Structural and morphological tuning of iron oxide polymorphs by ECR plasma-assisted thermal oxidation

The work presented involves the generation of oxygen plasma species at low pressure utilizing an Electron Cyclotron Resonance (ECR) plasma reactor, and their interactions with micron- and nano-sized iron films (M-Fe and N-Fe film respectively) prepared using ethyl cellulose processed at high temperature. A specially designed radiation heater (RH) was used to raise the surface temperature of the film rapidly, exactly at the film interface, where the plasma species interact with the surface. As a result of the interaction of oxygen plasma species and temperature, iron is oxidized to different polymorphs depending on the operating pressure and hence oxygen gas flow rate. The phase, as well as the morphology of the film was controlled by monitoring the oxygen flow rate using the unique Plasma-Assisted Thermal Oxidation (PATO) process. Different polymorphs, viz., Fe3O4, γ-Fe2O3, α-Fe2O3 and different morphologies, such as polygonal, compact facets, wire-like (1D) nanostructures at the surface were obtained for the films processed using PATO. The selected PATO-processed films were investigated for Field Electron Emission (FEE) properties. The 1D-grown surface of iron oxide obtained from the M-Fe film showed a turn-on field of 3 MV m−1 and emission current of 337 μA cm−2, whereas the pyramidal surface morphology obtained using N-Fe film gives a turn-on field of 3.3 MV m−1 with an emission current of 578 μA cm−2.


Introduction
Plasma, especially non-thermal plasma, has been used in the medical, 1 textile 2,3 and food processing 4-6 industries for more than a decade for sterilization, effluent treatment, surface cleaning, etc. It is also used in automobile industries for surface nitridation, 7 carbo-nitridation and carbonation of mechanical tools, also known as surface hardening. It is known that a typical type of non-thermal plasma 8 is used, depending on the application requirement. The non-thermal plasmas, viz., DC plasma, 9 RF plasma, 10 microwave plasma 11 and ECR plasma, have been explored for applications like thin lm deposition, surface modication, surface functionalization, surface hardening, etc. However, of these, the ECR plasma is relatively less explored. The ECR plasma reactor has its advantages as it operates at relatively lower pressure, higher electron density, requires no electrodes to generate plasma, and does not produce toxic gases or hazardous byproducts during the process, which make it an environmentally friendly process. The rst generation of indigenously developed ECR plasma reactor by our group has been explored for applications like the surface nitridation of GaAs, 12 En-41B steel 13 and M2 steel 14 using H 2 + N 2 (HN) plasma. The same ECR plasma reactor was used to deposit nano-crystalline diamond lms. 15 Appropriately biased hollow cathodic cylinders of Zn 16 and Mo 17 were introduced individually into ECR plasma to obtain oxide nanostructures and these nanostructures were further used for scanning tunneling microscopy (STM) studies. The ECR plasma source was used for the thin lm deposition of Ti and Fe-doped Ti, 18 as well as the surface modication of biocompatible polymers like poly(etherimide), 19 suitable for tissue engineering applications.
The presently used ECR plasma reactor is a modied version of the 1 st generation ECR plasma reactor that was reported earlier 20 with detailed diagnostics of the ECR plasma system for the spatial distribution of plasma properties, mainly the electron temperature (T e ), plasma density (n e ), etc., using a Langmuir probe. Knowledge of plasma properties is quite an important factor that is used to understand the materials processing. Further, the effects of the plasma species generated using Ar, O 2 and HN on nylon 6 have been studied. 20 Further, the ECR plasma reactor was used to modify the surface of UHMWPE (ultra high molecular weight polyethylene) using O 2 and HN plasma in order to study the adhesion and proliferation of bone-associated cells. The results demonstrate that the plasma treatment time was a sensitive parameter for dening the bone cell proliferation. 21 In addition, the inuence of the oxygen plasma treatment on the solar energy conversion performance of the porous ZnO-based dye-sensitized solar cells was studied. 22 To widen the utility of the present ECR plasma reactor, the manuscript mainly focuses on the feasibility of utilizing it for materials processing, in particular, to tune the surface morphology for FEE applications. The feasibility of the earlier ECR plasma reactor was investigated using hollow cathode-biased cylinders of Mo 17 and Ag 23 introduced individually into the plasma reactor to obtain oxide nanostructures and further used in eld emission microscopy (FEM) studies. Similarly, Kar et al. demonstrated the use of an ECR plasma reactor to grow carbon nanotubes on an Inconel substrate with varied process parameters to investigate the eld emission behaviour. 