,160 The addition of silver nitrate to poly(vinyl pyrrolidone) also exhibited negative-tone pattern ability; however, it was found that the nanocomposite formulation showed lower patterning resolution.¹⁶¹ Better results were obtained for a negative resist that consists of chloroauric acid (${\mathrm{HAuCl}}_4$) and poly(vinyl pyrrolidone) hybrid, giving a minimum Au feature size of $\sim 100\mathrm{nm}$ after electron beam lithography.¹⁶²

In a recent report, the addition of $\sim 2\text{-}\mathrm{nm}$-sized Ag nanoparticles to tert-butyl 2-ethyl-6-(4-hydroxyphenyl)-4-phenylheptanoate (Terpolymer) resist served as radiation sensitizer. Incremental addition of Ag nanoparticles up to 1 wt. % exhibited 10-fold sensitivity improvement with acetonitrile as developer.¹⁶³ The resist composition also showed $\sim 12\mathrm{nm}$ resolution with $\sim 1.5\mathrm{nm}$ LER. Zong et al. reported doping commercial electron beam resists with bimetallic (PtFe, PtCo, etc.) nanoparticles leading to significant improvement in their mechanical properties.¹⁶⁴ The modified resist exhibited increased pattern collapse mitigation and enhanced etch resistance for pattern transfer using ion-beam etching. Con et al. used coevaporation of Cr and polystyrene to incorporate metal into the polymeric matrix and reported stark improvement in the patterning contrast as well as dry etch resistance.¹⁶⁵ The authors were thus able to fabricate 100-nm Si-nanopillars with an aspect ratio as high as 35 using $\sim 90\text{-}\mathrm{nm}$ thick resist.

Several metallopolymers have been utilized for electron beam patterning. Earliest of the reports out of the Bell Labs utilized ferrocene containing polymers such as poly(vinyl ferrocene) for negative-tone electron beam resists.^166,167 One such platform utilized polyferrocenylsilane (PFS) backbone synthesized by ring opening polymerization of silaferrocenophanes.¹⁶⁸ Clendenning et al. demonstrated direct writing of magneto-optically active ceramics using cobalt-clusterized PFS.¹⁶⁹ This approach was further extended to PFS resists containing Ni and Mo clusters as pendent moieties.¹⁷⁰ In another report, metallopolymer with iron and phosphorus containing backbone, PFpP, was used for negative-tone electron beam patterning.¹⁷¹ Features as small as 17 nm were obtained [Figs. 9(a) and 9(b)]. Use of water-soluble poly(sodium 4-styrenesulfonate) (PSS) as a negative-tone electron beam resist was also reported having sensitivity in the similar range as PFS and PFpP.¹⁷³

Fig. 9

(a) Migration insertion polymerization synthesis of PFpP resist. (b) High resolution isolated line features of PFpP resist fabricated by $15\mathrm{nC}/\mathrm{cm}$ exposure using 20 kV electron beam (from Ref. 171). (c) Sn containing PBP resist is synthesized by free radical copolymerization of (4-(methacryloyloxy)phenyl)-dimethylsulfoniumtriflate (MAPDST) and ADSM. (d) 12.3 nm wide line patterns of the Sn containing PBP resist exposed with $400\mu \mathrm{C}/{\mathrm{cm}}^2$ with 20 kV electron beam and developed in 0.002 N TMAH (from Ref. 172).

2.4.2.

Ligand capped nanoparticle resists

In the earliest of the reports, sub-50-nm direct electron beam patterning was demonstrated using surfactant stabilized Pd, Pd/Pt,^174,175 and Au cluster.^176–179 Electrically active nanowire structures were also demonstrated by direct electron beam lithography on a Ru cluster polymer.¹⁸⁰ Langmuir–Blodgett films of alkanethiol-capped gold nanoparticles were also subjected to electron beam exposure to define nanowire structures at doses as low as $0.5\mathrm{mC}/{\mathrm{cm}}^2$, much lower than previous reports that used passivated gold nanoclusters.¹⁸¹

${\mathrm{ZrO}}_2$ and ${\mathrm{HfO}}_2$ nanoparticles with various acrylic acid ligands were investigated for their application in nanolithography by Ober’s group at Cornell University. A systematic study with different developer solutions led to the fabrication of resist patterns down to $\sim 32\mathrm{nm}$ with ketone-based solvent developers.¹⁸² The approach was further extended to resist formulations containing ${\mathrm{ZnO}}_{\mathrm{x}}$, ${\mathrm{TiO}}_{\mathrm{x}}$, ${\mathrm{InO}}_{\mathrm{x}}$ and ${\mathrm{SnO}}_{\mathrm{x}}$; with optimization of resist processing, 15-nm half pitch patterns were achieved.¹⁸³ However, the patterned features suffered from high LER along with significant residual scum artifact that could be the consequence of nanoparticle size and high concentration of photoacid generator (PAG).

