Possible mechanisms in atomic force microscope-induced nano-oxidation lithography in epitaxial La_0.67Ba_0.33MnO_3-δ thin films Free

The physical properties of perovskite manganites, such as colossal magnetoresistance (CMR), metal-insulator (MI) transition, and electronic/magnetic phase separation/coexistence are of continuing interest. The sensitivity of some compositions to an applied magnetic field has opened new possibilities for magnetic sensors and spintronic devices. A perovskite manganite application that has already found real world usage simply utilizes the dramatic resistance versus temperature displayed by appropriately doped MI compositions for bolometric application.^1–4 Perovskite manganite bolometers (fabricated by existing standard lithographic techniques) have been used to measure the total energy of the Linac Coherent Light Source free electron laser at the Stanford Linear Accelerator Center (SLAC) and to calibrate its diagnostic instruments.^5–9 Future optimization and integration of these sensors and devices will likely require the fabrication of nanostructures, which will prove challenging to existing standard lithographic techniques.

An alternative technique that can offer greater control at the atomic level is AFM induced nanolithography. This technique consists of utilizing the voltage biased conducting tip of an AFM to induce local chemical modification of the sample, with the AFM operated in a contact mode. The voltage bias between the sharp tip of the AFM and the sample generates an intense electric field in the vicinity of the tip. The intense electric field can trigger a local transformation in morphology and chemistry at the film surface. In the presence of ambient or added humidity, water collects between the tip and the surface, and subsequently, a two electrode nanoelectrochemical cell is formed. In perovskite manganite films, above a threshold voltage, the area beneath the AFM tip is modified.^10–19 The nanomodified areas can subsequently be transformed into a nanotrench by a simple wet etching with a mild acid without affecting the unmodified areas.^10–18 

The interplay of magnetic, electronic, and crystal structure, which results in the coexistence of electronic and magnetic phases in different mesoscopic areas of even a chemically homogeneous single crystal, and which is associated with the CMR and MI transition, still needs better understanding. Intrinsic nanoscale electrical transport studies on perovskite manganites enabled by AFM induced nanolithography have already yielded some insights in these strongly correlated electronic materials.^{10–12,16–18} Previous work has established that AFM induced nanolithography is an appropriate technique for many compositions of epitaxial perovskite manganite films on single crystal oxide substrates. Compositions that have been successfully patterned include La_0.7Sr_0.3MnO_3-δ,¹² La_0.67Ba_0.33MnO_3-δ,^10,11,19 La_0.8Ba_0.2MnO_3-δ (Refs. 16 and 13–15) films on (001) SrTiO₃ (STO) substrate, and La_0.275Pr_0.35,Ca_0.375MnO_3-δ (Refs. 17 and 18) on (001) (LaAlO₃)_0.3(Sr₂TaAlO₆)_0.7 substrate. In all cases, the AFM was operated in the contact mode, with most experiments done in the constant voltage mode, except for two reports^12,18 employing constant current mode.

Controllable AFM nanolithography has been performed on epitaxial perovskite manganite films on single crystal oxide substrates with both negative^10–12,19 (see Fig. 1) and positive^13–19 polarities of the AFM tip with respect to the sample. There is a general consensus among these previous works that electrochemical processes are involved in AFM nanolithography of perovskite manganites, but thus far there has been no clear understanding of the exact mechanisms involved (for either polarity). It is notable that in the (−) AFM tip case [but not in the (+) AFM tip case], large features of excessive outgrowths can occur under conditions of unregulated current, high humidity, low scan speed, and higher voltage (see Figs. 2 and 3). These features can even be large enough to be visible to the naked eye and are thus large enough to be meaningfully subjected to electron energy dispersive x-ray spectroscopy (EDS) elemental analysis. It was expected that the elemental composition information revealed by EDS could possibly provide insights into the chemical and physical processes involved in (−) AFM tip nanolithography. Such a large feature of surface modification in oxygen deficient La_0.67Ba_0.33MnO_3-δ films on STO(001) induced by a (−) AFM tip was investigated using EDS (with comparison being made to EDS performed on a unmodified region).¹⁹ Results of this EDS study indicate a significant weight percent increase in the oxygen content of the patterned regions relative to the pristine film. Furthermore, drastic weight percent increases in Ba and C, and weight percent decreases in Mn and La stoichiometries were observed. The present work focuses upon providing a framework of electrochemical and physical processes potentially involved in the modification of La_0.67Ba_0.33MnO_3-δ on STO(001) substrates, with the AFM tip voltage biased (−) with respect to the sample (see Figs. 1–3). Our postulated framework of electrochemical and physical processes provides qualitative explanations for the volume increases observed in the modified regions along with the acid etch ability observed in the modified regions. Additionally, the elemental weight percent changes predicted via our postulated framework of electrochemical and physical processes are in reasonable accord with the EDS observed elemental weight percent changes.

