1 Introduction

Si has been the dominant semiconductor material for decades with a number of applications including sensors, detectors, nanoelectronic and photovoltaic devices [1,2,3,4,5]. Material processing generally involves thermal treatments at elevated temperatures as well as irradiations/implantations by various particles. These processes introduce defects. Notably, Si-based devices are widely used in special environments where radiation damage is caused during operation, as for instance in accelerators, in large hadron collider, in nuclear medicine, and aerospace industry. In such applications the reliable performance of devices is highly demanded.

Regarding irradiations, it is well-known that the energetic bombardment of Si leads to the athermal production of primary defects that is vacancies (V) and self-interstitials (SiI). Importantly, the scientific understanding of radiation defects and their influence on the properties and the behaviour of any kind of material, for instance metals, semiconductors, ionic materials starts from the knowledge of the primary damage caused by irradiation [6, 7]. Although most of V and (SiI) defects annihilate between themselves in the course of irradiation, a fraction of them survive and recombine leading to the formation of either di-vacancies and multi-vacancy clusters or the formation of di-interstitials and multi-interstitial clusters [8,9,10,11,12]. Additionally, unbound V and SiI diffuse away to be subsequently trapped by impurities, such as oxygen (O) and carbon (C) atoms that are present in Si, forming stable defects. Important families of the latter defects are oxygen-related and carbon-related clusters as for instance VnOm, CiOi, CiCs, CiOi(SiI), CiCs(SiI) and others. The latter complexes have been investigated in detail by our group in previous works [13,14,15,16,17] and they are not of interest in the present study. Notably, neutron irradiation is an effective method [18] to introduce large concentration of primary defects. The established [19] separation of vacancies and self-interstitials in this kind of irradiation leads to larger concentrations of them that survive annihilation and therefore to larger concentration of the subsequently formed next generation complexes, which facilitate their study. Significantly enough, neutron irradiation introduces similar defects with ion implantation which is used widely in Si wafer processing. Since the signals from ion implantation induced defects are weak, neutron irradiation is used in order to study such defects [10, 20]. Obviously, the knowledge and control of these defects is very important to avoid detrimental effects induced by ion implantation and in general by any kind of irradiation used in various processing stages on the properties of Si and therefore to improve the quality of the material.

Regarding thermal treatments at elevated temperatures, it is well-known that they introduce various defects in the material, depending on the used temperature. For instance, treatments at temperatures around 1000 °C are technologically important due to the fact that very serious operations are performed on the material at this temperature range, as diffusion, oxidation and transmutation doping processes. In Cz–Si treated in such temperatures structural defects as precipitates, dislocation loops and stacking faults are formed [21]. Additionally, enhanced hydrostatic pressure applied in the course of thermal treatment has many effects on the material and its properties. In particular, it affects the diffusion kinetics of various dopants, influences the formation and evolution of various formed defects, as for instance it enhances the formation of thermal donors and oxygen precipitates. [22,23,24,25,26,27,28].

In general, defect signals are usually larger in material subjected to neutron irradiations, which facilitates their study including the determination of the mechanisms involved in their production, their thermal stability, their evolution with temperature, their origin, the reactions they participate in, as well as the products of these reactions. The aim of this investigation is the study of the properties and behavior of defects in neutron irradiated Si subjected to HTHP treatments.

2 Experimental details

In this work we used Cz-grown p-type Si samples with boron, oxygen and carbon concentrations of ~ 2 × 1015, 7.2 × 1017 and 1.52 × 1017 cm−3, respectively. The samples were cut from prepolished Si wafers of 2 mm thickness purchased from MEMC. They were initially subjected [29] to HTHP treatments (1000 °C, 11 Kbar for 5 h) carried out in pure argon atmosphere. Afterwards, the samples were irradiated with 5 MeV fast neutrons at a fluence of 1 × 1017 n cm−2 at ~ 40 °C. After the irradiation the samples were put in quartz cells and subjected to 20 min isochronal anneals, of ~ 10 °C steps, in open furnaces. The evolution of the radiation-induced defects was monitored by infrared spectroscopy measurements carried out, after each stage of the isochronal anneals sequence, at room temperature. A Jasco-IR 700 dispersive kind spectrometer operated with spectral resolution of 1 cm−1 was used. The background two-phonon absorption was always subtracted by using a reference sample of FZ–Si material of equal thickness with that of the Cz samples.