24 The cathode materials used in the electron guns of many devices have been a topic of interest from the viewpoint of basic understanding, as well as technological developments, for many years. To improve the performance of eld emitters, nanoscale materials have been used. 10,[25][26][27][28][29][30][31] Nanostructured materials possess high aspect ratios, as well as high surface activity, and are, therefore, suitable for applications like eldeffect electron emission, gas sensors, 32 catalysts, 33,34 magnetic storage devices, 35 anode material for lithium-ion batteries, 36 thermoelectric power generators, 37 nuclear radiation sensors, 38 etc. Looking at the eld emission properties of nanostructures grown on surfaces, researchers have reported that carbon nanotubes are excellent eld emitters. 24,39 However, onedimensional (1D) materials like carbon nanotubes require prolonged treatment at high temperatures and the eld emission current (performance) is degraded with time due to surface oxidation. To overcome this problem, various stable metal oxide nanomaterials like ZnO, 40 WO 3 , 41 CuO, 42 and Fe 2 O 3 10,25,30,43,44 have been synthesized and used as eld-effect emitters. Out of the various nano-materials, iron oxide has attracted the most attention due to its environmental friendliness, non-toxicity, excellent thermal stability, and low cost. 33,45 Iron oxide has 3 prominent polymorphs, viz., Fe 3 O 4 , g-Fe 2 O 3 , and a-Fe 2 O 3 , of which a-Fe 2 O 3 is the most stable under various ambient conditions, with n-type semiconductor behaviour and a band gap of 2.2 eV. These stable nanostructures have been synthesized using various methods like thermal oxidation, 30,35,46,47 plasma oxidation, 31,48,49 sol-gel-mediated reaction, 50 hydrothermal reaction, 51,52 the microwave-assisted hydrothermal method, 53-55 template methods, 56 chemical vapour deposition, 39 electrochemical deposition, 28,57 solvothermal deposition, 58,59 etc. It is noted from the literature that the eld emission properties mainly depend on the method of synthesis used to grow the hematite phase. In some cases, post processing is required to achieve a stable hematite phase. Table 1 summarizes the reported methods used to grow the hematite phase; more specically, the types of methods, processing time and eld emission properties. To improve the eld emission performance, Junqing et al. 25 reported that the sample needs additional current aging treatment, giving rise to a reduced threshold eld of 6.6 MV m À1 . Similarly, Wu et al. 43 mentioned a further decrease in the threshold eld to 7.2 MV m À1 once the samples were subjected to X-ray irradiation (a dose of 9.0 Â 10 14 phs cm À2 ). As mentioned by Liang Li, 44 the pulse laser-deposited lm needs further post-processing at 450 C for 3 h and gives the electron emission properties as mentioned in Table 1. The synthesis process reported by Zheng et al. 10 required 15 h of conventional thermal oxidation at 260 C and further RF oxygen plasma treatment to improve the eld emission properties. On the other hand, Li-Chieh et al. 30 performed the conventional thermal oxidation of iron lms having different thicknesses to grow the nanowires, where the substrate lm thickness and growth of the nanowires per unit area dened the emission properties.
As mentioned above, ECR plasma can produce different plasma species such as electrons, ions, atoms, molecules, etc., having different energies and densities. Also, the reactivity of the ionic plasma species generated in the ECR plasma reactor is expected to be different, which is sensitive to partial pressure and temperature at the reactive sites. Therefore, in the present study, we focus on the interaction of ECR plasma species leading to the phase tuning, as well as morphology tuning, of the surface of metal lms made up of micron-and nano-sized iron powders under suitable conditions. The ECR plasma being cold plasma, the interaction of plasma species with the iron surface was not seen clearly at room temperature because of the polymer coating on the precursor iron powders used during the lm formation. Therefore, a radiation heater (RH) was developed to raise the surface temperature of lms inside the ECR plasma reactor. The surface of the metal lms was oxidized using low-pressure oxygen plasma and rapid thermal heating with the heating rate of 12 C s À1 . The ECR plasma produced atomic, molecular, as well as ionic oxygen species, and elevated temperature facilitated the oxidation of the iron surface. The PATO process was carried out in a closed and controlled environment. The optimized sets of lms were characterized thoroughly using X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, X-ray diffraction (XRD) and Field Emission-Scanning Electron Microscopy (FE-SEM). The eld electron emission properties of selected lms obtained by PATO were investigated.