Since this work is focused on understanding the nanolithography processes in oxygen deficient La_2/3Ba_1/3MnO_3-δ thin films, it is pertinent to first examine the effects of varying film oxygen content. It is well known that oxygen can be very mobile in the perovskite structure (especially when it has oxygen vacancies). This implies that there is a relatively low energy barrier to oxygen being restored (at least into the top surface layers) in the oxygen deficient La_2/3Ba_1/3MnO_3-δ film; thus, this should occur at low voltages (most likely below the threshold voltage). It is possible that oxygen deficient La_2/3Ba_1/3MnO_3-δ (LBMO) regions first becomes fully oxygenated before participating in further electrochemical oxidation half reactions leading to even greater oxidation states.

In a recent work, it was reported that oxygen deficient La_2/3Ba_1/3MnO_3-δ films has a lattice constant of 3.97 Å while fully oxygenated La_2/3Ba_1/3MnO₃ films has a lattice constant of 3.90 Å.¹⁹ (The slight volume decrease as La_2/3Ba_1/3MnO_3-δ becomes La_2/3Ba_1/3MnO₃ is due to the radius of Mn³⁺ ion being larger than that of the Mn⁴⁺ ion.) It was also reported that the average roughness (R_a) value of the oxygen deficient films was 1.8 Å (with a standard deviation of 0.1 Å) and that the original film thickness was 1400 Å (which corresponds to ∼353 oxygen deficient unit cell layers).¹⁹ From these reported values, it can be estimated that if 26 or more unit cell layers were to become fully oxygenated, indentations/grooves forming should have been observable via AFM profiling. Actually, it has been reported [for the (−) AFM tip case] that at times grooves/trenches were obtained instead of peaks.^13–15 

This work will focus primarily on interpreting the EDS results of Tanyi et al. in which oxygen deficient La_2/3Ba_1/3MnO_3-δ films were used; the exact oxygen stoichiometry of the unmodified film was not reported,¹⁹ since it is not possible to determine the absolute value of the oxygen content in films using the EDS techniques that were employed. However, the EDS results clearly indicate relative changes in the oxygen content between modified and unmodified regions. For the sake of concreteness in balancing the chemical reactions presented in this work and in the subsequent weight percent change and volume percent change calculations, the maximum oxygen deficiency will be assumed for the unmodified films. One can argue that the maximum oxygen deficiency should occur when all the Mn exist in the Mn³⁺ oxidation state, which is x = 1/6 for this composition. Any oxygen deficiency over x = 1/6 is energetically unfavorable in the perovskite structure as this would require the presence of the Mn²⁺ oxidation state (Table I). Thus, an initial composition of La_2/3Ba_1/3MnO_2.83 is assumed in the postulated chemical reactions that are presented in the following sections.

Constant voltage mode nano-oxidation on epitaxial La_1-xBa_xMnO_3-δ films on STO(001) with (−) AFM tip bias has been reported by three groups.^{10–11,13–15,19} Li et al. also employed friction force microscopy (FFM) to study their La_0.8Ba_0.2MnO_3-δ films on SrTiO₃(001) to reveal that the region nanomodified under a (−) AFM tip is characterized by a larger friction force as compared to the unmodified regions. However, they were unable to obtain controllable and reproducible patterns with (−) AFM tip bias [as a result of which the primary focus of their work was on the (+) AFM tip case for which it was possible for them to consistently obtain reproducible and controllable lines].

