1. Introduction

Lithium niobate (LiNbO₃) is one of the most widely used complex oxides, exhibiting an important and practical set of materials functionalities such as ferroelectricity, piezoelectricity, electro-optic, and nonlinear-optical effects. A comprehensive investigation of radiation damage in such insulating oxides is important for several reasons. On one hand, certain technological applications of LiNbO₃ require the use of ion irradiation for the fabrication process [1]. On the other hand, a thorough understanding of the materials response to irradiation is essential for the design of radiation-resistant components for applications in space or extreme environmental condition such as use in nuclear irradiation systems or nuclear waste management [2]. Note that, in this connection, investigation of radiation damage in other related oxides has also been carried out. For example, studies of lattice disorder in Au-irradiated SrTiO₃ [3] and structural transformation in Xe-irradiated Gd₂Ti₂O₇ [4] have been previous investigated.

For a variety of reasons, much of this earlier work was focused primarily on damage from light-ion (He⁺) irradiation. For example, for LiNbO₃, damage during the He-ion processing, which is used for waveguide formation and selective etching [5], is of major concern in the fabrication of advanced devices. Fundamental studies of this damage have shown such light-ion bombardment leads to the formation of point defects, compositional change, long- and short-range strain and local volume swelling [6, 7]; these effects can alter the performance of electro-optical-devices [8].

However, recently, there has also been growing interest in using swift heavy ion (SHI) irradiation for materials modification and device fabrication in insulating and conducting oxides, semiconductors, polymers, and nanostructured materials such as nanowires and nanoparticles [9–12]. In general in this work, medium or higher mass ions (atomic number Z ≥ 3) with ~MeV or ~GeV energy ranges are used. Due to their large cross section for electronic energy loss, the interactions between projectile ions and the target materials are much stronger than for light ions such as H and He. Specifically it is known [13] that for swift heavy ions, even in the early portion of the ion trajectory in the target crystal, high electronic stopping power, S_e, (strong inelastic electronic excitation) dominates over nuclear stopping power, S_n, (elastic nuclear collision), forming ion tracks along the collision process. Such a major enhancement of collision rate has important consequences for applications in the fabrication of optoelectronic devices [1, 14]. For example, it has been reported [15, 16] that the required fluence (1 × 10¹² ~1 × 10¹⁴ cm⁻²) used to fabricate optical waveguides in LiNbO₃ is approximately two to three orders of magnitude smaller for heavy ions than for light-ion irradiation (~1 × 10¹⁶ cm⁻²). In addition, research has begun on interpreting the damage mechanisms from such energetic heavy ions, including Coulomb-explosion and thermal-spike models [17]. Pioneering theoretical and experimental work [18] has shown that SHI is a promising method for creating optical and electronic devices; however, comprehensive experiments comparing the individual and cumulative damage mechanisms from nuclear and electronic interactions with light and heavy ions has been absent – particularly at values of fluence below the threshold level for amorphization.

In this paper, a comprehensive and comparative study of damage in LiNbO₃ by low- and high-energy light- and heavy-ion irradiation is presented. The damage behavior for the two ion species (helium and iron) over two different energy ranges is investigated and compared using optical microscopy (OM), atomic force microscopy (AFM), Rutherford backscattering and channeling (RBS/C), micro-Raman spectroscopy, transmission and scanning transmission electron microscopy (TEM/STEM), and energy-dispersive X-ray spectroscopy (EDS). The energies are chosen such that for low-energies, the effect from S_e is minimal and the damage from S_n is dominant (except in the near-surface region). SRIM (Stopping and Range of Ions in Matter) [19] simulation is used as a guide to determine the energy range of interest. Experimentally it is found that for high-energy (~MeV) heavy-ion irradiation, both nuclear and electronic interactions contribute to damage formation. The combined interactions of both nuclear and electronic damage from swift heavy ions give rise to a lowered critical fluence for materials amorphization than for the case of low-energy (~100s keV) ions. Measurements of the evolution of damage in the materials clearly show the explicit contributions from electronic and nuclear stopping and their cumulative overlapping effects. Micro-Raman imaging is used to complement our study of lattice damage and shows that for both light- and heavy-ion irradiations, lattice disorder induces Raman activity in the normally forbidden Raman phonon-mode vibrations, though the ranges of interaction in the oxide crystals are different. TEM imaging further shows iron-ion irradiation results in the formation of nanometer-size damage pockets. Apparent changes in elemental composition have been observed in these defective regions using EDS line scans. Finally, an investigation of sequential irradiation by He⁺ and Fe⁺ was undertaken since sequential irradiation is important for ion processing and provides an additional degree of flexibility during materials processing. Our materials characterization probes show that sequential irradiation of light and heavy ions increases damage in comparison with a single-beam irradiation and can be used also to tailor the location of the damaged regions.

2. Experimental and results

2.1. Experimental

Two energy ranges were examined in our study. First, low-energy (350 keV) Fe⁺ irradiation was carried out at the Ion-Beam Facility at SUNY Albany. In this case, samples were tilted 7° from the beam direction to prevent ion channeling and were cooled during irradiation with a water-cooled sample plate. Second, high-energy (5MeV) Fe²⁺ irradiation was performed at the Environmental Molecular Sciences Laboratory (EMSL) at Pacific Northwest National Laboratory (PNNL). In this case, samples were mounted on a metal target holder; the damage was also created when using 7° tilting between the ion beam and the sample surface normal. All irradiation experiments were raster-scanned and performed in room temperature. A range of fluences from 1 × 10¹² to 5 × 10¹⁵ cm⁻² (~7.5 dpa, displacement per atom) were used for irradiation in order to be able to study damage evolution from the case of minor disorder to a continuous fully amorphous disorder. For the sequential irradiation of two ion species, He-ion irradiation, followed by Fe-ion irradiation, was used. Their corresponding ion energy and fluence are specifically described in each section below. Note that the different Fe-ion charge state in our low- (Fe⁺) and high-energy (Fe²⁺) irradiation was due the necessity of using an existing ion-source condition on each ion implanter.