Experimental section
2.1. Synthesis of iron nanoparticles using the thermal plasma route The iron nanoparticles were synthesized using a transferred arc thermal plasma reactor. The thermal plasma reactor consists of an anode, which also acts as the sample holder for the precursor, a cathode enclosed in plasma torch, and the whole assembly was enclosed in a double-walled water-cooled plasma reactor. It was possible to vary operating pressure inside the plasma reactor using a rotary pump. The micron-sized iron particles (LOBA, $24 to 60 microns, purity of 99.5% electrolytic grade) in the form of pellets were kept beneath the plasma plume having a length of 5-10 cm and diameter of about 1-2 cm with high thermal ux. The operating pressure during the synthesis was maintained at 1000 torr. The iron metal species were gasied due to the sufficiently high temperature and experienced rapid thermal quenching, i.e., the temperature decreased from $10 000 K to $1500 K within a 15-20 mm region from the center of the plasma plume. Due to the sharp temperature gradient, the evaporated species nucleated and grew in the plasma peripheral region. The formed nanocrystalline powder settled on the inner sides of the reactor chamber and was later scraped. The structural and morphological analysis indicated the formation of the BCC phase of iron with a maximum number of particles having a size of around 30 nm. The details of the synthesis and characterization of the iron nanoparticles were reported elsewhere. 60

Preparation of M-Fe and N-Fe lms
Micron-sized iron powder (M-Fe) ($24 to 60 microns, purity of 99.5% electrolytic grade) and nano-sized iron powder synthesized using the thermal plasma process (N-Fe, average particle size $ 30 nm) were used as a precursor material to prepare the slurries. The process followed to prepare the slurry and thick lms is shown in Fig. 1. Ethanol and acetylacetone were procured from Hayman (premium grade 100%) and Loba Chemie (99.5% pure), respectively. The owchart presented in Fig. 1 was followed to obtain the lms. Here, ethylcellulose was used as a binder to bind the precursor particles, which facilitated the adhesion of the precursor particles to the substrate, as well as the oxidation process. The choice of ethyl cellulose as a binder is based on the reports in the literature. 61

Radiation heater (RH)
The radiation heater (RH) was indigenously developed by using halogen lamps (PHILIPS 24 V/250 W, Projection lamp Type 13163). The RH consists of two lamps mounted on the circumference of the stainless steel ring of radius 10 cm. The lamps were mounted in such-a-way that the radiation was focused at the center of the ring where the lm was kept. The focal point of each lamp was $3.5 cm away from the lament. The lm was positioned (Fig. 2a) exactly at the center of the bright focal point of the lamps. The lamps were powered up using a specially designed step down transformer (24 V/10 A) connected through a dimmerstat. Fig. 2b shows the actual photograph of the RH heater assembly mounted inside the plasma reactor. The temperature prole produced by RH was recorded at the focal point as a function of the input power at the base pressure of 10 À5 mbar and 5 Â 10 À3 mbar. Fig. 2c shows the plot of the temperature proles obtained using RH under various operating conditions. Typically, the temperature of 850 AE 30 C was achieved by operating both the lamps at 230 W within 1 min.

ECR plasma reactor
The ECR plasma reactor was used to generate the oxygen plasma species. ECR conditions were achieved by using microwave radiations of frequency 2.45 GHz and a DC magnetic eld of 875 gauss. An electromagnet comprised of a pair of solenoids was used to generate the magnetic eld >875 G. Before generating the plasma, the system was evacuated to a base pressure of 10 À5 mbar using a turbo-molecular pump backed by a rotary pump and later, oxygen gas was lled in the reactor till the desired operating pressure of up to 1 to 15 Â 10 À3 mbar was reached. The resonance of the microwave eld and magnetic eld resulted in the formation of the glow discharge, where the ECR resonance condition was satised between the cyclotron motion of electrons (Lorentz force) and the input microwave frequency. The optical emission spectroscopic analysis of the oxygen ECR plasma indicated the presence of atomic oxygen species as prominent species along with other reactive species. To determine the plasma properties such as n e , T e , Debye length (l D ), etc., the Langmuir probe method was used. T e was found to vary in the range of 10-12 eV, n e was about 10 17 m À3 , with l D of about 20-100 mm as the Langmuir Probe moved from 15 to 31 cm away from the ECR zone. Further, the Electron Energy Distribution Function (EEDF) analysis indicated that the maximum value of T e varied in the range of 14-22 eV with an increase in the distance of the Langmuir probe from the ECR zone with a wide range of energy distribution. The detailed mapping of the ECR plasma properties was reported elsewhere. 20 The axial mapping of the spectroscopic measurement conrmed the presence of different plasma species, viz., atomic, molecular and ionic species. The M-Fe or N-Fe lms were kept 23 cm away from the ECR zone, where the focal points of the RH coincide and the estimated n e was found to be of the order of 10 17 m À3 as determined using the electrostatic probe method. Films made up of micro-and nano-sized precursor iron powders, designated respectively as M-Fe and N-Fe, were subjected to PATO. PATO was carried out in the presence of oxygen gas with different ow rates (5-200 sccm) and hence partial pressures of oxygen. The operating pressure during the PATO of M-Fe and N-Fe lms was varied in the range of 1 to 15 Â 10 À3 mbar. At a given operating pressure, oxygen plasma was generated and then RH was switched ON to raise the temperature to 750 C. The total time required to attain the desired temperature of 750 C was about a minute with a heating rate of 12 C s À1 , and further oxidation was carried out for 9 min. Variation of the oxygen ow rate and hence the operating pressure at elevated temperature resulted in phase and morphological tuning. The architected iron oxide surface was used to investigate the FEE properties. The operating conditions and lm designation are given in Table 2.

Characterization techniques
The chemical states aer the PATO processing of M-Fe/N-Fe lms were investigated by using the surface-sensitive X-ray Photoelectron Spectrometer (XPS, model VersaProbe III from Physical Electronics ULVAC-PH). The as-prepared M-Fe and N-Fe lms were characterized for their structural properties using the X-ray diffraction (XRD) technique where a Bruker AXS D8 Advance X-ray diffractometer with Cu-Ka radiation was used to record the XRD patterns before and aer the PATO process. The surface morphology of the lms was investigated using Field Emission Scanning Electron Microscope (FE-SEM, Carl-Zeiss MERLIN FE-SEM). The surface-sensitive Raman spectrometer (Renishaw inVia Raman Microscope) was used to investigate the polymorphs, viz., hematite, magnetite, maghemite, etc., formed at the surface. A laser having a wavelength of 532 nm with a power of 0.5% was used as an excitation source and Raman spectra were recorded in the range of 100-3200 cm À1 .