In contrast, controlled patterns on La_0.67Ba_0.33MnO_3-δ films on STO(001) with AFM (−) tip (in constant voltage mode) were reported by Pallechi et al.^10,11 In fact, these workers make no mention of using the other voltage bias choice (+) AFM tip. They noted that the best resolution was achieved in low humidity conditions and at higher scan speeds. A 10% HF dip for 10–20 s was used to selectively remove the (−) AFM tip nanomodified parts of the film, leaving the unmodified regions intact.^10,11

In a recent work, La_0.67Ba_0.33MnO_3-δ films on STO(001) were modified employing both tip polarities in constant voltage mode.¹⁹ As was previously found^13–15 with the (+) AFM tip, it was possible to consistently obtain reproducible and controllable lines, while with the (−) AFM tip at the same writing speed, excessive outgrowths were observed, which prevented the tip from moving.¹⁹ As was previously found,^10,11 it was possible to write controllable and reproducible lines with the (−) AFM tip by increasing the writing speed.¹⁹ In an effort to better understand the chemical changes induced by the (−) AFM tip nanolithography process, a large area of excessive outgrowths was deliberately produced on a La_0.67Ba_0.33MnO_3-δ film by (−) AFM tip modification and this area was subjected to EDS) elemental analysis.¹⁹ EDS elemental analysis was performed on both the pristine and modified film regions; the results are summarized in Table II.¹⁹ An increase in O, a drastic increase in C, a decrease in La, a drastic increase in Ba, and a decrease in Mn, weight percentages in the patterned regions relative to the unpatterned regions were observed. From these EDS results, Tanyi et al. was able to definitively state that oxygen stoichiometry was increased in the film after (−) AFM tip nanomodification, consistent with oxidation occurring at the La_2/3Ba_1/3MnO_3-δ film in this polarity; however, there were no explanations for any of the other EDS detected elemental weight percent changes in the film after the (−) AFM tip nanomodification.¹⁹ 

In the sections that follow, a self-consistent framework is presented, which is consistent with the physical and chemical changes observed in the experiments above. First, a simple reduction reaction is proposed at the (−) AFM tip, which will be followed by a series of postulated oxidation reactions at the (+) LBMO film surface.

The two electrode nanoelectrochemical cell consists of the AFM tip (−) bias as the cathode (electrode at which electrons enter the electrolytic liquid), the LBMO film surface (+) bias as the anode (electrode at which electrons are withdrawn from the electrolytic liquid), and the water meniscus formed between the tip and the surface as the electrolytic liquid (Fig. 1). Assuming that the tip material does not participate in any reactions (as is consistent with no mention of any chemical changes in the AFM tip by previous workers), a primary half reaction that can occur at the AFM tip is the reduction of water (Table III).

Table IV presents a list (obtained from standard references) of known compounds composed of the available elements (La, Ba, Mn, O, and H); moreover, this list only includes compounds in which elements are in equal or higher oxidation states as compared to their oxidation state in La_2/3Ba_1/3MnO_2.83.

Table V lists six possible half reactions [(0), (1), (2′), (3′), (4), and (5)] at the LBMO (La_2/3Ba_1/3MnO_2.83) film surface, reacting with progressively greater amounts of water and resulting in known compounds with progressively larger number of electrons released. (The unprimed denotes a half reaction in which Ba and Mn remain separate in different product compounds while the primed denotes a half reaction in which Ba and Mn are combined into one product compound.) Note that of the cations, only Mn undergoes changes in formal oxidation state; the La and Ba do not undergo any changes in formal oxidation state. (La stays as La³⁺ and Ba stays as Ba²⁺.) It is primarily the Mn with its large range of available oxidation states that is responsible for the existence of this series of possible oxidation half reactions. Note that after half reaction (1a) in which the La decomposes into La₂O₃, other than possibly incorporating water [reaction (1c) ¹/₃La₂O₃ + H₂O → ²/₃La(OH)₃], it passively stands by with no further chemical reaction. However, the BaO and MnO₂ formed after half reaction (1a) has many more possibilities. Reaction (1b) shows BaO incorporating water to form the hydrate Ba(OH)₂ ·8H₂O. Ba²⁺ can also form combined compounds with Mn, with the Mn cation in progressively higher oxidation states. [From half reaction (1) to half reaction (2′), the Mn oxidation state jumps from Mn⁴⁺ → Mn⁶⁺; there does not appear to be any compound composed of the available elements with manganese in an oxidation state of Mn⁵⁺.] In half reaction (2′), some of the Mn is present in the combined compound BaMnO₄ with a Mn⁶⁺ oxidation state (to our knowledge, this is the only compound composed of the available elements with manganese in the Mn⁶⁺ oxidation state). In half reaction (3′), some of the Mn is present in the combined compound Ba(MnO₄)₂ with a Mn⁷⁺ oxidation state. In half reaction (4a), Ba and Mn separates into independent product compounds, that is, into BaO and Mn₂O₇, with all of the Mn present in a Mn⁷⁺ oxidation state. Mn₂O₇ is exceptionally unstable and would most likely react with water to form more stable compounds: Mn₂O₇ + H₂O → 2HMnO₄ [included in half reaction (4b)]. Half reaction (5a) involves the formation of barium peroxide BaO₂ (oxygen in the O⁻¹ oxidation state). BaO₂ can then incorporate water to form the hydrate BaO₂ · 8H₂O [half reaction (5b)].