RBS/C measurements were carried out at Albany and at PNNL with the He⁺ probing beam at initial energies ranging from 2 to 4 MeV, depending on the depth of damage in LiNbO₃ to be analyzed. In these measurements, a detector was positioned at 170° to collect backscattered helium. In channeling measurements, samples were carefully angle oriented to minimize backscatter for channeling measurements. The backscattered count (yield) was then recorded as a function of channel (energy) after a total charge of ~10 μC was delivered to the sample. The high-energy iron-ion irradiation and the subsequent ion-beam measurements were carried out in the same target chamber for in situ analysis. Selective samples were annealed up to 600°C under ambient conditions to study the recovery mechanism after exposure to the ion flux.

For micro-Raman spectroscopy, a 532 nm-wavelength excitation laser was used in a backscattered confocal-mode geometry. The spectrometer grating had 1800 lines/cm, giving a ~1.5 cm⁻¹ spectral resolution. The scanning was carried out in steps of 100 nm along the facet (edge) of the sample to acquire the damage depth profile. To carry out patterning, a surface mask was created on a LiNbO₃ sample surface prior to irradiation. The mask was either a contact mask (0.5 mm-thick metal sheet with ~500 μm circular-grid openings in diameter) or a photoresist pattern (Shipley S1818) with a thickness of ~2 μm.

For three-dimensional AFM imaging, a Veeco/Digital Instruments Multimode SPM (Scanning Probe Microscope) with a NanoScope III controller was used. The scan was operated in contact mode under ambient conditions, with a scan velocity of 60–80 μm/sec depending on the dimension of the scanned area.

TEM specimens were prepared by the focused-ion-beam (FIB) in situ lift-out (INLO) technique [20]. With this method, thinned electron-transparent (200-kV) wedge-shaped lamellae were readily obtained. Electron microscopy was performed with a JEOL JEM2100F, high-resolution analytical transmission electron microscope at 200 kV, and a Hitachi HD2700C scanning transmission electron microscope (STEM).

Table 1 summarizes the irradiation parameters and the probing methods performed in our experiments. Results are discussed in each section below.

Table 1. Summary of the irradiation parameters and probing methods in our experiments. Results are discussed in each section of the paper.

View Table

2.2. Surface characterization by optical microscopy and AFM

It is known that high concentrations of irradiated species can alter the surface topography of targets [21]. Such changes, which are due to the presence of the resulting interstitials and associated defect clusters, can be readily observed using optical microscopy (OM) or atomic force microscopy (AFM). Consider first deeply implanted species such as from high energy (MeV) light ions. In this case, the ions (e.g., helium) have relatively low electronic stopping power such that irradiation-induced damage observed in targets of such crystals is mostly from nuclear collision, occurring deep (micrometers) beneath the crystal surface. This buried damage region has been explored at the atomic scale using high-resolution electron microscopy. As one example, Ofan, et al. [21] showed that creation of copious nuclear displacements regions and He nanobubbles at the ion-implantation region result in the pileups of dislocation defects in local twin bands, forming along the threefold crystallographic axis. In addition, under these conditions, it was found that if helium-ions stop at depths of ≥ 2.5 μm, no change in surface topography was observed for the as-implanted material [22] other than a ~1–5 nm overall surface swelling [23]. However, for shallow implantation depths, i.e., for incident ion energies ≤ 350 keV, crystallographically driven fissure and cracks are observed at the edges of dislocation defects [22, 23]. Our previous micro-Raman study [7] also showed that helium-ion irradiation induces a relatively long-range strain field, i.e., a length scale of a few micrometers. If pileups and interstitials are close enough to the surface, high built-in stress will form and give rise to local surface exfoliation. Examples of such topographical changes occurring in our experiments are illustrated in Fig. 1 . Figures 1(a) and 1(b) show the results after room-temperature 195 keV He⁺ irradiation, corresponding to a stopping range (R_p) of ~670 nm. Figure 1(a) is an optical image of the surface showing the induced dislocation defects; Fig. 1(b) shows an AFM investigation revealing the raised topography of ~20 ± 5 nm along the edge of such dislocation, as indicated with a white arrow.

 

Fig. 1 Ion induced changes in surface conditions observed with optical and atomic force microscopy. Panels (a) and (c) are optical images and panels (b) and (d) are AFM scans. Panels (a) and (b) show the effects of 195 keV He⁺ irradiation (R_p ~670 nm) with a fluence of 4 × 10¹⁶ cm⁻²; panels (c) and (d) show effects of 5 MeV Fe²⁺ irradiation (R_p ~2 μm) with a fluence of 1 × 10¹⁵ cm⁻². The x and y axes in panels (b) and (d) both have scales of 5 μm/div, while the z axis has a scale of 35 nm/div in (b), and 250 nm/div in (d), respectively. Both panels (b and d) are displayed with a viewing angle of 60 degrees from sample surface. These images show clearly that shallow helium irradiation can give rise to surface cracking along lattice crystallographic axes, as indicated by the white arrow in (b), while iron irradiation results in isotropically oriented distributed surface deformations such as nanometer-scale hillocks. The height of these heavy-ion-induced protrusions is ~200 ± 40 nm, e.g., one of which is indicated by the white arrow in (d). The degree of the surface deformation scales with the near surface amorphization resulted from high electronic stopping. Note that the AFM scanned regions are indicated in the corresponding optical images in (a) and (c) with red arrows.

Download Full Size | PDF

In contrast, heavy ion irradiation, even at relatively low energies, is found to result in isotropically oriented surface damage or changes, i.e., generally not oriented with respect to the crystal directions. These changes are manifested as hillocks, swelling, and fracture. Earlier, it had been reported [9] that swift heavy-ion irradiation results in amorphous ion tracks; the appearance of protrusions or hillocks in these targets correlates with the appearance of this disorder formation after irradiation. Similar phenomena have also been observed in other oxides such as α-SiO₂ and Y₃Fe₅O₁₂ (YIG) [24]. In addition, calculations of the elastic stress fields [25] around these nano-tracks showed that the raised surface deformations are a consequence of the electronic-excitation-induced strain generated at the free surface of the sample.

