1. Introduction

Thermoluminescence (TL) from wet synthetic silica has been explored by Guzzi et al. [1]. They have shown that in this silica presenting a low level of impurities, only one small TL peak appears at 2.7 eV after neutron bombardment. The photoluminescence induced under proton bombardment also presents a smaller second peak at 1.9 eV [2]. In both cases the luminescence from wet synthetic fused silica is very low after irradiation. The first peak is assigned to the oxygen deficient center and the second to the nonbridging oxygen hole center. These two peaks correspond to known and listed [3] intrinsic silica defects having luminescent emission in the visible wavelengths range. In this paper we report on the intense, near infra-red Transient Incandescence (TI) emission from laser beam annealed fused silica samples implanted with protons. The TI peak presents the characteristics of a TL peak that occurs at a thermal rate threshold. We show that it rises at the stress relaxation.

2. Experimental set up and implantation conditions

The experimental set-up used to probe silica surface emissivity has already been described [4]. The TEM₀₀ mode from a 10.59 µm wavelength continuous CO₂ laser beam is focused on the front surface from a Suprasil 2 grade silica disk. On the silica surface, the measured 300µm at 1/e waist diameter is wider than in previous experiments. Prior to the front surface implantation, the 50mm diameter and 5 mm thick silica disks, polished on both sides, are cleaned in an automatic machine that provides sequences with soap cleansing and de-ionized water rinsing. As silica, like most oxides, absorbs at 10.59 µm we expect the temperature to rise, initially from room temperature, under the rapid laser anneal the duration of which is controlled by a mechanical shutter.

When mapping the incandescence, the silica sample is moved in front of the laser beam at a constant speed of 2mm/s. The near infrared thermography diagnostic is fixed and collects radiatively emitted photons from the front surface plane through the transparent sample. The uncooled InGaAs detector’s spectral response, 0.9 to 1.65 µm is well suited for silica’s transparency window. This configuration enables a total filtering of the 10.59 µm excitation photons and an on-axis imagery. The temporal sampling of incandescence along the Y axis is at 20 Hertz which results in spatial resolution of 100 µm over a 40 mm line. Successive scanning lines are juxtaposed at a 200 µm pitch along the X axis, and finally create a 2D, (XY) plane, 40*40 mm² TI mapping. The TI images are range-limited and coded over 8 bits, 256 grey levels.

Two fused silica samples, E1 and E2 have been implanted with protons at room temperature, with an Axcelis NV-8200P medium current ion implanter under the Table 1 conditions.

Table 1. Implantation conditions

View Table

The sample E1 has been used to observe the incandescence temporal behavior when sites are submitted to a thermal step and the sample E2 with the incandescence mapping technique. In E2, the three proton doses are implanted, at the maximum acceleration energy available with the ion implanter, in juxtaposed regions which are defined by masking tape. In each sample, the implanted protons induce a stoichiometry perturbation of less than 1% inside a small embedded layer of a few tens of nanometers. What is more important is the silica compaction induced by the lattice atom displacement. In bulk silica, the ion implantation induces a densification and a tensile stress below the surface [5].

3. Results and discussion

The incandescence temporal behavior, from the E1 sample, is recorded at four adjacent sites submitted to a one second duration thermal step at a different laser power. In each temporal spectrum, obtained from a different site, Fig. 1 , the incandescence exhibits a peak, more intense and at shorter time as the laser power increases.

 

Fig. 1 Incandescence signal from juxtaposed sites in stressed silica with increased excitation power. The dotted lines represent the calculated background incandescence at each excitation power.

Download Full Size | PDF

After the peak occurrence the incandescence intensity settles, either decreasingly for the lowest laser power or increasingly for the highest laser power, to a quasi-stationary emission. At one second the shutter closes and the incandescence signal decreases to zero in a few milliseconds. The TI peaks are absent from the unimplanted fused silica incandescence spectra [4] or even from post recorded spectra on the same sites. The background incandescence deduced at the different laser powers, with a fitting procedure described in [4], is represented by the dotted line curves in Fig. 1. The TI intensity can be extracted and plotted versus the inverse time at which the peak occurs, in Fig. 2 with the right ordinate scale. The TI intensity is inversely proportional to the time at which it occurs. This is a first characteristic of a TL peak [6]. The higher the heating rate is, the sooner and the more intense the TI is. On the other hand, when the heating rate is lower, the TI should occur after a longer time with a lower intensity. This is not what is determined from the experimental data. The incident laser power is plotted versus the TI intensity in Fig. 2 with the left ordinate scale. The TI intensity is proportional to the laser power but below a laser power of 0.95W, the extrapolated peak intensity is null. In a unimolecular relaxation mechanism like TL, the peak intensity is proportional to the heating rate [6]. So the TI peak linearity with the laser power, above a determined laser power, finds its justification as a TL emission only if a thermal rate threshold exists. No heating to a critical temperature would lead to this result because the TL peak intensity is at least inversely proportional to the temperature [6]. In our experimental conditions, the laser power of 0.95 W corresponds to a heating rate of $\stackrel{•}{T}\approx 2\cdot {10}^{4}K\cdot s^{-1}$. This is two orders of magnitude above the thermal ramps used in previous laser induced TL experiments [7].

 

Fig. 2 Incident laser power and peak inverse time versus transient incandescence intensity

Download Full Size | PDF

Below this heating rate threshold there is no TI but an incandescence background.

If, instead of a thermal step on a fixed site, we scan dynamically the E2 sample, the thermal excitation on the spatially sampled sites changes to a thermal pulse [8]. The E2 sample has been line scanned first at a 0.8W laser power, below the thermal rate threshold, and then at a 1.4W laser power, above the thermal rate threshold, with a 0.1mm X spatial offset in order to probe the same sample on intertwined lines.