3 Results and discussion

Figure 1a presents the IR spectra of Cz–Si immediately after irradiation and measured at 77 °C, for the as-grown (M1) as well as for the initially HTHP treated material (M3). Figure 1b presents the IR spectra at 350 °C in the course of the isochronal annealing sequence both for the M1 and the M3 samples, although in the insert a smaller fragment of the spectrum after annealing at 500 °C is depicted. Very strong signals as those of the well-known bands of Cs and Oi at 605 and 1107 cm−1 respectively are not shown, as well as strong signals from the VO and the CiOi defects at 828 and 862 cm−1 respectively are shown partly, in order to exhibit more clearly weak signals from other defects and facilitate their study. The present work is focused on the study of (i) two bands at 592 and 883 cm−1 (Fig. 1a) arising in the spectra immediately after irradiation and showing similar annealing behavior in as grown and HTHP Si material (ii) two bands at 535 and 556 cm−1 (Fig. 1b) arising in the spectra at ~ 280 °C in the course of the isochronal anneals and showing again similar annealing behavior and (iii) four bands at 562, 642 654 and 678 cm−1 (Fig. 1b), arising in the spectra at ~ 250 °C and disappearing together. Their decay is followed by the emergence of a band at 729 cm−1 (Fig. 1b), which upon decaying out is accompanied by the emergence of a band at 1044 cm−1 (insert of Fig. 1b).

Fig. 1
figure 1

Segments of the IR spectra of the as-grown sample (M1) and the HTHP treated sample (M3) at 77 °C (a) and at 350 °C (b), correspondingly. The inset in (b) depicts a part of the spectra at 500 °C

Full size image

(i) The 592 and 883 cm−1 pair of bands emerge immediately after irradiation and begin to decay at ~ 275 °C in the course of the 20 min isochronal annealing sequence and disappears just below at ~ 350 °C, without the concomitant growth of new peaks (Fig. 2). One way to determine the origin of unknown IR bands is to make comparisons with previously reported bands associated with certain defects, in an attempt to exclude some of them and limit the range of investigated structures as potential candidates. Studies of IR spectra of neutron irradiated Cz–Si contained carbon reveal the presence of well-known bands originated from oxygen-related (mainly VO, V2O and V3O), carbon-related [mainly CiOi, CiCs, CiOi(SiI) and CiCs(SiI)] defects, as well as V2 and self-interstitial clusters. VO, V2O and V3O measured at room temperature have LVMs at ~ 830, 826 and 839 cm−1 respectively [30, 31]. The three bands appear after neutron irradiation and anneal out [31] at ~ 350 °C, as our 592 and 883 cm−1 bands. We notice at first, that there is a substantial deference in the LVM values. 592 cm−1 band has a value far away from that of the above three vacancy-oxygen bands. The value of the 883 cm−1 band is also quite larger than those of the three bands. Importantly, our two bands behave as a pair and demonstrate a significant annealing behavior in HTHP treated material showing a shift in the annealing behavior towards lower temperatures, in comparison with untreated Cz–Si (Fig. 2). We shall discuss this behavior extensively below. The 592,883 cm−1 bands apparently originate from the same defect. However, each of the three mentioned bands exhibit one characteristic LVM band which is related to the VO(830), V2O(826) and V3O(839) structures, respectively. Thus, our pair of bands is not related with the VO, V2O and V3O group of defects. Additionally regarding the V2 defect, the 2767 cm−1 band related to electronic excitation of V2 anneals out [30, 31] at ~ 250 °C. Apparently, it cannot be connected with the origin of the 592, 883 cm−1 bands. Let us now discuss the carbon- related defects. The two main defects are the CiOi, and the CiCs. CiCs pair is considered [32] to be bistable and eleven LVMs has been correlated with it. The bands are very weak and detected only in low temperature measurements [32], besides a band at ~ 544 cm−1 which has been seen [33] also at room temperatures. The LVM values of these bands are not close [32] to the values of ours two bands and therefore cannot be considered as candidates. With similar arguments the CiOi, with at least six LVM bands [34], and the CiOi(SiI) and CiCs(SiI), with two bands each [34], are also excluded. The last important family of defect is the self-interstitial clusters.