In our experiments, both room-temperature low-energy (350 keV) Fe⁺ and high-energy (5 MeV) Fe²⁺ irradiation were performed, corresponding to a stopping range of ~0.2 μm and ~2 μm, respectively. For example, after irradiation with 350 keV Fe⁺ to a fluence of ~1 × 10¹⁴ cm⁻² or with 5 MeV Fe²⁺ to a fluence of ~1 × 10¹³ cm⁻², lattice disorder is apparent (see also RBS/C session below). In particular, Figs. 1(c) and 1(d) show the examples of optical and AFM images of the samples after 5 MeV Fe²⁺ irradiation with a fluence of 1 × 10¹⁵ cm⁻². From Figs. 1(c) and 1(d), it is clear that this energetic iron-ion irradiation gives rise to randomly oriented surface deformation. The AFM scan shows that the change in topography in this case exhibits a maximum roughness of ~200 ± 40 nm. As is seen in our AFM and RBS/C measurements shown in the next section, such surface elevation is correlated with the width of damaged material in the surface layer. In addition, later in the paper, it is shown via calculations and RBS/C measurements that the degree of surface deformation and local height elevation after a 5 MeV Fe²⁺ irradiation appears to scale with the depth of the near-surface amorphization as a result of the strong stopping energy (electronic) of these heavy ions.

2.3. In situ channeling Rutherford backscattering (RBS/C)

Our experiments have also used beam-scattering methods, i.e., RBS and channeling to observe nuclear displacement and interfacial intermixing in the bulk and surfaces of irradiated samples. Earlier our group and others [26] used RBS/C to study damage in keV and MeV He-ion-irradiated LiNbO₃ to show the importance and effect of fluence and annealing on this damage. For example, after 1 × 10¹⁶ cm⁻² He⁺ bombardment, an implant-induced peak in the RBS/C measurements was detected near the He⁺ stopping range. This peak became more prominent and sharper with low-temperature (175°C–275°C) annealing, reaching a maximum at ~250°C. This peak was thus attributed to thermal diffusion of He to form buried He bubbles. Further, higher-temperature annealing (T > 350°C) was found to induce He out-diffusion, resulting in a reduction of backscattered yield [26].

RBS/C measurements were further performed to study the induced damage from Fe-ion-irradiation. In particular, different energies of the incident He-ion probe were used to investigate damage at different depths using damage in the Nb sublattice. Previously reported data led us to anticipate more significant damage than with the use of He due to the larger cross section of Fe [27]. Figure 2(a) shows He-ion RBS scattering after low-energy (350 keV) Fe⁺ irradiation with a set of values of fluence from 1 × 10¹² cm⁻² to 1 × 10¹⁵ cm⁻². It is clear that as the exposure to irradiated species increases, dechanneling increases, resulting in an increased backscattered signal. This result is in agreement with an increase in the degree of damage from induced defects and crystalline damage. In particular, at this Fe⁺ energy, the critical fluence for the appearance of lattice disorder is ~5 × 10¹³ cm⁻². With a fluence of ~1 × 10¹⁵ cm⁻² (~1.5 dpa), the backscattered signal reaches the level of purely random scattering, indicating that, in this case, bombardment leads to a fully amorphized layer extending from surface to the depth of the ion stopping range [18].

 

Fig. 2 RBS/C measurements on 350 keV Fe⁺-irradiated LiNbO₃ using the Nb sublattice. Experimental and SRIM data showing that the dominant damage mechanism in low-energy Fe⁺ irradiation is due to nuclear collisions. (a) RBS/C versus Fe-ion fluence showing the evolution of crystal damage with the variation in fluence. The sharp increase in background with a fluence of ~1 × 10¹⁴ cm⁻² shows a threshold-like behavior for crystal amorphization. After a fluence of ~1 × 10¹⁵ cm⁻², the dechanneling signal reaches the level of that from a randomized lattice, indicating the material is fully amorphized. (b) The calculation using Eq. (1) of an ~1 × 10¹⁴ cm⁻² irradiation curve is shown by the blue line. The depth of the peak is ~110 nm below the surface in agreement with a SRIM simulation of most damaged location (~125 nm). (c) Cascading and overlapping of the defects. The data show that 5 × 10¹⁵ cm⁻² iron irradiation results in broader amorphous layers (additional ~35 nm increase in thickness) than the case with a fluence of 1 × 10¹⁵ cm⁻². (d) Data showing that annealing at different temperatures recovers the sample crystallinity.

Download Full Size | PDF

The data in Fig. 2(a) can be simply analyzed using the well-known approximate expression to estimate the relative distance between the location of the peak of the most-damaged region and the surface [28]. Assuming binary collisions:

Figures 2(c) and (d) provide additional data on yield v.s. scattered He⁺ energy (channel) to further characterize damage after bombardment with 350 keV Fe ions. Figure 2(c) shows that as the irradiation fluence is increased, further increases occur in the extent of the amorphized layer, e.g., a fluence of 5 × 10¹⁵ cm⁻² gives rise to a broader amorphized layer than that of 1 × 10¹⁵ cm⁻²; see Fig. 2(c). This broadening is attributed to the high concentration of induced defects from nuclear-collision cascades [29]. Using Eq. (1), we find that this higher-fluence irradiation (5 × 10¹⁵ cm⁻² Fe⁺) results in an additional ~35 nm increase in the thickness of the amorphized layer. Figure 2(d) displays the effects of annealing at different temperatures up to a maximum temperature of 600°C, with a 30 min annealing duration at each temperature. As the annealing temperature increases, dechanneling is reduced, indicating a partial recovery of the sample crystallinity with annealing. Note that this data suggests that the crystallinity would be fully restored with the use of even higher annealing temperature and longer duration [7]. Also note that from SRIM, the electronic stopping (S_e) of 350 keV Fe in LiNbO₃ is ~0.5 keV/nm. This number is approximately an order of magnitude smaller than the reported threshold value of S_e [18] that has been reported to cause electronic-collision-induced damage (see below). Thus, the dominant mechanism giving rise to materials amorphization in low-energy regime is attributed to nuclear collision.