Further, the eld emission properties for the morphologically tuned M-Fe and N-Fe lms were investigated by using a planar diode conguration. Initially, the lm of interest was xed/stuck onto a copper rod (acts as a cathode) using carbon tape. The rod was connected to a linear motion drive, which was then used to adjust the cathode-anode separation during the eld emission (FE) measurements. A typical diode conguration was used, where a semi-transparent cathodoluminescent phosphor screen (diameter $50 mm) was held parallel to the cathode. The FE working chamber was closed and evacuated to 1 Â 10 À8 mbar of base pressure using appropriate vacuum systems. The FE measurements were carried out at a constant cathode-anode separation of 1 mm (1000 mm), the emission current was recorded using a KEITHLEY electrometer (model 6514) by varying the applied voltage between the cathode and anode in the range of 0-40 kV (Spellman, USA), having step size of 40 V.

Results and discussion
3.1. Plasma-assisted thermal oxidation (PATO) process 3.1.1. XPS analysis. Fig. 3 shows the high-resolution XPS spectra recorded for the M-Fe oxidized lms in the energy range corresponding to characteristics of Fe2p 3/2 and Fe2p 1/2 of iron oxide. These peaks were deconvoluted to investigate the polymorphs of iron oxide present at the surface of the M-S1 to M-S3 lms. ESI Table 1 † shows the summary of deconvoluted peaks corresponding to the Fe2p states and satellite peaks.
The XPS peak corresponding to Fe2p state shows two prominent peaks associated with Fe2p 3/2 and Fe2p 1/2 . The peaks observed at around 710.7 eV and 724.5 eV correspond to Fe2p 3/2 and Fe2p 1/2 , respectively, 62-64 whereas broad peaks observed at 718.8 eV and 733 eV are the shake-up satellite peaks of Fe2p 3/2 and Fe2p 1/2 , respectively. 65 Satellites associated with the Fe2p core level spectra were used to determine the oxidation states of iron. 63 It was mentioned by Radu et al. 66 that the clearly visible satellite peak at 718 eV indicates the presence of maghemite or hematite, and the absence of the satellite peak indicates the presence of the magnetite phase. 67,68 Fig. 3a-c shows the presence of a satellite peak (between 711 and 724 eV) at 718 eV with increasing intensity for M-S1 to M-S3 lms, indicating the presence of the maghemite or hematite phase. Fujii et al. reported that there was hardly any difference between the XPS spectra of hematite and maghemite; however, there are certain points that can distinguish between these phases. The intensity ratio of the satellite observed at 718 eV with the main peak of Fe2p 3/2 observed at 711 eV was less for maghemite as compared to hematite, and secondly, the peak position of 2p 3/2 was slightly shied towards lower binding energy in case of maghemite with respect to hematite. 67 Therefore, to distinguish between maghemite/hematite, the XPS lines corresponding to the Fe2p spectrum needed to be carefully analyzed. From Fig. 3 and ESI Table 1, † the peak observed at 719 and 733 eV represents the signature of satellite peaks; the area under the tted curve greatly increased from M-S1 to M-S3. The highest intensity/area of satellite peaks with the prominent Fe2p peak of the M-S3 lm represents the hematite phase, conrming that the Fe 3+ cations are octahedrally coordinated in a crystal structure. 66 A comparison of the XPS spectra of lm M-S1 and M-S3 showed that the broad peak observed at 719 eV had a very low intensity and was slightly shied to the lower binding energy of the Fe2p 3/2 main peak, indicating the presence of the maghemite phase according to Fujii et al. 67 The high-resolution XPS spectra recorded in the energy range corresponding to the Fe2p state for N-Fe lms processed using PATO are shown in Fig. 4. The XPS tted peak position, FWHM and the area under the curve were tabulated for the respective peaks and are shown in ESI Table 2. † The Fe2p highresolution XPS (Fig. 4a) for the N-S0 lm shows the poor signature of the 2p 3/2 and 2p 1/2 peak at around 710 eV and 724 eV along with the almost negligible signature of the satellite peaks, indicating the presence of the magnetite phase. Similarly, the N-S1 (Fig. 4b) lm showed the prominent signature of the 2p 3/2 and 2p 1/2 peaks but the absence of the satellite peak at 718 eV. 57,66,67 Further, XPS recorded for N-S2 (Fig. 4c) and N-S3 (Fig. 4d) indicated the presence of satellite peaks along with the main peaks corresponding to Fe2p 3/2 and Fe2p 1/2 . Based on the XPS data, it was inferred that the surface polymorph of the N-S2 and N-S3 lms was the hematite phase. The intensity/area under the curve of the Fe 3+ satellite peak (719 eV) increased but that of Fe 2+ satellite peak (733 eV) decreased in N-S3 as compared to N-S2. The N-S2 and N-S3 lms have similar features to that of the M-S3 lm, indicating the presence of the hematite phase at the surface. The signicant difference between Fe 3 O 4 and a-Fe 2 O 3 phases was indicated by the doublet peak of Fe2p 3/2 and Fe2p 1/2 of the Fe2p spectral position. The shi in the Fe2p doublet peak by 0.5 eV towards lower binding energy represents the magnetite phase (N-S1) as compared to the hematite phase (N-S2/N-S3); this spectral shi 66 is presented in ESI Table 2. † The effect of oxygen pressure on N-Fe lms was distinguishable using XPS analysis, unlike M-Fe lms.