Fig. 2
figure 2

The thermal evolution of the 592 and 883 cm−1 bands

Full size image

Our two bands show similar thermal stability [35,36,37,38] as that of the W center in Si. Experimentally the center is observed by photoluminesce (PL) measurements, (W 1.018 eV PL line) [35,36,37,38,39] and electron paramagnetic resonance (EPR) measurements, (Si–B5 EPR line) [40]. To the best of our knowledge no IR bands related with the W-center have been reported. The W-center is expected to be present in neutron irradiated Si, but it is not detected by IR technique. The origin of the later line is self-interstitial clusters, and although the exact atomic configuration is still controversial the most probable candidate is a tri-interstitials (I3) center [37,38,39]. A single LVM line with 70 meV energy (~ 566 cm−1) has been associated with the later defect [41] and density functional calculations have correlated [11] LVM frequencies at ~ 625, 586 and 550.6 cm−1 with the W line. The one band of the pair at 592 cm−1 is located within the above frequency range. However, there are important differences with the properties and behavior between the 592 cm−1 and the W line. Namely, for the tri-interstitial cluster migration energy of ~ 1.7 eV was calculated [42, 43], although we found an activation energy of 2.5 eV for the 592 cm−1 line. The other band of the pair at 883 cm−1 has a frequency value far away from that of the W line. We consider that the two bands originate from a defect involving a tri-interstitial structure perturbed by an impurity atom. Interestingly, a lot of EPR signals related with interstitial-impurity defects involving various impurities as for instance O, C, Al, B and Ga have been previously reported [8]. Note that the carbon impurity readily associates 34 with self-interstitials. However, not all the self-interstitial combine with carbon. In the literature, it has been proposed that small interstitial clusters, as for instance I3 as well as I4 defects, have multiple configurations [44, 45] and for the case of I3 center a combination between I3 and carbon atoms cannot be excluded. Apparently the attribution of our two bands to an I3C complex is tentative pending further experimental and theoretical investigation.

It can be assumed that a structure embodying a tri-interstitial complex together with a loosely bound carbon interstitial impurity, may give rise to the 592, 883 cm−1 pair of bands. In this assumption, the first LVM frequency could originate from the vibration of the tri-interstitial structure where the presence of Ci has caused a shift of the expected 566 cm−1 line of the W complex towards the observed 592 cm−1 line. The second LVM frequency could originate from the vibration of the Ci impurity [34] where the presence of the tri-interstitial has caused a shift of the 922 cm−1 line (or 932 cm−1) of Ci towards the observed 883 cm−1 line. Such an assignment may also account for the observed deviation between the reported above value of the activation energy of the W line and that of our findings. Notably, the W line is shifted [35, 37, 39, 46] by nearby trapped impurities, the presence of which can cause changes in its properties and behavior. Of course, the above attribution relating the 592 and 883 cm−1 bands with a self-interstitials structure complexing with an impurity atom, possibly carbon, should be verified by further experiments.

(ii) The 535 and 556 cm−1 bands emerge in the spectra in the course of the 20 min isochronal annealing sequence around ~ 320 °C and disappear around ~ 430 °C (Fig. 3). Their growth does not accompany the decay of another peak in our spectra. In an attempt to establish their origin, we need again to exclude connection with certain structures. In about the same temperature range as that of our two bands defects as VO2, V2O2 and V2O3 appear [47, 48] in the IR spectra. These structures have IR bands above 800 cm−1 and are excluded to be related with our bands at 535 and 556 cm−1. Similarly for V2. Regarding the CiOi and CiCs pairs, which also decay in the spectra around the temperatures where the 535 and 556 cm−1 bands emerge, we have to notice the following. They anneal out mostly by dissociation [34]. There are no any relative IR signals growing in the spectra after the anneal out of the CiOi, CiCs pairs. Thus any association of our two bands with defects produced with the destruction of the CiOi, CiCs pairs cannot be established. What remains to be examined are self-interstitial clusters and in particular to investigate any possible correlation with the 535 and 556 cm−1 bands.