In Table VI, the possible overall oxidation half reactions of the La_2/3Ba_1/3MnO_2.83 film (0, 1, 2′, 3′, 4, and 5 from Table V) are recast in a sequential manner. Table VI begins with the half reaction (0), which is the oxygenation of the initially oxygen deficient La_2/3Ba_1/3MnO_3-δ film (that for concreteness is assumed to be La_2/3Ba_1/3MnO_2.83, with all manganese in the Mn³⁺ oxidation state). Recall that in half reaction (1), the Ba and Mn product compounds are separated, while in half reactions (2′) and (3′), the Ba and Mn product compounds are combined, and finally, in half reactions (4) and (5), the Ba and Mn compounds become separated again. In half reaction (2′), the BaO from (1) and some of the MnO₂ from (1) are combined into BaMnO₄ (Tables V and VI). In half reaction (3′), the BaMnO₄ from (3′) and some of the MnO₂ from (1) are combined into Ba(MnO₄)₂ (Tables V and VI). In half reaction (4), some of the MnO₂ from (1) and the Ba(MnO₄)₂ from (3′) react to form separated compounds BaO and Mn₂O₇. In half reaction (5), the BaO is converted to the peroxide BaO₂. In the sequence of linking half reactions from (0) → (1) → (2′) → (3′) → (4) → (5), the lanthanum and barium oxidation states remain the same throughout, while the manganese oxidation state varies as Mn³⁺ → Mn⁴⁺ → Mn⁶⁺ → Mn⁷⁺→ Mn⁷⁺.

Combining the six overall oxidation half reactions at the (+) film (Table V) with the primary reduction half reaction at the (−) AFM tip (Table III), one obtains the complete reactions in Table VII. Table VII also includes predictions for the volume changes (calculated using tabulated densities), weight percent changes, and the elemental weight percent changes in the solid materials associated with each overall reaction. Higher humidity is expected to increase the rate of the nanolithography (in some cases even to uncontrollable levels, as reported by previous workers) as H₂O is a required reactant. The observed (in some cases extreme) volume increases, and EDS detected O weight percent increases (Table II) are qualitatively explained as the result of La_2/3Ba_1/3MnO_2.83 decomposing into other higher oxidation state product solids with ever increasing oxygen content and most with lower density. The largest volume increase that can be predicted is +167.8% [Table VII, Reaction (1c)]; due to incomplete reference density information, it was not possible to calculate the volume change of the patterned regions relative to the pristine surface for all of the postulated complete reactions. The maximum predicted O weight percent increase of +23.5% [Table VII, Reaction (5c)] for the patterned regions relative to the pristine surface falls somewhat short of the EDS measured O weight percent increase of +26.6% (Table II). This indicates that there may be other product compounds (besides La₂O₃, BaO, and BaO₂) that can also incorporate H₂O. This can also indicate that besides the incorporation of H₂O, there are product compounds that can incorporate CO₂ from the environment. Note that one molecule of CO₂ has two atoms of O while one molecule of H₂O only has one atom of O. For example, LaCO₃OH (Table IV) is a known compound; also, note that the PDF file for the BaO₂·8H₂O hydrate states that it loses water rapidly at lower humidity's and absorbs CO₂ from the air. The EDS detected drastic C weight percent increase of +109.4% (Table II) is highly supportive of the incorporation of environmental CO₂.

