Correlation between Morphology and Cathodoluminescence in Porous GaP

A very good correlation between morphology and cathodoluminescence in porous GaP is described.

4.1.2 Correlation between Morphology and Cathodoluminescence in Porous GaP

SEM micrograph of the top surface of a  GaP sample anodically etched for 10 min at a constant voltage of 10 V


Figure 4.1: a) SEM micrograph of the top surface of a GaP sample anodically etched for 10 min at a constant voltage of 10 V. It shows the so-called catacomb-like porous structure; b) A schematic representation of the catacomb-like pores obtained in n-GaP during the anodization process at constant voltages. The primary pore is growing perpendicularly to the surface, whereas the secondary pores grow radial away from the primary pore. The distance between secondary pores is 2W, where W is the width of the space charge region in the semiconductor.


The SEM images presented in Figure 4.1 and Figure 4.2a show the development of porous regions in n- GaP samples subjected to anodic etching under potentiostatic conditions for 10 and 120 min respectively. In both cases a constant voltage of 10 V was applied to the sample. At the beginning of the process the etching starts at surface imperfections forming the so-called catacomb-like pores [28]. After initial pitting of the surface, further dissolution proceeds in directions both perpendicular and parallel to the surface. A pore starting at a surface imperfection and growing along a current-line, i.e. mainly perpendicular to the initial surface, is called a primary pore. Pores originating at the surface of a primary pore and propagating away from it are called secondary pores. The development of secondary pores occurs underneath the initial surface.

As it can be seen from Figure 4.1a, the secondary pores in n-GaP propagate radially away from the primary pore forming a symmetric set of 'catacombs'. Since the etching proceeds at the same rate in all directions, the boundary of the porous region or in other words the porous domain is circular. In comparison with the domains observed in n-GaAs and n-InP, where the domains can be rectangular and the pores are strictly oriented along <111B> directions, the catacomb-like pores in GaP expose no specific crystallographic features.



SEM and  CL micrographs taken from the top surface of  n- GaP

Figure 4.2: SEM and CL micrographs taken from the top surface of n- GaP (n=3×10e17cm-3) sample anodically etched for 120 min at a constant voltage of V = 10 V. For CL images the beam energy was 15 keV, and the beam current was 0.025 nA a) SEM micrograph. The white borders of the domains are twice the width of the space charge region. The domains are not round anymore as in Figure 4.1 because the etching stops when two domains meet each other. Therefore, the round shape is destroyed. b) CL panchromatic images, i.e. all wavelengths are used for imaging. The domains again can be easily observed on the CL micrographs and the similarity to the SEM is evident. In this case the borders between the domains are dark, which means that the CL efficiency is lower.


A schematic representation of the catacomb-like pores in n- GaP is presented in Figure 4.1b. The distance between the secondary pores is twice the width of the space charge region. The first set of radial pores begins growing underneath the initial surface. Note that, when looking with the naked eyes at anodized samples, one gets the impression that the surface is intact.

Following etching for an extended period, the secondary pores from different domains eventually meet, leaving nearly straight walls between neighboring porous domains (Figure 4.2a). The thickness of these walls is also defined by twice the thickness of the surface depletion layer during anodization [28]. As it can be seen in Figure 4.2a, the lateral dimensions of porous domains depend upon the local density of surface imperfections (e.g. dislocations, scratches etc.) initiating the formation of primary pores.


 Panchromatic  CL image taken from the top surface of a  GaP sample anodized under temporary variations of the applied voltage


Figure 4.3: Panchromatic CL image taken from the top surface of a GaP sample anodized under temporary variations of the applied voltage from 5 to 15 V. The beam energy is 15 keV, the beam current is 1 nA. The variation of the voltage is accompanied by the change in porosity and by the modulation of the CL intensity (see the rings). It is quite obvious that the CL intensity depends on the porosity.


Figure 4.2b shows a panchromatic CL image taken from the same region of the as-anodized n- GaP sample from Figure 4.2a. It clearly shows a porosity-induced increase of the emission efficiency of gallium phosphide. The light areas in the panchromatic CL image result from enhanced luminescence collected for wavelengths between 250 and 900 nm, with response maximum at ~550 nm (2.25 eV). It is obvious that there is a good correlation between the enhanced CL emission (bright areas) in the panchromatic CL image and porous domains in the SEM image. The walls between porous domains are less luminescent and therefore are easily distinguishable in Figure 4.2b as dark curves.


CL spectra of bulk and  porous  GaP measured at accelerating voltage of 25 keV

Figure 4.4: CL spectra of bulk and porous GaP measured at accelerating voltage of 25 keV and current intensities 10 nA (triangles) and 50 nA (squares). The CL intensity from the bulk material is lower than that from the porous one. The band at ~2.25 eV is more pronounced when increasing the intensity of the incident electron beam.


To demonstrate the dependence of the emission efficiency upon the degree of porosity, a panchromatic CL image was taken from a sample rendered porous under etching conditions, when the applied voltage was varied in time between 5 and 15 V. The modulation of the applied voltage is accompanied by an increase and decrease of the anodic current density that causes synchronous modulation of the degree of porosity. As one can see from Figure 4.3, successions of high and low voltages lead to the observation of annular bright and dark traceries around most of the observed etch pits. Note that the dark areas in Figure 4.3 correspond to regions unaffected by anodization.