The primary contribution of nuclear collision to damage is expected to change as the Fe-ion-energy increases since for heavy ions, the electronic stopping power, S_e, is also known to increase with ion energy. To examine this phenomenon, experiments with high-energy Fe ions were carried out using 5 MeV Fe-ion irradiation at the EMSL facility at PNNL. Our SRIM calculations showed that 5 MeV Fe ions have a stopping range of ~2 μm in LiNbO₃ and an initial electronic stopping S_e of ~3.6 keV/nm. The results of our experimental measurements with RBS/C analysis to determine the evolution of ion damage with fluence are given in Fig. 3 ; the channeling data in Fig. 3(a) used a comparatively low energy, 2 MeV He⁺ probe to study damage closer to the surface, while the experiment results shown in Fig. 3(b) used higher energy, 4 MeV He⁺, to investigate the overall damage profile. For the low-energy He probe in Fig. 3(a), after a fluence of >1 × 10¹³ cm⁻² Fe²⁺, a strong back-scattered dechanneling He signal is observed. This dechanneling is known from [25] to indicate damage and displaced atoms generated near the surface region. Note that the backscattered signal of 1 × 10¹³ cm⁻² from the Fe²⁺-irradiated sample is close to the level of a fully randomized sample, indicating that the critical fluence for the appearance of damage formation occurs between 1 × 10¹² cm⁻² and 1 × 10¹³ cm⁻² Fe²⁺. In addition, for a fluence of 1 × 10¹⁴ cm⁻², it is apparent that this damaged surface layer becomes fully amorphous, since the backscattered signal level is observed to reach that of a randomly-oriented material. Using the expression given in Eq. (1) and the data for the 1 × 10¹³ cm⁻² curve (from Channel 750~850, or Energy 1450~1650 keV), the thickness of this damaged material on the surface is found to be ~245 nm. Note that this thickness also agrees with the measured elevated height of the surface deformation (~200 ± 40 nm); see AFM data in Fig. 1 in the previous section. This agreement suggests that the observed hillocks feature originates from a region of amorphous material in the near surface region.

 

Fig. 3 In situ RBS/C measurements on 5 MeV Fe²⁺ irradiated LiNbO₃ using Nb sublattice. (a) Results of RBS using a 2 MeV He⁺ probe to study damage evolution on surface. The data show that the critical fluence for damage formation is ~1 × 10¹³ cm⁻² and that an irradiation fluence of ~1 × 10¹⁴ cm⁻² gives rise to amorphization. Note that this critical fluence is approximately an order of magnitude smaller than that achieved using low-energy iron, as seen in Fig. 2. (b) RBS/C data as a function of fluence. Two dechanneling mechanisms are clearly seen; that in the higher channel (>700) is attributed to electronic damage, and that in channels 500~600 is attributed to nuclear collision. The contribution of both effects lowers the critical fluence for the emergence of lattice disorder and material amorphization.

Download Full Size | PDF

These observations can be interpreted using prior observations obtained in both light- and heavy-ion damage. For example, as briefly mentioned above, the interaction of heavy ions with their target lead to extended defects [30]. In fact, beyond a threshold in stopping-power, high electronic excitation coupled to atomic motion results in track formation, through mechanisms based on Coulomb explosion and thermal spiking [17]. With a further increase in irradiation fluence, additional lattice disorder is formed and broadened along the projectile incident axis. Concomitantly, elastic strain is also induced. Once the “swelling threshold” is exceeded [24], the generation of the overlapping defects and the associated built-in stress give rise to structural transformations both in the bulk and the near-surface of the material. Such structural transformations and the built-up stress thus lead to volume expansion (swelling); these effects result in the surface protrusions seen in our samples via the AFM scans in Fig. 1(d). Note that this volume expansion is related to the phase transition from crystalline to amorphous state [24]; however, a more detailed set of imaging experiments (e.g., TEM) must be undertaken before a more definite statement can be made on the relation between the height of surface swelling and the extent of surface damage.

In Fig. 3(b), besides the damage at the surface seen in Fig. 3(a), another and less well-defined backscattered signal is also detected in deeper regions of the sample (Channel 500~600). The use of Eq. (1) along with the measured energies in the figure shows that the location of this peak is ~1.86 μm below surface, which corresponds to the location where the most probable nuclear collisions take place (~1.9 μm, with R_p ~2.0 μm) based on SRIM calculation. Thus the existence of this second higher-yield signal is attributed to disorder-induced dechanneling from nuclear collisions.

Previously, it was shown that if electronic ionization exceeds a threshold energy, the energetic ions will produce nanometer-wide latent tracks in the targets [31]. In particular, Olivares, et al. reported [18] that for 5-MeV-Si-irradiated LiNbO₃, such an electronic stopping threshold (S_e,th) is ~5 keV/nm. If S_e ≥ S_e,th, a latent (amorphous) track is then generated. In addition, simulations based on a thermal-spike model and a semi-empirical formulation [32] showed that such a threshold S_e,th is, in fact, dependent on the reduced energy E_r (E_r = ion energy / ion mass) of the ion species. If the reduced energy E_r of the chosen ion is ~0.1 MeV/amu, the threshold S_e,th to induce electronic-stopping damage will be ~3 keV/nm. With this information in mind, consider our high-energy 5 MeV Fe²⁺ irradiation experiments. From SRIM calculations, it is seen that 5 MeV iron atoms (E_r ~0.1 MeV/amu) would yield an initial electronic stopping value of S_e ~3.6 keV/nm, which is higher than the predicted threshold S_e,th (~3 keV/nm) for electronic damage, thus indicating that it is reasonable to expect electronic origin for damage. This damage would be present at the surface and extend into the bulk until sufficient ion energy is lost such that the electronic stopping value drops below threshold. Our low-energy He probe in the RBS/C measurements confirms and shows the existence of such a damaged surface layer. Further, measurements with high-energy (5 MeV) iron ions show that the critical fluence for the appearance of damage formation (~1 × 10¹³ cm⁻²) and of an amorphized layer (~1 × 10¹⁴ cm⁻²) are approximately an order of magnitude lower than for these values using low-energy (350 keV) Fe⁺ [~1 × 10¹⁴ cm⁻² and ~1 × 10¹⁵ cm⁻², respectively; see Fig. 2(a)]. Such enhanced damage from swift heavy ions with lower critical fluence is attributed to the fact that in addition to damage from nuclear collisions taking place in this ion range, high electronic stopping damage is also occurring. With the existence of these two contributions, i.e., nuclear-collision events and high electronic stopping, defect accumulation and overlapping are thus enhanced, lowering the critical fluence for damage formation. For low-energy irradiation, on the other hand, only nuclear stopping dominates, thus the critical values are higher. Similar phenomena have also been reported using other ions such as O, F and Cl [33, 34].