Fig. 3
figure 3

The thermal evolution of the 535 and 556 cm−1 bands

Full size image

At this point, it is worth noting that the well-known X (1.040 eV) PL center arises in the spectra just above ~ 200 °C and disappears just above ~ 550 °C. [36, 49] The formation of the X line is linked to the anneal out of the W line and it is attributed to a tetra-interstitial cluster [36, 50]. The first thought was since (X center, I4) grows in the spectra at the expense of the (W center, I3) [36], it was reasonable to expect that I3C converts to I4C with the last defect to be a possible candidate for the origin of the 535 and 556 cm−1 lines. However, the emergence of the latter two bands in the IR spectra, in the course of the isochronal anneals sequence, does not coincide with the beginning of the decay out of the first pair of bands at 592 and 883 cm−1. The latter bands begin to arise in the spectra almost at temperatures where the 535 and 556 cm−1 lines have almost disappeared from the spectra (compare Figs. 2, 3). Thus a conversion of an I3C structure to a I4C structure cannot be supported by our data. Theoretical calculations and experimental findings [11, 41, 50, 51] have correlated LVM frequencies for the X line in the range of 530–580 cm−1. Interestingly, the LVM bands at 535 and 556 cm−1 found in this work lie within this frequency range. Nevertheless, the thermal stability of the two bands is quite different from that of the X line clearly indicating that they do not have a common origin. However, it is important to note that the LVM values of the two bands are within the spectral range where LVMs from self-interstitial cluster occur. Additionally, measurements on implanted Si have shown [51] the existence of various self-interstitial clusters having thermal stability between that of W line and that of the {311}defects around 650 °C, acting as precursors for the formation of the latter defect. More generally, self-interstitials tend [52] to coalesce into more complex defect structures in the implanted material. The multiplicity of observed defects corroborates the assumption [53] that a range of interstitial complexes exists in the annealed samples. In the case of heavily neutron irradiated Si, large self-interstitial clusters form or/and self-interstitials are trapped near disordered regions [54] formed during irradiation. These disordered regions release the stored self-interstitials at higher temperatures, in the course of annealing. Notably, two IR bands at 530 and 550 cm−1 were detected in neutron irradiated Si [54] and theoretical calculations [50] have correlated them with the tetra-interstitial defect. This assignment was supported by ESR data [49]. Interestingly, our pair of bands at 535 and 556 cm−1 have similar values correspondingly with the above pair at 530 and 550 cm−1 and one could connect our bands with the tetra-interstitial complex (X center). However, 530 and 550 cm−1 bands evolve [50, 51] in the range of 200–550 °C. In comparison, our two bands at 535 and 556 cm−1 evolve in the range 320–430 °C. This indicates that the two pair of bands may have differences in their origin. In this line of thought, the suggestion that our two bands may originates from a complex comprising an I4 defect with an impurity atom nearby seems reasonable. Notice that according to our proposal regarding I3C defect, the carbon atom that binds with the I3 defect is at an interstitial position. Reasonably, carbon cannot be excluded from the structure responsible for the 535 and 556 cm−1 lines. It may be suggested an I4C defect that comprises for instance a substitutional carbon atom at position next to an I4 structure. The latter impurity has an IR band [34] ~ 605 cm−1 a value which is close to our bands. This attribution is pending further experimental and theoretical verification.

Interestingly, in an experiment of neutron irradiated Si next subjected to 20 min isochronal anneals of 20 °C steps, the detection of seven lines was reported [55] in the energy range (0.955–1.018) eV formed by annealing in the temperature range of (140–240) oC, and were annihilated at 380 °C. It was stated that their origin is related with structures of comparatively complex nature. These defects show similar thermal behavior with that of our two bands at 535 and 556 cm−1.