The two incandescence mappings are shown in Fig. 3(a) . In each implanted zone, a square area of 4*4 mm², outside “polishing defect areas,” can be defined, in which incandescence is spatially averaged for both laser powers. The incandescence intensity in these different zones versus the implanted dose is presented in the Fig. 3(b). The incandescence is proportional to the implanted doses but only at the 1.4W laser power. The TI proportionality with the dose is a second characteristic of a TL peak. The extrapolation at zero dose shows that there is no significant residual TI. A TL emission occurs above a thermal rate threshold in heated stressed fused silica. But this TL emission has rather unusual characteristics. The TI is detected in the NIR wavelength range where no known silica intrinsic defects are identified. The proton implantation might create new intrinsic defects in silica with a radiative emission in the NIR but these defects, if they exist, are not observed at a lower thermal rate in our experiment. The TI is also more intense than the background incandescence. In the laser induced TL [9], increasing the heating rate increases the signal to noise ratio of existing TL peaks. Somehow a compromise has to be found between the benefit of high heating rates and the shift of the TL peak to higher temperatures in which a noticeable incandescence background appears. The TL emission is usually in competition with the incandescence rise. Pre or post irradiation anneals are needed to optimize the sensitivity of TL materials. In our experiment, the TI is more intense than the background incandescence at temperatures around 1000K [10]. This is in contradiction with the observation of usual TL peaks from wet synthetic silica or other materials. Most of the TI peculiarity arises from the existence of a kinetic thermal threshold. In the next paragraphs we show that the TI rises at the internal stress relaxation.

 

Fig. 3 (a) 2D incandescence mapping of half the E2 silica disk implanted with three protons dose at a laser power of 0.8W, left, and 1.4W, right. The gray scale intensity is reduced by 4 on the 0.8W map in order to better see the polishing defects. (b) Transient incandescence intensity from the red squared areas versus the implanted dose at the two laser powers.

Download Full Size | PDF

Three main radiation-induced effects are known to induce changes in the mechanical stress in bulk silica:

The extrapolated proton dose at which the maximum tensile stress is reached [5] in implanted silica is ~5 10¹⁵ at/cm² at 250 keV. With lower implanted doses (≤1.4 10¹⁵ at/cm²) and implantation energy (≤240 keV), the initial tensile stress from our samples is below the maximum tensile stress which means that the maximum densification is not achieved. This is the starting state from which a thermal pulse is applied. The effect of a thermal pulse into implanted silica is to induce a further densification as seen from molecular dynamics simulations [14]. As the implanted silica layer remains constrained by the silica substrate, the thermally induced densification leads to the build-up of a dynamic tensile stress which increases but cannot overstep the maximum without structural relaxation occurring. A critical density of ~2.8 g/cm³ is reached at an internal local compressive stress [15] between 5 and 10 GPa where the silica bulk modulus is at a minimum. This is where the plastic flow and stress relaxation occur: “Silica densification is like a self–catalyzing process that invites additional compaction and inevitably leads to a catastrophic condition at the molecular level,” [15]. TI is the signature of the entropy release that occurs at the stress relaxation. It was not observed or mentioned under previous ion radiation exposures [12] because the ion implantation thermal rate is nine orders of magnitude above our laser thermal rate which push the TI into the picosecond regime. That the stress is relaxed after the 1.4W laser scanning can be seen in Fig. 4 for the E2 sample at the implantation step between the 7 10¹⁴ and 1.4 10¹⁵ at/cm² implanted zones. In silica, the increasing ion dose induces a higher compaction and a negative height step [16]. When relaxed by the laser scanning, there is a positive height step as observed by a Wyko optical interferometer. This effect is better seen on this interface with the most important stress discontinuity and a TI intensity that saturates the InGaAs detector.

 

Fig. 4 Relief mapping of the interface between the 7 10¹⁴ and 1.4 10¹⁵ H⁺/cm² implanted areas. On the X profile the laser scanning at a 200 µm pitch can be observed. On the Y profile, a positive step height of 10 nm can be measured between the two zones.

Download Full Size | PDF

Anomalous vibrational modes are soft modes with large transverse displacements like the bonded proton in silica which has twice the vibrational amplitude of the other lattice vibrations [17] at the “boson peak,”. Theoretically, such anomalous modes should collapse under an internal compressive stress [18]. As the hyperquenched glass annealing is also characterized by a decrease in the vibrational density of state at the boson peak [19], the origin of the stress relaxation might be found in the anomalous soft modes collapse. This hypothesis can be verified by monitoring the atomic exhaust from heated stressed fused silica under vacuum. As the plastic flow moves atoms toward the surface to relieve the in-plane stress [20], it may be possible to observe the simultaneous implanted ions exo-diffusion as well as the TI occurring at the thermal rate threshold. The TI relationship to the plastic flow would also explain its intensity because the plastic flow has very low or null activation energy [20]. The “Bose phonons,” can also be responsible for the lattice deformation which enables the radiative emission from point defect centers. Putting a name on the phonons that initiate the TL enables us to put a terahertz limit on the electron escape frequency in TL models [21].

Transient incandescence (TI) has been discovered through the very fast CO₂ laser annealing of protons implanted fused silica surface. TI shows the characteristics of a thermoluminescence (TL) phenomenon that occurs at a thermal rate threshold. From the surface strain inversion occurring at the laser scanning between two differently stressed areas, we are able to conclude that TI rises at the stress relaxation.