RBS/C measurements were also performed on He⁺ and Fe⁺ irradiated samples to examine any cumulative damage effects present during sequential irradiation by the two ions. This form of irradiation is of particular interest, considering that in many recent ion-irradiation processing experiments [35, 36], sequential irradiation has been shown to be potentially useful in providing flexibility in influencing or controlling ion-damage profiles. In particular, depending on the irradiation condition, additive damage production or induced healing from sequential irradiation or simultaneous dual-beam co-irradiation have been reported [37].

Thus, irradiation of only He⁺, only Fe⁺, and sequential irradiation with both species (He⁺ first followed by Fe⁺) were carried out. The corresponding ion energies and fluence are 200 keV He⁺ with a fluence of 1 × 10¹⁶ cm⁻² or ~0.25 dpa (R_p ~700 nm), and 350 keV Fe⁺ with a fluence of 1 × 10¹⁴ cm⁻² or ~0.15 dpa (R_p ~200 nm), respectively. The RBS/C measurements are displayed in Fig. 4 as green (He), blue (Fe), and orange (He and Fe) curves, respectively. Measurements with virgin (black) and randomized (red) samples are also included for comparison. From Fig. 4, it is clear that in our sequential irradiation, it leads to higher backscattered counts compared with the Fe⁺(only) irradiation. This higher yield in the sequentially irradiated sample is tentatively attributed to the fact that the presence of the localized He inclusions underneath (shown at Channel ~600) can result in significant long-range strain field and dislocation network extending toward the surface [6, 7, 26, 38]. Such pre-existing stress and defect networks after He-ion irradiation can thus enhance the formation of Fe⁺-induced damage. Note that this cumulative damage is also observed using micro-Raman scanning and TEM imaging as described below. These results indicate that damage from cumulative irradiation may alter materials properties more significantly than expected. Thus, the irradiation conditions of each ion species need to be optimized to achieve the expected effects from individual irradiation.

 

Fig. 4 RBS/C measurements on virgin (black), He⁺-irradiated (green), Fe⁺-irradiated (blue) and sequentially irradiated (orange) samples. A result from a fully randomized sample is also included for comparison. The corresponding energy and fluence of the irradiation conditions are described in the text. It is clear that the sequential irradiation results in cumulative damage. In particular, Fe⁺-induced damage is enhanced in the presence of long-range strain and dislocation network from the He inclusions at the stopping range.

Download Full Size | PDF

In summary, our RBS/C experiments have shown that for iron-ion irradiation, the critical fluence for the appearance of damage is energy-dependent, with values two to three orders of magnitude smaller than for the case of helium irradiation. In addition, iron-ion-induced materials amorphization has been observed, while for helium, the materials remain crystalline for the highest fluence used in the experiments (~5 × 10¹⁶ cm⁻² He⁺). Also, sequential irradiation results in cumulative damage from each ion species. Note that SHI effects are to be expected using a wide range of other heavy atoms, such as Xe, Ar, etc.

2.4. Confocal micro-Raman imaging and scanning

Confocal micro-Raman spectroscopy is a powerful nondestructive approach to examine lattice modification and local micro-structural change. Its imaging capability provides direct visualization of crystallographic defects and damage formation. Our previously reported scanned micro-Raman imaging study of He⁺ irradiated LiNbO₃ [6, 7] showed that light-ion irradiation gives rise to lattice disorder and induces elastic strain, with the most damaged region of the crystal at approximately the ion stopping range. Such damage is manifested in Raman spectra through the broadening of peak widths, shifts in spectral features, decrease in active-mode intensities, and appearance of forbidden-mode signals. The decrease and increase in active- and forbidden-mode signals can lead to high imaging contrast, which has been proven to be effective for locating and determining the spatial-damage distribution of particle irradiation [6].

In our experiments here, Raman imaging was done with samples irradiated by both light and heavy ions, and the images are displayed in Fig. 5 . Figure 5(a) shows Raman mapping of 195-keV He⁺-irradiated LiNbO₃ performed through the top, irradiated surface (Z + face). In the figure, the measured mode profiles are overlaid on the corresponding optical image, with the scanned region outlined in red. The figure shows an example of damage-induced changes in both active and forbidden modes. The intensities in the image of the active- and forbidden-mode signals exhibit an opposite behavior in the presence of lattice damage with the active mode decreasing and the forbidden mode increasing, respectively. For example, in Fig. 5(a), the E(TO₈) active mode at ~581 cm⁻¹ and A₁(TO₄) forbidden mode at ~631 cm⁻¹ were used in the scan geometry, which is $Z(X,XY)\overline{Z}$in Porto notation [39]. In the figure, active-mode imaging (left inset in green) shows the distribution of dislocation lines. Forbidden-mode imaging (right inset in red) is then used to show where local surface cracks occur. Our results are in agreement (see section on surface characterization) with the fact that shallow He⁺ implantation can, under certain conditions, give rise to local surface cracking along crystallographic axes. This blistering-induced exfoliation and cracking is accompanied by significant lattice disorder such that the normally-forbidden phonon-vibrational activity is allowed.

 

Fig. 5 Confocal micro-Raman imaging to visualize the modifications of phonon vibrations. In (a), 195 keV He⁺ irradiation was used. This shallow irradiation gives rise to lattice disorder, with most damage buried along lattice crystallographic axes. Surface cracking can also be observed. The imaging clearly shows at these locations, the forbidden mode (631 cm⁻¹) is “turned on”, while the active mode signal is decreased (581 cm⁻¹). In (b), 5 MeV Fe²⁺ irradiation gives rise to surface deformations. The locations of these deformations can also be illustrated using Raman forbidden-mode mapping.

Download Full Size | PDF

A similar analysis can also be applied for heavy-ion irradiation. In Fig. 5(b) and (c), Raman mapping was carried out with 5 MeV-Fe²⁺-irradiated LiNbO₃. Figure 5(b) is an optical image of a sample with the irradiated regions patterned with a surface mask. In this sample, the center circular region was irradiated with iron ions, whereas the outside is masked and thus unirradiated. As discussed in Section B and shown in Fig. 1, high-energy iron-ion irradiation gives rise to surface deformation such as the raising of local regions and hillock formation. Using the same normally-forbidden phonon mode at ~631 cm⁻¹, it is seen that the bright red-spot areas correspond to regions where the surface is severely deformed. Note that when the scan is in a region away from the irradiation region, the forbidden mode signal is minimal (≤ 10% of the maximum value). This result indicates that the patterning is, in general, well confined to the irradiated region of interest. From our measurements, it is then clear that the use of Raman imaging enables the detection of these nanometer-scale surface non-uniformities after either light- or heavy-ion irradiation.