(iii) The effect of high pressure treatment. In this section we shall consider the effect of HTHP treatment on the annealing behavior of the two pair of bands discussed above in sections (i) and (ii). Figures 4 and 5 show the Arrhenius plots for the calculation of the activation energies of the (592, 883) cm−1 and (535, 556) cm−1 pairs of bands, respectively. The data fit with a first order kinetics. Regarding the first pair the calculated activation energies for the 592 cm−1 band are 2.5 and 1.8 eV for the as-grown and the HTHP Si sample, respectively. Concerning the activation energies for the 883 cm−1 band the corresponding values are 2.3 and 1.8 eV for the as-grown and the HTHP Si sample, respectively. The similar values of the activation energies of the two bands are an additional indication of a common origin. The resemblance in the change of the activation energy for the two bands as a result of the HTHP treatment further supports this assumption. Regarding the second pair the calculated activation energies for the 535 cm−1 band are 2.0 and 1.4 eV for the as-grown and the HTHP Si sample, respectively. The activation energies for the 556 cm−1 band are 2.2 and 1.6 eV for the as-grown and the HTHP Si sample, respectively. Again these two pair of bands seems to have a common origin.

Fig. 4
figure 4

Arrhenius plots for the decay of the 592 and 883 cm−1 bands

Full size image
Fig. 5
figure 5

Arrhenius plots for the decay of the 535 and 556 cm−1 bands

Full size image

Importantly, a reduction is observed in the activation energies of the bands in the HTHP treated material in comparison to those of the as-grown one, which is also manifested in the evolution curves by a shift of the annealing temperature to lower values. It is well-known that the application of high pressure has a significant impact on a crystal [56,57,58,59,60,61,62,63]. Among others, it can cause changes in the lattice constant, in the bond lengths and angles of a crystal as well as in its elastic constants. It also causes structural phase transitions, and in the case of semiconductors modification of the band gap allowing for band gap engineering. Additionally, it can lead to the generation of new defects, increase the concentration of already existed defects, induce changes in their structure, and affect their formation energies and their diffusion parameters. In general, high pressure either hydrostatic or uniaxial induces changes in the electrical, optical, mechanical and structural properties of the material helping in getting insight in the properties of solids. In essence, the material parameters change as a result of the HP treatment and the material is not exactly the same after the application of the pressure [57, 59]. Regarding the annealing of defects, the effect of HTHP treatment can cause changes in the activation energy of the process which are either of kinetic origin by altering the reaction channels involved or of thermodynamic origin by changing directly the energy barrier that characterizes the annealing.

In the first case, one possible reason for the reduction of the activation energies is the appearance of an additional trap in the HTHP Si material that creates an additional reaction channel, thus enhancing the annealing of the relative defects. Note that thermal treatments of Si at high temperatures lead to the formation of oxygen precipitates [21] although HTHP treatments enhance the formation of these precipitates [64] Their formation is accompanied [21] by the emission of self-interstitials, to accommodate for the developed strains in the material. They are bound at the precipitate/matrix (Si/SiO2) interface and these regions upon annealing are potential sources of self-interstitials liberating them at higher temperatures, thus initiating additional reactions. Furthermore, it has been reported [56] that in HT-HP treated Si subjected to neutron irradiation some small defects of unknown nature are created which upon annealing may provide the additional agent to modify the decay out procedure of our defects. In the second case, it is possible that the applied HTHP treatment affects directly the activation energy of the annealing by changing the energy barrier of the process. Actually, the crystal environment of the defects has changed as a result of the HTHP treatments and this is assumed to affect the annealing behavior of the radiation induced defects. Admittedly, as a result of the treatments the induced changes in the crystal matrix are expected to affect the movement of the defects upon annealing and therefore the activation energies of the defects. The phenomenon characterizes each particular defect, depending on its structure and the mechanism involved in the annealing out process. Indeed, theoretical calculations [65] have predicted changes in the diffusivity of dopant-vacancy and dopant-interstitial pairs in compressed medium reflecting changes in the values of the corresponding activation energies. In our case we do not know positively the exact structure of our defects that is the chemical nature and the geometry of the defects that give rise to our bands. Also we do not know the mechanisms that govern their annealing, that is if the defects migrate as an entity or for instance dissociate. It is reasonable however to anticipate that the rate r of the annealing process can be generally expressed [66,67,68] by a relation r = ro exp (− ΔE/kT) where ro is an effective frequency and ΔE is the activation energy characterizing the process. In the framework of this analysis the applied stress of 11 Kbars it is assumed [69] to have a minor effect on the variation of ro since the variation of the lattice constant is small (~ 0.1 Å) and we shall only consider changes in the activation energy of the process. In that case the decrease of the activation energy in the compressed material is manifested in the spectra by the annealing out of the bands at a lower temperature, as it occurs in our case for the HTHP material. It is noteworthy that an analogous phenomenon has been observed for the case of the oxygen in Si, where the application of hydrostatic pressure results in a linear decrease [69] of the diffusion barrier of oxygen in Si. The formation energies of the defects are also expected to be affected [70, 71] by the applied external pressure. This is manifested in our spectra by the formation of the bands at a lower temperature. It is depicted by the shift of the evolution curves of the (535, 556) cm−1 bands to a lower temperature range (Fig. 3). In the case of the (592, 883) cm−1 pair of bands it seems that the effect of the pressure on the formation energy is almost negligible, since the temperature that the bands begin to emerge in the spectra is practically the same (Fig. 2).