One of the widely known classes of functional nanoparticles is that of colloidal quantum dots owing to their tunable narrow bandwidth photoluminescence. However, their integration into scalable device platform is challenging since they can lose their performance when agglomerated. Few recent studies have utilized electron beam exposure as a means to directly pattern quantum dots into high-resolution optoelectronic devices. Nandwana et al. demonstrated electron beam direct patterning of trioctylphosphine oxide functionalized CdSe/ZnS quantum dots, where electron beam-induced crosslinking of the ligands rendered exposed regions insoluble in toluene.¹⁸⁴ Subsequently, CdSe/CdS quantum dot patterns down to 30-nm lines and 50-nm disks were also reported.¹⁸⁵ Wang et al. demonstrated a versatile approach for direct-patterning of quantum dots by exploring a range of different ligand chemistries. They were able to show ${\mathrm{CeO}}_2$ nanoparticle patterns down to 30 nm linewidth, whereas CdSe quantum dot patterns down to 70 nm linewidth were also reported.¹⁸⁶ More recently, direct patterning of quantum dots using an electron beam as well as EUV exposure has also been reported.¹⁸⁷ Here upon exposure, oleic acid, the capping ligand for quantum dots, and a carboxylic acid, underwent cross-linking, rendering the exposed area insoluble in an apolar developer. Electron-beam-induced cross-linking of long-chain ligands (carboxylate and oleylamine) was also utilized for direct patterning of ${\mathrm{CsPbBr}}_3$ nanocrystals.¹⁸⁸ This type of cross-linking is similar to what has been observed earlier in metal carboxylate-based resists (see section “Metal carboxylate-based resists”).

2.4.3.

Metal-containing polymer-bound PAG resists

CARs have been load-bearers for semiconductor high volume manufacturing for decades due to their fast photo-speed, largely owing to highly radiation-sensitive small molecule additive called the PAG. However, as the critical dimension started to drastically decrease, PAG aggregation and acid diffusion posed a pertinent challenge for pattern fidelity. One of the solutions that was implemented was PAG grafting into the backbone of the copolymeric resists namely—polymer-bound PAG (PBP) resists.^189,190 However, with the emergence of EUV lithography, these primarily organic compositions started falling short of achieving adequate radiation absorption and promoting the incorporation of inorganic/metal species into the resist mixture.

A PBP composition containing ferrocene pendent group was reported to exhibit improved electron beam sensitivity and resolution compared to the composition without ferrocene group.^191,192 A resist concept using W and Mo polyoxometalates as pendent group in PBP composition has also been explored.¹⁹³ A recent report detailed doping of PBP resist with thiolated Ag nanoparticles that demonstrated improved sensitivity to electrons and lower LER for sub-15 nm features.¹⁹⁴ The performance improvement was attributed to the photomultiplication effect induced by the Ag nanoparticles supporting plasmonic resonance. Similarly, copolymer synthesized using acetyldibutylstannyl methacrylate (ADSM) for incorporating Sn containing pendent group was also demonstrated, providing resolution down to $\sim 12.3\mathrm{nm}$ with $\mathrm{LER}<1.3\mathrm{nm}$ [Figs. 9(c) and 9(d)] after development in 0.002 N tetramethylammonium hydroxide (TMAH).¹⁷²

2.5.1.

Metal-organic framework-based resists

Metal-organic frameworks (MOF) are organic–inorganic hybrid crystalline porous compounds that consist of metal ions coordinated to organic ligands to form one-, two-, or three-dimensional cage-like structures. These materials can be deposited with high degree of regularity leading to highly regular absorbance. The use of zeolitic imidazolate framework particles loaded into phenyltriethoxysilane sol–gel matrix was reported for generating negative-tone patterns using deep x-ray exposure.¹⁹⁵ MOF inspired Zn-cluster resist was also demonstrated to pattern dense negative-tone structures of 25-nm half pitch with electron beam lithography and 13-nm half pitch with EUV exposure.¹⁹⁶ Recently, Tu et al. demonstrated direct patterning of MOFs using electron beam and x-ray lithographies.¹⁹⁷ Direct patterning avoids damage due to etching and leaves the porosity and crystallinity of the patterned MOF intact. Unlike the previous reports, positive-tone resist behavior was observed here due to the crystalline to amorphous transition of exposed regions becoming soluble in solvent such as dimethyl sulfoxide. They were able to pattern sub-50 nm features of MOFs with potential applications in high-performance dielectrics, and selective and sensitive sensors.

2.5.2.

Infiltration synthesis of hybrid resists

Vapor phase infiltration synthesis: an ex-situ approach for converting a conventional polymeric resist into a hybrid resist was recently demonstrated.^198,199 Infiltration synthesis approach is based on vapor phase precursors typically used for atomic layer deposition (ALD); however, contrary to adsorption driven thin film deposition, precursors are made to diffuse into polymeric resist where they bind to reactive functional groups within the polymer. As a result, metal oxide moieties are grown inside the resist matrix. In the initial report, PMMA resist was converted into $\mathrm{PMMA}\text{-}{\mathrm{AlO}}_{\mathrm{x}}$ and its performance for electron beam lithography was studied. With an increasing amount of infiltration, the resist exhibited decrease in sensitivity while improving the patterning contrast by sixfold. At the same time, stark improvement in the dry etching resistance was observed giving Si etch selectivity estimated to be in excess of 300. Patterned features with $\sim 30\mathrm{nm}$ linewidth were transferred in an Si substrate to fabricate Si fins with an aspect ratio of $\sim 17$. Subsequently, an approach for improving the resist sensitivity by Hf-infiltration has also been introduced.²⁰⁰ In a recent report, infiltration of various metal oxides into a high sensitivity resist was reported with a resolution $\sim 30\mathrm{nm}$.²⁰¹ While controllability and ease of implementation being the high points, further maturation of this resist platform is required.

Table 1 provides a succinct summary of electron beam lithography of metal-containing resists reported over the years.

Table 1

Summary of key metal-containing resists reported over the years.