A micro-Raman edge scan, with a scan geometry of$X(Z,ZY)\overline{X}$ in Porto notation, was also carried out on Fe²⁺-irradiated LiNbO₃ to acquire the damage profile as a function of irradiation depth. In addition, probing was also performed on He⁺ and Fe²⁺ sequentially irradiated samples to examine any cumulative damage effects by the two ion species. Selection of a confocal geometry for the Raman microscopy enabled a high-resolution scan step (~100 nm) and a tight spot size (≤ 1 μm) to be achieved in our experiments for the damage measurements. Thus, a sample of Z-cut LiNbO₃ was first irradiated with 3.8 MeV He⁺ to a fluence of 5 × 10¹⁶ cm⁻², followed by a second irradiation using 5 MeV Fe²⁺ to a fluence of 1 × 10¹⁵ cm⁻². This sequential irradiation process was performed at room temperature with a 7° degree tilt from the sample normal to prevent channeling. Subsequently, the sample facet was carefully polished until an optical finish was achieved, i.e., ~λ/5 of the Raman excitation wavelength, to investigate the change of phonon vibrational activities on the irradiated samples.

Examples of these micro-Raman edge-scan measurements are displayed in Fig. 6 . Figures 6(a) and 6(c) are optical images of side views of Fe²⁺-only-irradiated (1 × 10¹⁵ cm⁻² fluence), and He⁺ (5 × 10¹⁶ cm⁻² fluence) and Fe²⁺sequential-irradiated LiNbO₃, respectively; such cross-sectional optical images enabled the direct visualization of any change in the optical properties upon irradiation and are presented in this section for the purpose of easy comparison to the results of Raman light scattering. Figures 6(b) and 6(d) are the results of the corresponding edge scan for Raman features between 500 and 1000 cm⁻¹. The insets in Figs. 6(b) and 6(d) display a plot of the measured peak intensities of the A₁(TO₄) active mode at ~631 cm⁻¹ versus the scan distance. Note that a different set of selection rules applies for this edge-scan optical configuration, as compared with those used in Fig. 5.

 

Fig. 6 Raman edge scan of Fe²⁺ irradiated and He⁺ and Fe²⁺ sequentially irradiated LiNbO₃. Panels (a) and (c) are side views of optical images; panels (b) and (d) are the corresponding Raman edge-scan results. The insets show the intensity of the A₁(TO₄) active mode as a function of irradiation depth. In (a), a darker layer is seen, indicated by yellow arrows. Raman scanning in this layer shows a region of low-crystallinity, displayed as the red spectra in (b). In (c), besides the darker layer close to surface, additional dim line-like region at ~10 μm below is also observed, indicated by red arrow. The appearance of this line is attributed to helium-ion-induced nuclear damage. In (d), the evolution of the spectra at different locations is apparent. The inset in (d) clearly shows a transition region (~2 to ~4 μm in distance, region II). The width of this transition region is determined by the cumulative effects of the sequential irradiation.

Download Full Size | PDF

Consider first the case of an Fe-ion-only irradiated sample. In Fig. 6(a), an apparently damaged layer within the surface region (indicated by yellow arrows) is shown, with a thickness corresponding to the range of 5 MeV iron ions in LiNbO₃ (R_p ~2 μm). The formation of this damaged darker layer is a result of the overlap of multiple iron-ion tracks, as alluded to above. This defective region of defects modifies the crystal structure over a depth extending from the surface to the end of the ion range. In addition, such Fe-ion-induced change of lattice crystallinity also modifies the phonon vibrational activities as manifested in the Raman spectra shown in Fig. 6(b). In Fig. 6(b) (not the inset), spectra corresponding to different locations or regions within the depth scan are shown. Different line colors are used for scans in different locations: red in the surface region where iron ions interact with LiNbO₃ (≤ 2 μm from surface); blue in the deeper unirradiated crystal region. Note that the locations of these measured spectra are also shown by spots of the corresponding color on the optical images in Fig. 6(a). It is clear from Fig. 6(b) that the regions that are Fe-ion-irradiated (from surface to ~2 μm below) show a decrease of active-mode signals, suggesting strong lattice crystalline disorder. Furthermore, as seen in the inset in Fig. 6(b), the intensity of this low-density crystalline active mode A₁(TO₄) at ~631 cm⁻¹ remains nearly constant from the surface to the Fe-ion stopping distance. This distance is in agreement with the thickness of the optically-damaged layer shown in the optical side view of the darker layer in Fig. 6(a). Once the probe depth reaches that of the unirradiated region (~2 μm below the surface), the spectral signal becomes that of the original unirradiated crystal. Note that, as mentioned above, after a 1 × 10¹⁵ cm⁻² iron bombardment, RBS/C measurements show a nearly amorphous layer is created, since the backscattered yield is at the level of a randomized lattice. However, as is apparent from the Raman scan, remnants of crystalline-like peak signal still appear. For example, the intensity of the A₁(TO₄) active mode discussed here exhibits an amplitude of ~10% at the damaged location of that in the unirradiated virgin region. This result indicates the existence of a mixed phase containing both amorphous-like and low-density crystalline material. In addition, a ~5 cm⁻¹ peak redshift is also observed, indicating the presence of the induced strain field. Similar effects have also been reported for fluorine-irradiated LiNbO₃ [29]. Also note that in addition to the active phonon-mode profile, the appearance of an additional Raman spectral feature from 800 to 900 cm⁻¹ is also observed. The emergence of this feature is attributed to irradiation-induced activation of known forbidden modes [6, 40].