(iv) Fig. 6 demonstrates the evolution with temperature of a group of four bands at 562, 642, 654 and 678 cm−1 having similar annealing behavior. They appear in the spectra at ~ 250 °C and disappear at ~ 400 °C. Their decay out is accompanied in the spectra by the concomitant growth of a band at 729 cm−1 which upon disappearing give rise to another band at 1044 cm−1. Comparing as-grown (Fig. 6a) and initially HTHP treated samples (Fig. 6b), any temperature shift in the evolution curves was not observed. Again, with the same reasoning, as with the other groups above, any association of the four bands with VnOm and C and O-related structures should be excluded. In the present status of analysis of the results, a potential candidate that could be correlated with to these four bands is defects involving self-interstitials in their structure.

Fig. 6
figure 6

The thermal evolution of the 562, 642, 654 and 678 cm−1 bands as well as those of the 729 and the 1044 cm−1 bands

Full size image

Let us see first the effect of HTHP treatment on the annealing behavior of the bands. The data fit with a first order kinetics. Figure 7 presents the Arrhenius curves of the group of the 562, 642, 654 and 678 cm−1 bands. It is immediately seen that the values of the activation energies of the first two bands at 562 cm−1 (1.1 eV), 642 cm−1 (~ 1.04 eV) are very similar both for the as-grown and the HTHP treated samples. For the third band at 654 cm−1 there is a small difference in the activation energies, namely 1.4 eV for the as-grown and 1.2 eV for the HTHP treated sample. These values for the 654 cm−1 band are at ~ 20% larger than those of the 562 and 642 cm−1 bands, demonstrating a stronger influence of the pressure on the annealing process of the 654 cm−1 band. However, regarding the fourth band at 678 cm−1 the activation energy is 0.76 eV for the as-grown while 0.91 eV for the HTHP treated sample. Namely, the activation energy is larger in the HTHP sample. This behavior is the opposite than that of the other three bands. Notably, depending on the nature of the diffusing element the activation barrier for diffusion may decrease as in the case of oxygen [69] or increase as in the case of interstitial hydrogen molecule [72].Additionally, it has been shown [65] that in a compressive medium, the diffusivity of dopant-interstitial pairs decreases while of dopant-vacancy pairs increases. The different trend in the variation of the activation energy of the four bands upon HTHP treatment suggests that the bands although they have similar annealing behavior they may have different origin, at least the 678 cm−1 band.

Fig. 7
figure 7

Arrhenius plots for the decay of the 562, 642, 654 and 678 cm−1 bands

Full size image

Searching now for clusters involving self-interstitials in their structure it important to note that boron exhibits a strong tendency to react promptly with self-interstitials. We note that substitutional boron has [73] a vibration frequency at 623 cm−1 which is in the region of our four bands. Thus, it cannot be ignored the formation of boron-self interstitial (Bn–SiIm) clusters as a result of neutron irradiation and subsequent thermal treatments of boron-doped Si. The formation and evolution of such clusters has been extensively studied [73,74,75,76,77,78,79] both theoretically and experimentally and in relation with the phenomenon of the transient enhanced diffusion process of boron. There is a lot of theoretical work about the Bn–SiIm defects and experimental work dealing with the effect of boron on the formation of various PL detected defects as for instance the W and X lines in Si [73,74,75,76,77,78,79]. To the best of our knowledge no IR signal related with Bn–SiIm defects have been reported. Although we cannot correlate our bands with any particular the PL or/and IR signal related with certain boron-self interstitial defects, we wish to emphasize that there are strong indications for the generation of boron interstitial clusters (BICs) upon irradiation and/or implantation, which in the course of thermal treatments lead to the creation [73,74,75,76,77,78,79] of various (Bn–SiIm) structures. Thus, the latter structures appear as potential candidates for been relating with the 562, 642, 654 and 678 cm−1 bands. Our assignment is tentative pending further verification by other experimental techniques.