3.1.2. Raman spectroscopic analysis. In order to understand the surface stabilized polymorphs of iron oxide more clearly aer the lms were processed using PATO, another surface-sensitive spectroscopic technique, viz., Raman spectroscopy, was used. Raman spectroscopic analysis was used to conrm the phases and distinguish them. Fig. 5a and b show the Raman spectra for M-Fe and N-Fe lms, respectively, processed by PATO. Table 3 represents the Raman active modes associated with different polymorphs of iron oxide. 69 The Raman spectra in the range of 100-1800 cm À1 for the M-Fe lms processed by PATO are shown in Fig. 5a. In the case of lms M-S1 to M-S3, Raman peaks were observed at 226 cm À1 (A 1g ), 245 cm À1 (E g ), 292 cm À1 (E g ), 411 cm À1 (E g ), 491 cm À1 (A 1g ) and 501 cm À1 (T 2g ) belong to the active modes 67 of the hematite phase as shown in Table 3. Here, the Raman active mode is indicated in the parentheses. Another peak was observed at 612 cm À1 (E g ) for M-S2 and M-S3, assigned to the hematite phase. The Raman peak observed at 1324 cm À1 in these lms was due to the two-magnon scattering of hematite, which is not the feature of the magnetite or maghemite phase. 70 The Raman active modes present in the M-S1 to M-S3 showed hematite as a prominent phase. The careful observation of the spectra depicted a broad peak present in the range of 580-757 cm À1 in the case of M-S1, which weakened in the case of the M-S3 lm with a broad peak at 666 cm À1 in both cases. The weak Raman peak observed at 666 cm À1 is associated with the A 1g Raman active mode of the magnetite phase. A few more distinct Raman active peaks were observed in the case of M-S1 at 560 cm À1 71 and 700 cm À1 , 72 which were assigned to the T 2g mode of the magnetite and A 1g mode of the maghemite phases of iron oxide, respectively. Raman spectroscopic data for M-Fe lms are summarized in Table 3, depicting the strong signature of the hematite phase in all the lms of M-Fe (M-S1 to M-S3). A weak signature of magnetite and maghemite was seen in the case of M-S1, whereas a weak signature of magnetite only was seen in the M-S2 and M-S3 lms. Fig. 5b represents the Raman spectra for PATO-processed N-Fe lms as a function of operating pressure in the ECR plasma reactor due to the increased oxygen ow rate. As depicted in Table 3, Raman peaks are used to identify different polymorphs present at the surface of the N-S0 to N-S3 lms. The broad peak at 318 cm À1 (E g ) indicates the Raman-active band of the maghemite phase, and other peaks at 590 cm À1 (T 2g ) 74 and 661 cm À1 (A 1g ) belong to the magnetite phase present in the N-S0 lm. No peaks were observed at higher wavenumbers, and hence the N-S0 lm showed the prominent phase of magnetite with maghemite as a secondary (weak) phase. In the case of N-S1, the rst broad peak was shied to a higher wavenumber as compared to the N-S0 with a peak position at 350 cm À1 , and the other peak at 701 cm À1 (A 1g ) represents the maghemite phase, whereas the peak observed at 590 cm À1 (T 2g ) indicated the presence of the magnetite phase. Hence, the N-S1 lm consists of magnetite and maghemite phases at the surface, with magnetite as a prominent phase. From the N-S2 lm, the Raman active modes observed at 226 cm À1 , 244 cm À1 , 291 cm À1 , 410 cm À1 , 498 cm À1 and 608 cm À1 were the signature of the hematite phase. Here, a broad peak was observed at 608 cm À1 , ranging from 527-764 cm À1 ; the broadening of the peak may be due to the presence of secondary phases like magnetite/maghemite and hence, the N-S2 lm does not consist of only the hematite phase at the surface. For the lm synthesized at higher oxygen pressure, i.e., the N-S3 lm, the Raman active peaks were similar to those of the N-S2 lm with an additional weak Raman active band observed at 661 cm À1 belonging to the magnetite phase. In the case of the N-S3 lm, prominent signatures of a hematite phase were observed. Interestingly, the peak at 1324 cm À1 was seen only in N-S2 and N-S3 lms, which conrmed the presence of the hematite phase (Fig. 5b). Overall, the Raman analysis reected strongly about the phases of iron oxide observed aer PATO processing, especially in the case of M-Fe. Moreover, Raman analysis suggested the mixed phases observed at the respective surfaces, which was difficult to interpret from XPS analysis.  