Consider now the sequentially irradiated (He⁺ and Fe²⁺) case. Figure 6(c) displays the optical cross-sectional image of this irradiated sample. From Fig. 6(c), a similar damaged surface layer from iron ion is seen as in Fig. 6(a) (indicated by the yellow arrows), but in this case a weak defect “line” has also appeared at location of ~10 μm below surface; this line is indicated by the red arrow. The appearance of this line is attributed to the nuclear damage from the 3.8-MeV-He⁺ irradiation. Note that the thickness of the He⁺-induced displacement-damage region is much narrower, that is, the region dominated by nuclear interactions. Judging from the optical images, it is apparent that the combination of electronic and nuclear interactions between heavy ion and the target results in damage over a much broader spatial region and thus has a greater impact on optical properties of the crystal than is the case for the impact of light ions. Also note that the thickness of the damaged surface layer (yellow-arrows region) is identical in Figs. 6(a) and (c). As will be seen below using Raman scanning, this same thickness indicates that damage from iron ions is dominant in the sequentially irradiated case.

Figure 6(d) shows the Raman spectra of the sequentially irradiated sample taken at different locations: red and blue taken at the surface (≤ 2 μm) and deeper unirradiated (> 10 μm) regions, respectively. The black spectrum is a scan taken ~5 μm below the surface after He⁺ exposure, while the orange spectrum is taken at a ~10 μm depth, corresponding to the He⁺ stopping range. The locations of these scans are also shown in Fig. 6(c) by spots of the corresponding color. It is clear from Fig. 6(d) that four regions are observed: (I) an Fe-ion-induced-damaged surface layer from the surface to a ~2 μm depth, (II) the transition region from ~2 μm to ~4 μm, (III) long-range elastic strain and local He-induced lattice damage from ~4 μm to ~10 μm, and (IV) a deep unirradiated region at a ~10 μm depth. Using the intensity of the same A₁(TO₄) active mode, as is observed in Fig. 6(b), the damage profile along the z axis of the sample was measured and is displayed in the inset. It is clear that the intensity profile in region (I) is essentially identical to the profile in Fig. 6(b), that is, the sequentially irradiated damage profile in (I) shows a similar spectra profile as in the surface region of the Fe-ion-only irradiated sample. This indicates in the sequentially irradiated sample, Fe-ion irradiation is the dominant contributor to the damage in the surface region. However, also notice that the Raman-peak position is red-shifted (by additional ~4.5 cm⁻¹) as compared with the Fe-ion-only irradiated case. This additional peak redshift is attributed to the long-range elastic strain induced by the buried He inclusions. Compared with the solely Fe-ion-irradiated sample, the presence of this He-induced strain field also results in ~10% lowered intensity in region (I) (see the inset) of the sequentially irradiated sample. This cumulative effect agrees with the higher backscattered counts observed in the RBS/C measurements as discussed above. More discussions of this strain/stress states are described in the next paragraph. Finally, once the probe is in a region beyond the sequentially irradiated region, i.e., region (IV), the spectral becomes that of an unirradiated sample.

Region (III) shows an apparent extension of damaged region. This feature is attributed to a long-range strain/stress field, which gives rise to a prominent Raman peak shift over its value measured in the unirradiated crystalline region. The existence of such a long-range strain field in LiNbO₃ after helium irradiation [7, 23] was reported in earlier X-ray diffraction and Raman studies. In fact, in both of these reports, it was found that the length scale of the stress field can extend several micrometers from the buried helium-ion stopping region. Note that damage would be expected to be maximum at the helium-ion range where the magnitude of strain is the greatest [7, 23]; in fact, a local intensity minimum is seen at the stopping region of the helium ions, i.e., ~10 μm below surface.

Finally, region (II) is the transition between the two irradiation zones. The width and the materials properties of this transition region [i.e., region (II) in the inset in Fig. 6(d)] are strongly dependent on the distance between the ion ranges of the two species. If lower iron-ion energies and/or higher helium-ion energies are used, the cumulative effect from the sequential irradiation is minimized, thus creating an unaffected unirradiated-like transition layer.

2.5. Transmission Electron Microscopy (TEM) imaging

To further examine any radiation-induced modifications in the material crystallinity and related properties, cross-sectional transmission electron microscopy (TEM) was performed on the lattice structure of the irradiated sample. Typical imaging of the samples is displayed in Figs. 7 and 8 . These samples included Fe⁺-only irradiated and He⁺ and Fe⁺ sequentially irradiated LiNbO₃. The irradiation parameters were 350 keV Fe⁺ with a fluence of 1 × 10¹⁴ cm⁻² for Fe⁺-only irradiated LiNbO₃; for sequentially irradiated samples, 200 keV He⁺ with a fluence of 1 × 10¹⁶ cm⁻², followed by 1 × 10¹⁴ cm⁻²-Fe⁺ at 350 keV (i.e., the same Fe⁺-only-irradiation condition) were used.

 

Fig. 7 HRTEM micrographs of Fe⁺-irradiated LiNbO₃. The iron ion fluence is 1 × 10¹⁴ cm⁻² at an ion energy of 350 keV. Figure 7(a) shows a cross-sectional image at a lower magnification ( × 80k); damage distribution is seen from the surface to a depth of ~210 nm. Figures 7(b)-7(d) are images at a higher magnification ( × 500k), showing more detailed defective regions at different depths below the surface: Fig. 7(b) at the surface, Fig. 7(c) at a depth of ~100 nm, and Fig. 7(d) at a depth of ~200 nm. It is clear from the zoomed-in micrographs that the size and the concentration of these nanometer-scale defects are depth-dependent, with the average size of ~4–6 nm and the highest concentration at a depth of ~100 nm. Notice that this depth is in accord with results from our previous RBS/C measurements; see Fig. 2(b). Figure 7(e) is an EDS line scan profile measuring O and Nb concentrations across two defective regions (top and bottom insets are the corresponding Z-contrast (dark field) STEM image and the signal after background subtraction, respectively). It is clear that non-uniform distribution of elemental composition is observed in such regions.

Download Full Size | PDF

 

Fig. 8 HRTEM micrographs of Fe⁺-irradiated [Fig. 8(a)] and He^+- and Fe^+-sequential irradiated [Fig. 8(b)] LiNbO₃. The energies of the He⁺ and Fe⁺ ions are 200 keV (R_p ~700 nm) and 350 keV (R_p ~200 nm), with fluence of 1 × 10¹⁶ cm⁻² and 1 × 10¹⁴ cm⁻², respectively. A comparison of these two types of irradiation makes it clear that sequential irradiation gives rise to a more apparent image contrast in the defective regions. Further, several of these features exhibit faceted shapes (see the yellow labels for example). These more apparent features are attributed to the formation of higher stress field and crystallographically oriented dislocation network, resulted from He⁺ irradiation.