Furthermore, the decay of our 562, 642, 654 and 678 cm−1 bands is accompanied in the spectra by the emergence of a band at 729 cm−1 which appears in the spectra between 330 and 480 °C. The analysis for the 729 cm−1 band gives activation energy of 2.0 eV for the as-grown and 1.9 eV for the HTHP treated sample (Fig. 8). In its turn, the decay of the 729 cm−1 band is followed by the growth of another band at 1044 cm−1 which appears in the spectra between 430 and 580 °C, (Fig. 6). Furthermore, the activation energy for the annealing out of the 1044 cm−1 band is 1.4 eV for the as-grown, although 1.1 eV for the HTHP treated sample (Fig. 9), indicating a stronger influence of pressure on the annealing of this band in comparison with the 729 cm−1 band. Moreover, the amplitude of the band in the HTHP sample is about half as that in the as-grown sample. It seems that the application of pressure in the course of the heat treatment affects also its generation.

Fig. 8
figure 8

Arrhenius plot for the decay of the 729 cm−1 band

Full size image
Fig. 9
figure 9

Arrhenius plot for the decay of the 1044 cm−1 band

Full size image

We tend to suggest, as a general picture derived from the discussion above, that the bands under investigation could be related to small self-interstitial structures complexing with impurity atoms as C, B. We should note at this stage, that there are important theoretical works [80,81,82] investigating the effect of temperature and pressure on the formation and aggregation of self-interstitial clusters in Si. Their results could provide additional information on the behavior and properties of such complexes. Still, the exact knowledge of their identity is difficult. Notably, the origin of the well-known PL lines W (1.018 eV) and X (1.040 eV) is still unclear [38, 39, 83] despite the intense studies in the last decades. Even their evolution may be perturbed [83] by the presence and action of other coexisting defects which mask their exact behavior. Apparently, further work comprising additional experimental techniques and coupled with theoretical calculations is necessary to make definite assignments for the identity of the defects that give rise to of all these bands.

The efficiency of Si-based electronic devices depends crucially on the defects introduced in the course of material processing. Thus, it matters a great deal to understand the properties and behavior of defects in the material, allowing for their control. The aim of this work was the study of defects introduced in Si as a result of irradiation and thermal treatments, two very important stages of material processing. High pressure was used as a tool in the course of thermal anneals to acquire information in establishing the origin of the produced defects. On the other hand, stress is inevitable in Si material used for the fabrication of integrated circuits. Notice that strained Si has been employed [84] to improve the performance of very large scale integrated (VLSI) circuits. To this end, any information gained by the effect of pressure on the properties of the material is beneficial for making better the performance of relative devices.

4 Conclusions

We have investigated, by means of IR spectroscopy, defects introduced by neutron irradiation in Si material initially subjected to HTHP treatments. We observed some previously unreported IR bands. In particular, our study was focused on three groups of bands. The first group, a pair of two bands at 592 and 883 cm−1 having the same annealing behavior, was correlated with a tri-interstitial cluster perturbed by an impurity atom possibly carbon, at an interstitial position. The second group, a pair of two bands at 535 and 556 cm−1 having the same annealing behavior, was correlated with a self-interstitial cluster presumably with an impurity atom captured nearby. This impurity may be a carbon substitutional atom. The activation energies are smaller in the HTHP treated samples than those of the as-grown ones for each band, respectively. It is suggested that the application of high pressure in the course of the thermal treatment promotes their annihilation. The third group, four bands at 562, 642, 654 and 678 cm−1 having the same annealing behavior was studied in detail. The HTHP treatments indicated that these four bands may have not exactly the same origin, in particular the 678 cm−1 band. Small self-interstitial structures complexing with B, in short (Bn–SiIm) clusters were considered as potential candidates for the origin of the bands. Further theoretical and experimental investigations are necessary to verify the suggested attributions for the above three group of bands.