3.1.3. X-Ray diffraction analysis. Surface analysis is essential to understanding the surface properties; however, the surface and bulk may behave differently in general and more specically, during the oxidation of iron using the PATO process. Since an oxidative environment is provided to iron powder mixed with ethylcellulose lms to grow the polymorphs of iron oxide, it was expected to show different bulk properties. Depending on the temperature and oxygen abundance, different polymorphs of iron oxide, in different parts of the grain were expected to be present. X-ray diffraction was employed to understand the bulk characteristics of the PATOprocessed lms. Fig. 6 show the X-ray diffraction patterns of M-Fe and N-Fe lms recorded aer PATO processing. It is clear from the Raman spectroscopic analysis that the M-Fe lms consist of hematite as a prominent phase, with a secondary (weak) phase of magnetite/maghemite. The X-ray diffraction lines (Fig. 6a) observed for the lms M-S1, M-S2 and M-S3 indicate the formation of the hematite phase. In the case of the M-S2 lm, the hematite phase seemed to be the prominent phase, whereas the M-S3 lm indicated the prominent phase of magnetite/maghemite and hematite as the secondary phase, which supplements the inferences from Raman analysis (Fig. 5a). Careful analysis of the XRD pattern of the M-S3 lm indicated the presence of metallic iron along with the oxide phase. This observation indicates the incomplete oxidation of iron lms. Generally, the oxidation of metallic iron takes place in the following sequence, Fe / FeO / Fe 3 O 4 / g-Fe 2 O 3 / a-Fe 2 O 3 . This indicates in the case of the M-S3 lm that the oxidation process was incomplete. During the process of the hematite phase formation from iron through magnetite and maghemite, oxygen was liberated. The incomplete reaction resulted in the formation of the mixed-phase at the surface. It was reported that at the temperature of about 300 C, the magnetite phase was transformed into maghemite, which is the metastable phase. 74,75 Both magnetite and maghemite possess the spinel cubic crystal structure; therefore, by a topotactic transition, maghemite was formed on top of the magnetite surface. As the temperature increased above 500 C, the magnetite/maghemite was transformed into the hematite phase and their crystal structures were totally different. The difference in the crystal structure introduced the possibility of stressinduced morphology. Fig. 6b shows the X-ray diffraction patterns for the N-Fe lm analyzed aer PATO processing. The N-S0, N-S1, N-S2 and N-S3 lms were processed by PATO under oxygen as the plasmaforming gas with ow rates of 5, 15, 30 and 100 sccm, respectively. From Fig. 6b, the X-ray diffraction pattern for the N-S0 lm depicts the formation of magnetite (Fe 3 O 4 ) as the prominent phase; for the N-S1 lm, the diffraction lines show magnetite/maghemite as the prominent phase along with traces of the hematite (a-Fe 2 O 3 ) secondary phase. In the case of the N-S2 lm, the diffraction lines showed hematite as the prominent phase along with Fe 3 O 4 /g-Fe 2 O 3 as the secondary phase. The diffraction lines of the N-S3 lm show only hematite as the prominent phase. This trend of phase formations was expected because as the oxygen pressure increased, the iron oxide phase transformation occurred from the magnetite to the hematite phase. From Fig. 6, it is interesting to note that the oxygen pressures of 30 sccm and 100 sccm-processed M-Fe and N-Fe lms, viz., M-S1, M-S2 and N-S2, N-S3, showed hematite as a prominent phase. The higher oxygen pressure in the case of M-Fe and lower oxygen pressure in the case of N-Fe showed magnetite as a prominent phase.
3.1.4. Morphological analysis. Fig. 7 shows the FE-SEM micrographs for M-Fe and N-Fe lms processed by PATO. Before the PATO process, iron particles were passivated by an ethylcellulose matrix. In the temperature range of 200-500 C, the ethylcellulose matrix was dissociated and evaporated and at that instant, the oxygen plasma species interacted with iron to form different polymorphs as well as morphologies of iron oxide, depending on the surface reactivity. Fig. 7 indicates the micrographs of M-Fe and N-Fe lms processed by PATO at optimized conditions. ESI Fig. 1 † shows FE-SEM pictures of all the remaining lms of M-Fe and N-Fe processed by PATO in the present study. The distinct morphology of the onedimensionally grown structure on the surface of iron oxide particles was observed in the case of M-S3. The average length of the one-dimensionally grown structure was found to be in the range of 1.2-1.8 mm with a diameter of $0.1 mm. It was also noted that the wire-like growth was not unidirectional. The pyramidal, well-faceted morphology with nano-sized grains was seen in the case of lm N-S2. The pyramidal morphology had dimensions in the range of 0.3-0.5 mm.