Download Full Size | PDF

Figure 7 shows a high-resolution TEM (HRTEM) images of Fe-irradiated LiNbO₃. Figure 7(a) is a micrograph with × 80k magnification depicting the damage distribution from the surface to a ~210 nm depth in the sample, and Figs. 7(b)-7(d) are zoomed-in cross-sectional images at a × 500k magnification at surface [Fig. 7(b)], at ~100 nm below the surface [Fig. 7(c)], and at ~200 nm below the surface [Fig. 7(d)], respectively. It is clear from these micrographs that Fe⁺-induced nanometric-size damage pockets are formed. The size and the concentration of these defective regions are depth-dependent, with the size ranging from ~2 nm to ~15 nm and their maximum concentration located at a distance ~100 nm below the surface. Note that this latter depth agrees well with the location of the highest ion-dechanneling signal observed by RBS/C, as mentioned above in Fig. 2(b). These defects, originating from large cascading collision events, generally lead to strong displacement damage including the generation of Frenkel pairs and extended defects [21]. Such dense local lattice distortions would thus lead to the strong back-scattered RBS/C signal that has been observed in our RBS studies. This defective region persisted to a depth of ~350 nm but was absent for depths > 500 nm. Notice that, despite the high density of defects, overall sample crystallinity was still present in all samples. This “sub-amorphization” result is also in good agreement with our RBS/C measurements, since the fluence and energy used here, i.e., 1 × 10¹⁴ cm⁻²-Fe⁺ with an energy 350 keV, is below the threshold value for full amorphization as discussed above. Finally, note that our TEM images also showed the presence of a strain field. The existence of this elastic strain also contributes to the dechanneling in our RBS/C measurements, as well as a peak shift in Raman phonon modes, as seen elsewhere in our results (Fig. 6).

An important question, which arises when examining the image in Fig. 7(a), is the origin of the spherically shaped defects. In general, image contrast results from non-coherent diffraction by the lattice planes. In the case of nonreactive rare-gas bombardment in oxides, ~5 nm rare-gas bubbles have been reported near the end of the ion range [22, 41–44]. In addition, our irradiation might also be expected to result in an alteration of local crystal composition. For example, changes in elemental concentration, such as the loss of Li, has been reported in helium-irradiated LiNbO₃ [26] using nuclear reaction analysis (NRA). In the present experiments, Fe-ion bombardment would be expected to lead to changes in crystal decomposition through the combined effects of both electronic and nuclear interactions along the ion path. To investigate the nature of the spherical features, Z-contrast imaging (HAADF) and energy-dispersive X-ray spectroscopy (EDS) were performed. The results, displayed in Fig. 7(e) and the insets, show apparent non-uniform distributions of Nb and O in these defective regions. Figure 7(e) displays raw data and its top and bottom insets show the HAADF image and the signal after background subtraction, respectively. The background signal was obtained by performing a separate EDS line scan across regions with no such defects at the same depth. In particular, from our EDS line scan profile and the STEM HAADF data, the dark regions in the inset HAADF image in Fig. 7(e) show a decrease in oxygen and niobium; note that in the bright-field STEM images, these are light images. The presence of such image contrast, which occurs in both bright- and dark-field imaging, suggests the existence of a compositional difference compared with the undamaged region. The fact that compositional decrease in the damaged regions occurs indicates perhaps the formation of a lower-density phase or micro-cavity; such a spherical region might result from the diffusive loss of oxygen to the surrounding area. Currently more extensive experiments, including the use of different iron-ion fluence and energies, are being carried out to provide more quantitative change in concentration and lattice structure.

TEM imaging on sequentially irradiated (He⁺ and Fe⁺) sample was also carried out and compared with Fe⁺-only irradiated samples. Images of these samples, taken at the similar depth (~30 nm below the surface), are displayed in Fig. 8. Figure 8(a) shows an Fe⁺(only)-irradiated sample, while Fig. 8(b) displays a sequentially irradiated sample. Both micrographs were taken at a magnification of × 500k. A comparison of the two figures make it clear that sequential irradiation gives rise to sharper contrast in the damage pockets. This greater damage is also reflected in RBS/C scattering and micro-Raman scan as discussed above. In addition, several of the defective regions in the sequentially irradiated sample exhibit faceted shapes, see the yellow labels in Fig. 8(b) for example. The presence of these apparent faceted shapes are attributed to the formation of a greater stress field in the He⁺ bombarded sample as well as a He⁺-induced crystallographically oriented dislocation network [4, 38].

3. Conclusions

A comprehensive and comparative investigation of radiation damage from light- and heavy-ion bombardment has been undertaken. First, surface conditions were characterized using optical microscopy and AFM. Irradiation-induced surface deformation was observed. It is found that shallow helium-ion irradiation can give rise to local cracking along the threefold crystallographic axis, while iron-ion irradiation results in isotropically-oriented hillocks. Second, compared with light-ion or low-energy heavy-ion (Fe⁺) irradiation, in situ RBS/C measurements show that high-energy (swift) heavy-ion irradiation induces electronic damage; besides nuclear collision, additional contribution to damage from electronic stopping lowers the critical fluence for the appearance of damage formation and materials amorphization. Third, confocal micro-Raman mapping showed clearly the difference in the formation of damage and distribution from light- and heavy-ion bombardment. Raman measurements also showed mixing of amorphous-like and low-density crystalline phases. Fourth, TEM imaging measurements also uncoverd the formation of depth-dependent damage pockets from Fe-ion irradiation, with an average size of ~4–6 nm. In addition, EDS line scans allowed the observation of a change of elemental composition in these defective regions. Finally, sequential irradiation of both light and heavy ion performed in our experiments resulted in enhanced accumulation of damage. Such enhanced cumulative damage effect was found in RBS/C scattering, Raman scanning, and TEM imaging, and was attributed to the formation of long-range strain/stress field from He⁺ irradiation.

From a more general perspective, while this study has focused on two specific light and heavy ions, it is clearly applicable to other species. For example, similar effects have been found by our group and others when using the Ar⁺ as the “heavy” ion. In addition, applications of this work are currently being pursued in heavy-ion etching of oxides as well as in exfoliation of thin oxide films for photonic-crystal and nanowire structures [45].