The mild acid etch ability of the modified regions is easily explained by the solubility of many of the postulated product oxides in acids (Table IV). Most notably, La₂O₃ appears early on as a product [in reaction (1)] and La₂O₃ together with La(OH)₃ are both soluble in acid.

As is well known, mass and atomic identities are conserved in chemical processes, and there does not appear to be any gaseous species at ambient conditions involving the cations of interest La, Ba, Mn (including in combination with O, and H). Furthermore, there have been no reports of any chemical changes occurring with the AFM tip material. Therefore, the electrochemical reactions can only predict minor decreases in the cation weight percentages as the product solids increase slightly in weight by taking up O and incorporating H₂O (Table VII). The maximum decrease in La weight percent (−12.4%) predicted by the electrochemical reactions (Table VII), falls short of the EDS measured (Table II) La weight percent decrease (−17.5%). The maximum decrease in Mn weight percent (−7.4%) predicted by the electrochemical reactions (Table VII), falls far short of the EDS measured (Table II) La weight percent decrease (−22.9%). The discrepancy between the La predicted and measured weight percent change and especially the larger discrepancy between the Mn predicted and measured weight percent change are indications that other mechanism(s) may be occurring (although the additional incorporation of CO₂ by product solids can rationalize somewhat larger decreases in predicted La and Mn weight percent changes). As for the Ba weight percent change, the electrochemical reactions (Table VII) can only predict small decreases, completely inconsistent with the drastic increase in EDS measured (Table II) Ba weight percent increase (+71.3%). The discordant EDS Ba weight percent result definitely points to the existence of additional physical and/or chemical processes. Additional mechanisms will be presented to further explain the magnitudes of the EDS measured decreases in La and Mn weight percentages, and to explain especially the drastic increase in EDS measured Ba weight percent.

In EDS, the sample is bombarded with electrons to knock out core electrons, thus stimulating the emission of characteristic x-rays from the specimen. EDS is generally thought of as a bulk analysis technique that is not well suited for nanoscale analysis; however, by careful selection of electron beam energy and tilt angle, it is possible to obtain a higher spatial resolution and greater surface sensitivity than is typically achieved.² Lower accelerating voltages result in shallow electron beam penetration and narrow lateral electron beam spread in the sample, while higher tilt angles have the concurrent effect of moving the interaction volume closer to the sample surface. Thus, it is possible to obtain spatial resolution on the order of 1000 Å or less by using very low accelerating voltages (<5 kV) and high tilt angles. Such care was taken in the EDS analysis performed on a 1400 Å La_2/3Ba_1/3MnO_3-δ film reported by Tanyi et al. (Table II).¹⁹ A low accelerating voltage of 5 kV was used to assess the O (and C) content of the film (Table II).¹⁹ Since O is present in both the LBMO film and the STO substrate, it was important to use low energy electrons to avoid confusing the results with O x-rays from the STO substrate.¹⁹ In fact, Monte Carlo simulations predict that for 5 keV electrons, O x-ray spectra originated almost entirely from the film.¹⁹ Indeed, this is confirmed in Table II by the EDS measured value of O weight percent in the unmodified area (19.6%) being very close to the calculated value of O weight percent in La_2/3Ba_1/3MnO_2.83 (19.0%). Furthermore, as C x-rays are lower in energy (and shorter in range) than O x-rays, for the 5 keV electrons used, the C x-ray spectra are expected to originate only from the LBMO film layer.