Phase and morphology tuning
During the PATO process, the lms were exposed to oxygen plasma consisting of plasma species, viz., atomic oxygen and ionic species of oxygen, for 10 min. The key observations based on the experimental results obtained using X-ray diffraction, Raman spectroscopy, X-ray photoelectron spectroscopy and microstructural analysis using FE-SEM are as follows. (i) The structural analysis of M-Fe lms indicated the formation of (a) the hematite phase up to the oxygen ow rate of 100 sccm (M-S1) with maghemite/magnetite as a secondary phase, and (b) a mixed phase of hematite (weak phase) and magnetite/ maghemite as a prominent phase for an oxygen ow (iii) XPS analysis of all three lms, M-S1 to M-S3, indicated that they were well dened, and were assigned to the maghemite to hematite phase at the surface. (iv) The morphology of M-S1 was found to be like cauliower, M-S2 like cabbage, and quite dense onedimensional growth was observed at the surface of the M-S3 lm. On investigating the N-Fe lms processed using PATO, (v) the structural analysis based on X-ray diffraction depicted the following. (a) For N-S0, the lm formation of a well-dened magnetite phase; (b) for N-S1, the formation of magnetite as a prominent phase along with the signature of hematite; (c) for N-S2, the trend was found to be reversed with hematite as a prominent phase along with the superimposed magnetite/ maghemite phase; (d) for N-S3, the complete hematite phase was observed. (vi) Raman spectroscopic analysis indicated the presence of the magnetite phase, in the cases of N-S0 and N-S1, as the prominent phase, with a mixture of maghemite being observed in N-S0 and N-S1. In the case of N-S2, a prominent signature of hematite was observed along with a weak signature of magnetite, and nally, in the case of N-S3, a single hematite phase was observed. (vii) The weakly dened XPS spectrum of N-S0 was due to the magnetite phase, which was found to be strengthened in the case of N-S1. The presence of a satellite peak at 718 eV indicated the hematite phase formation at the surfaces of N-S2 and N-S3. (viii) The morphologies of N-S0 and N-S1 were polygonal, with well-grown grains in the case of N-S1. These grains had pores. The grains of N-S2 were small but wellfaceted and grains of N-S3 were small and blunt. Fig. 8 gives the summary of the bulk, surface and morphological properties of M-Fe and N-Fe lms processed by PATO under different oxygen pressure conditions.
The structural and morphological properties observed can be explained as follows: usually, the Fe 3 O 4 (magnetite) phase is formed at a relatively low temperature and in an oxygen-starving atmosphere. Magnetite undergoes oxidation and is transformed into a-Fe 2 O 3 through g-Fe 2 O 3 (maghemite) phase as a result of heat treatment. It is given as Fe 3 Fig. 7 The Field Emission-Scanning Electron Micrographs (FE-SEM) recorded after PATO processing for M-S3 and N-S2 films kept at different processing conditions as mentioned in Table 2, with a 1 mm scale and magnification of Â30 000 (each inset shows the magnified micrograph of the respective film surface; the rest of the films processed under PATO are shown in ESI Fig. 1 †).
This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 32088-32101 | 32097 Paper RSC Advances O 3 2À ) and a vacancy is created however, the spinel crystal structure is maintained. The formation of a-Fe 2 O 3 occurs through the oxidation of magnetite/maghemite. Firstly, oxygen species at the surface create iron vacancies and secondly, iron at the core diffuses through the interfaces of magnetite, and these are consumed during the formation of the hematite phase.
In the present case, the formation of different polymorphs of iron oxide is facilitated by the diffusion of oxygen into the iron grain and outward diffusion of iron from the core. During the oxidation process, Fe atoms at the surface gets oxidized and are subjected to a concentration gradient, and hence Fe atoms continuously diffuse out from the core to the surface. Moreover, Fe atoms become ionized during the diffusion process normal to the surface where Fe 2+ /Fe 3+ diffuse outward (outward diffusion coefficient: 9.7 Â 10 À15 cm 2 s À1 ) and O 2À diffuses inward (inward diffusion coefficient: 5.2 Â 10 À16 cm 2 s À1 ). This ionic diffusion happens through vacancy exchange rather than direct atom exchange. 76 In the present case, the role of ethylcellulose, the reactivity of the particles due to different surface to volume ratio, surface temperature and oxygen partial pressure are decisive parameters for the structural and morphological properties. In the case of bare particles (both micron-sized as well as nano-sized), the diffusion of oxygen and iron into each other is quite fast and results in the formation of a stable hematite phase. However, the phase and morphology were found to be different in the presence of ethylcellulose for PATO-processed lms. In the case of M-Fe lms processed by PATO, the diffusion process facilitated the formation of hematite when the abundance of oxygen was low (i.e. up to the ow rate of 100 sccm), whereas for a higher ow rate, viz., 200 sccm, the formation of hematite on magnetite was facilitated.
It is interesting to note the different morphologies observed in the case of M-Fe and N-Fe lms. Among the lms investigated, whisker growth was observed in the case of M-S3, whereas a faceted morphology was observed in the case of N-S2. The growth of a-Fe 2 O 3 whiskers followed the same mechanism as that of CuO nanowire formation reported by Yuan et al. 77 The growth of a-Fe 2 O 3 whiskers is mainly associated with the stress generation and relaxation at the a-Fe 2 O 3 /Fe 3 O 4 interface. With decreasing oxygen partial pressure towards the Fe core of the grain in the presence of the ethylcellulose matrix, the oxidation rate was slowed down and the Fe 3 O 4 /g-Fe 2 O 3 and a-Fe 2 O 3 interfacial reaction rate was reduced, which resulted in a small stress gradient across the a-Fe 2 O 3 layer. Inuenced by this driving force, the Fe cations diffused along the grain boundary region and were deposited at the bottom of the a-Fe 2 O 3 phase. The a-Fe 2 O 3 layer at the surface grows at the expense of the thin Fe 3 O 4 /g-Fe 2 O 3 layer at the interface. It was reported that the Pilling and Bedworth ratios (the ratio of the volume of metal oxide to the volume of consumed metal) for FeO, Fe 3 O 4 and Fe 2 O 3 were 1.68, 2.10 and 2.14, respectively. 77 Since the specic volume of Fe 3 O 4 /g-Fe 2 O 3 was smaller than that of Fe 2 O 3 , the compressive stresses were generated and accumulated at the bottom of the Fe 2 O 3 layer. This led to an increased stress gradient and facilitated the outward diffusion and promoted the delivery of Fe cations onto the top of Fe 2 O 3 grains via combined grain boundary and surface diffusion, where the grain surface served as structure templates for the nucleation of Fe 2 O 3 whiskers. 77 The drastic growth of the a-Fe 2 O 3 layer at the surface in the presence of ethylcellulose slowed down the diffusion of oxygen to the core and hence, in the case of M-S3, unreacted Fe traces were observed in the X-ray diffraction pattern. Due to the sufficiently high reactivity of nanoparticles, this feature was not observed in N-Fe lms.