However, due to the fact that La and Ba spectra peaks overlap at lower electron energies, higher energy 12 keV electrons had to be used for the cations' assessment (Table II).¹⁹ X-ray signals from the STO substrate are present at 12 keV as at this energy the electron beam does penetrate into the STO substrate. However, relative comparisons are still valid as there are no common cations between the film and the substrate.¹⁹ 

In general, when interpreting the EDS results, the limitations of EDS must be kept in mind as it relies on the likelihood of an x-ray escaping the specimen, and thus being available to be detected and measured. The probability of an x-ray escaping the specimen depends on the energy of the x-ray, and the amount and density of material it has to pass through. Thus, a depth dependence is inherent in the EDS technique (and can result in reduced accuracy in inhomogeneous and rough samples). To illuminate the depth dependence intrinsic in the EDS technique, let us compare the ideal weight percent values of La_2/3Ba_1/3MnO_2.83 and stoichiometric SrTiO₃ with the EDS detected weight percent values in the unmodified region in Table II. It can be seen for all the elements tested (with the exception of C), that the EDS weight percent values in the unmodified area are intermediate between that of the ideal weight percent value of La_2/3Ba_1/3MnO_2.83 by itself and the ideal weight percent value of stoichiometric SrTiO₃ by itself (as opposed to mirroring the values in stoichiometric SrTiO₃ by itself). Even though overall the sample is composed of overwhelmingly much more STO substrate material than it is of LBMO film material, the fact that the LBMO film is on the top allows it to contribute disproportionally more to the overall signal.

The depth sensitivity of the EDS technique can be used to reconcile the EDS results showing a decrease in the La content in the modified region (Table II, −17.5%) that is significantly greater in magnitude than that which can be predicted by the sequence of proposed reactions (Table VII, −12.4%). Since La₂O₃ is produced early on and does not participate further in any other reactions [besides incorporating H₂O to become La(OH)₃], one can argue that it gets buried underneath the other still reactive oxides that are still involved in chemical reactions that lead to increases in mass and volume.

A well-studied source of printed circuit board (open circuit) failure is electromigration. Electromigration is a physical process whereupon the electrons moving as a result of the applied field collides with and transfers momentum to the lattice ions, causing the lattice ions to be pushed away from their original positions. This effect is greater with weakly bond ions and with smaller mass ions. In printed circuit boards, this has been observed to occur especially within the disordered low density grain boundary regions in Al and Cu interconnects. The amorphous nature and the lower density (Table IV) of the product solids in the proposed reactions (Table VII) can be expected to encourage electromigration; especially for the lighter elements, thus electromigration should have some effects on the EDS detected values (Table II). The direction of electron motion for the voltage bias (− AFM tip and + substrate) focused upon in this work would tend to drive the atoms/ions from the film surface further down into the sample resulting in these atoms/ions being harder to detect via the depth sensitive EDS technique. The less massive ions would tend to be moved farther away from their original position in comparison to the more massive ions making the less massive ions harder to detect via EDS. O (which is less massive than Al) and Mn (which is less massive than Cu) are the lightest ions and should be the most affected by electromigration while Ba and La are the heaviest ions and should be the least affected. Thus, some depression in the EDS detected Mn weight percent and O values in the modified area (Table II) due to electromigration are expected. Hence, electromigration can be used to partially reconcile the EDS results showing a decrease in the Mn content in the modified region (Table II, −22.9%) that is significantly greater in magnitude than that which can be predicted by the sequence of proposed reactions (Table VII, −7.4%). However, electromigration predicts that the EDS detected increase in O weight percent value in the modified region (Table II, +26.6%) should be less than (instead of being larger than) that which can be predicted by the proposed reactions (Table VII, +23.5%). This discrepancy is probably due to the fact that compounds such as Mn₆O₁₂(H₂O)_4.16, LaCO₃OH and BaO₂ · (8-x)H₂O · xCO₂ (Table IV) were excluded in the O weight percent changes calculated from the sequence of proposed reactions (Table VII). Recall that the EDS detected drastic C weight percent increase of +109.4% (Table II) is highly supportive of the incorporation of environmental CO₂ and that each molecule of CO₂ has two atoms of O while each molecule of H₂O has one atom of O.

Another well studied source of printed circuit board (short circuit) failure is electrochemical migration. Of the metals commonly used in printed circuit boards, electrochemical migration is the most pronounced for Ag interconnects. Ag leaves its initial location in ionic form, migrates under the applied electric field to another location, and is subsequently reduced to the metal state. This multistep process can only take place in a humid environment. First, silver readily form an oxide coating upon exposure to air (2Ag + ½ O₂ → Ag₂O) and the appreciable solubility of Ag₂O in water (Table VIII) is the source of Ag⁺. Second, the lower voltage required to reduce the Ag⁺ ion to the metal state compared to the voltage required to reduce water (Table IX) means that Ag⁺ is preferentially reduced.

For the case of interest in this work, examination of the solubilities of the various postulated product solids (Table VIII) from the proposed framework of reactions, one immediately notices that many of the barium compounds [especially Ba(MnO₄)₂] are quite soluble. Thus, the relatively high solubility of these proposed barium product compounds provide the source for Ba²⁺ in the electrolyte (water). However (unlike the Ag⁺ case), the higher voltage required to reduce the Ba²⁺ ion to the metal state compared to the voltage required to reduce water (Table IX) means that the Ba²⁺ ion does not get reduced to the metal. [This is consistent with no mention in the previous works of any chemical changes in (−) AFM tip.] Therefore, one can explain the pronounced increase in the EDS detected Ba weight percent (Table II, +71.3%) as the result of Ba²⁺ cations migrating up toward (but never plating out at) the (−) AFM tip at the film surface. As this barium has been drawn up to the film surface, they are easier for EDS to detect. Note that the high solubility of the proposed barium manganese combined product compounds (Table VIII) also results in availability of the negative manganese containing ions MnO₄⁻ and MnO₄²⁻. Since these ion clusters are negative, they would under the applied electric field tend to migrate away from the (−) AFM tip, and thus be driven deeper into the film. As this manganese has been driven deeper down into the film, it would be harder for the EDS to detect. So, the electrochemical migration mechanism also provides another explanation (in addition to electromigration) as to why the magnitude of the EDS detected Mn weight percent decrease (Table II, −22.9%) is larger than the maximum decrease that can be predicted by the electrochemical reactions (Table VII, −7.4%) alone.

The chemical and physical processes involved in the process of (−) AFM tip induced nanomodification on epitaxial perovskite La_2/3Ba_1/3MnO_3-δ thin films in a humid environment have previously not been understood. A conceptual framework consisting of a postulated series of electrochemical reactions, retention of H₂O and CO₂ by the postulated product solids (BaO₂, La₂O₃), the process of electromigration, and the process of electrochemical migration is qualitatively and to some extent quantitatively consistent with previously reported observations. The volume increase in nano-oxidized patterned regions is explained by a series of postulated electrochemical half reactions in which positively biased La_2/3Ba_1/3MnO_3-δ (in the presence of H₂O) decomposes into other higher oxidation state product solids with ever increasing oxygen content and with lower densities. Retention of H₂O and CO₂ by postulated product solids also increases the volume. (The EDS detected a drastic increase in C weight percent; Table II is highly supportive of the possibility of retention of CO₂ by the product solids.) The fast and (for some cases uncontrollable) process of nano-oxidization at a higher humidity is explained by the consumption of H₂O in the postulated series of reactions (Table VII). The ability of nano-oxidized patterned regions to be preferentially etched away with acids is explained by the postulated reactions (Table VII) as La₂O₃ appears early on as a product and La₂O₃ [and La(OH)₃] are soluble in acids. The increase in EDS detected O weight percent in nano-oxidized patterned regions is explained by the higher oxygen content product solids along with incorporation of environmental H₂O and CO₂ by the postulated product solids. The decrease in EDS detected La weight percent in nano-oxidized patterned regions can be explained by the increase in weight of the product solids (due to uptake of O along with incorporation of environmental H₂O and CO₂), along with the depth sensitivity of the EDS process. In the postulated series of reactions (Table VII), the La₂O₃ [and La(OH)₃] is produced early on and does not participate further in any other electrochemical reactions; thus, it can be argued that the La gets buried underneath the other still reactive oxides that are increasing in mass and volume. Lastly, the dramatic EDS detected increase in Ba weight percent and large value of EDS detected decrease in Mn weight percent in nano-oxidized patterned regions can be explained by the process of electrochemical migration, which is reliant upon the exceptionally high solubility in water of the postulated product Ba(MnO₄)₂, and subsequent drift of the dissolved ions in the applied electric field.

The authors acknowledge support from the NSF Grant No. ECCS 1128586 and seed funding from the School of Emerging Technologies at Towson University.