Field emission (FE) analysis
The eld emission electron ux recorded on the phosphorouscoated conducting screen (on the anode) as a function of the applied electric eld for M-S3 and N-S2 lms is shown in Fig. 9a. It was seen that the turn-on eld for the M-S3 lm was low as compared to the N-Fe lm, whereas the turn-on eld for the N-S2 lm was low among N-Fe lms. Table 4 shows the summary of the eld emission properties of the selected lms of M-Fe and N-Fe processed by PATO. Among all of these lms under investigation, the M-S3 lm with moderate whiskers showed a low turn-on eld. Nanostructure morphologies, more specifically, one-dimensional nanostructures are expected to show superior eld emission properties. However, highly dense emission sites show poor eld emission performance. 30 Li-Chieh et al. 30 reported that the increases in the population density of nanowires per square centimeter indicate the drastic change in the turn-on eld, which supports the experimentally observed results in the present case.
Field emission properties for N-Fe lms are shown in Fig. 9a and b. The PATO-processed N-S2 lm having a nanostructured facet (pyramidal morphology) possesses a low turn-on eld of 3.6 MV m À1 with higher current density (578 mA cm À2 ). The current density in N-S0 was highest in N-Fe because of its surface being stabilized with the magnetite phase, which is a better conductor than the hematite phase. Different morphologies of iron oxide grown over M-Fe and N-Fe-based M-S3 and N-S2 lms were found to be good eld emitters among the lms under investigation.
On comparison of the experimentally observed eld emission properties (Table 4) with those reported in the literature as summarized in Table 1, it was seen that the PATO-processed hematite surface had the better eld emission performance as compared to that mentioned in Table 1 without any postprocessing. Moreover, PATO provides a controlled oxygen environment, green process, processing in a clean environment, less processing time and in situ heating to form the desired polymorph of iron oxide at the surface as compared to the methods listed in Table 1. The role of organic matter in encapsulating the iron precursor cannot be avoided. Usually, the iron oxide surface requires prolonged treatment at elevated temperatures to obtain the hematite phase and the typical morphology suitable for eldeffect electron emission. 10,25,30,43,44 From Table 4, PATO-processed lms revealed that the method of processing strongly affects the eld emission performance; the PATO-processed lms with unique morphologies were better than the conventionally processed lms. The difference in eld emission performance in the PATO-processed M-Fe and N-Fe lms was due to the surfacestabilized phase and morphology obtained in a given oxygen  This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 32088-32101 | 32099 environment. Moreover, it is well known that the surface with the higher aspect ratio always demonstrates better eld emission behaviour; hence, M-S3 indicated a turn-on eld of 3 MV m À1 as compared to that of N-S2, showing a wire-like and pyramidal morphology, respectively.

Conclusion
The present work sheds light on the phase-and morphologytuning of iron oxide using the Plasma Assisted Thermal Oxidation (PATO) process via oxygen ECR plasma, which is the rst of its kind. With ECR plasma being cold plasma, a radiation heater was designed, which operates at low pressure and is capable of raising the surface temperature to the desired value within a minute. Low operating pressure, less processing time, rapid heating and controlled oxygen environment are the features of the PATO process used for tuning the specic phase and morphology of iron oxide. For the M-Fe lm, the phase transformation from hematite to the mixed iron oxide phase, and the morphology tuning from polygonal to whiskers were observed on increasing the oxygen ow rate from 30-200 sccm. The N-Fe lms showed a phase transformation from magnetite to hematite through maghemite with an increased oxygen ow rate from 5-100 sccm. In the PATO-processed iron oxide lms, the stress generated at the bottom of a-Fe 2 O 3 seems to be the driving force for morphological tuning, whereas temperature and partial pressure of oxygen are required for phase tuning. The PATOprocessed lms M-S3 and N-S2 are comprised of mixed phases of iron oxide as analyzed by XRD and Raman spectra with the a-Fe 2 O 3 phase at the surface. The eld emission study showed that the PATO-processed M-S3 and N-S2 lms are the best eld emitters. In a nutshell, the present work and its analysis showcase the feasibility of the ECR plasma reactor for tuning the phase and morphology in nanostructure synthesis.

Conflicts of interest
There are no conicts to declare.