In Figure 4.4 the CL spectra from bulk and porous GaP fabricated at a constant voltage of 10 V are shown. The spectra are dominated by a broad band at ?1.5 eV and a lower intensity band at ?2.25 eV. It was found that increasing the beam power, i.e. Eb×Ib, increased the intensity of the emission at 2.25 eV in absolute terms and relative to the ?1.5 eV emission in both bulk and porous GaP. These two broad emission bands may be attributed to sulfur and some residual impurities including C, O, Zn and Cd [69]. The emission observed at ?2.25 eV has been investigated by a number of groups [70, 71, 72] and was associated with the radiative recombination at sulfur and carbon atoms on phosphorus sites. Note that, according to the earlier published data on luminescence, deep level transient spectroscopy and optically detected magnetic resonance in n- GaP [73, 74], the 1.5 eV band has been attributed to the radiative recombination of non-equilibrium carriers via donor-acceptor pairs, the donor being a shallow centre.


SEM and panchromatic  CL images in cross-section


Figure 4.5: a) SEM and b) panchromatic CL images in cross-section taken from a sample anodized galvanostatically at two current densities: j1=80 mA/cm2 for 60 min and j2=1 mA/cm2 for 240 min. Due to the differences in size of the pores at the two current densities, it is not possible to see explicitly the second layer (with crystallographically oriented pores). However, a great difference in CL intensity is observed between bulk and the porous layer with Current-line Oriented Pores. More than that, in the CL image it can be observed that there is a difference in CL intensity between the bulk and the porous layer obtained at low current densities (crystallographically oriented pores): CL(bulk)>CL(crystallografic pores).


Further evidence of the impact of porosity upon the luminescence efficiency was obtained by investigating samples prepared under galvanostatic etching conditions at different anodic current densities. Figure 4.5 shows SEM and CL images in cross-section taken from a sample subjected to successive anodization steps at two current densities: j1=80 mA/cm2 for 60 min and j2=1 mA/cm2 for 240 min. As discussed in Chapter 3.2, etching at high anodic current density leads to the formation of the so-called Current-line Oriented Pores (Figure 4.5a). Under these conditions the high density of etching pits at the initial surface excludes the formation of porous domains. When the current density is switched down to j2=1 mA/cm2, the rate of dissolution and the degree of porosity sharply decreases. As discussed in Chapter 3.2, the pores in this case grow along <111>B directions (see Figure 3.15).


CL spectra of  porous layers produced at current densities j1=80 mA/cm2

Figure 4.6: CL spectra of porous layers produced at current densities j1=80 mA/cm2 (curve 1) and j2=1 mA/cm2 (curve 2). The beam energy is 15 keV, the beam current is 0.025 nA.


In perfect agreement with the data obtained in potentiostatically anodized samples a strong correlation between high porosity and high CL intensity is found. As it can be seen from Figure 4.5, the top layer fabricated at the current density j1 is more luminescent than the second layer produced at the current density j2. Even non-uniformities in the thickness of the top porous layer and small particles related to the cleavage process are clearly distinguished in the CL image. The CL spectra from both layers show luminescence in the near infra-red region with the band maximum at ?1.5 eV when excited by a low-current electron beam (Figure 4.6). The near-band-edge CL at ?2.25 eV was observed at high beam currents. Under intense excitation, however, the CL from the porous layers was considerably attenuated by the electron beam. The dynamic of CL attenuation with the time is illustrated in Figure 4.7.



The time evolution of the  CL attenuation.

Figure 4.7: The time evolution of the CL attenuation. The insert shows the time dependence of the intensities of the near infrared (1.5 eV) band and UV emission (integrated from 2.75 to 4.2 eV). The beam energy is 25 keV, the beam current is 50 nA.


Another interesting feature of the anodization process at high constant current densities is the oscillations of voltage in time, similarly to the oscillations observed in n-InP, see Chapter 3.4. Figure 4.8 illustrates the time dependence of the voltage measured on the sample during anodization. Although it is difficult to find a direct correlation between peaks visible in the voltage/time diagram and horizontal trajectories in the SEM image, it is highly probable that the self-induced voltage oscillations are responsible for the synchronous modulation of the pore diameters as in the case of InP.


CL images in cross-section taken from a sample anodized galvanostatically at two current densities

Figure 4.8: a, b) SEM and c) panchromatic CL images in cross-section taken from a sample anodized galvanostatically at two current densities: j1=100 mA/cm2 for 30 min and j2=1 mA/cm2 for 240 min; d) As in the case of n-InP at high current densities, self-induced voltage oscillations are observed. Voltage oscillations induce the pore diameter oscillations (see b an c), although the correlation between the diameter and voltage oscillations is not so evident as in the case of n- InP.


A subject of particular interest is the difference in CL intensities between the bulk and the porous layer with crystallographically oriented pores. Although, neither in Figure 4.5b nor in Figure 4.8c the difference in CL intensity between the bulk and crystallographically oriented pores is evident, it is confirmed by the CL spectra. This is shown in Figure 4.9, where CL spectra of bulk GaP and porous layers produced at high and low current densities are illustrated. Although all the spectra are dominated by a band with the maximum at approximately 1.5 eV, they give further evidence that the top porous layer exhibits the most intense luminescence. As to the second layer produced at the current density 1 mA/cm2, it shows less luminescence in comparison with both the top porous layer and bulk GaP.


CL spectra of bulk  GaP

Figure 4.9: CL spectra of bulk GaP (curve 1) and porous layers produced at current densities j1=100 mA/cm2 (curve 2) and j2=1 mA/cm2 (curve 3). The beam energy is 25 keV, the beam current is 50 nA. These spectra demonstrate clearly that the porous layers obtained at very low anodization current densities are less luminescent than the bulk GaP (curve 3).


Taking into account the effect of H-passivation of pore walls in Si, one may expect the {112} planes (forming the triangular shape of the crystallographically oriented pores) in III-V compounds to have a high stability against dissolution, i.e. they represent an easily passivated set of planes enveloping <111> directions. We suggest that in this case, as in the case of GaAs and InP, the crystallographically oriented pores obtained at the current density 1 mA/cm2 expose {112} planes which are efficiently passivated in the electrolyte. On the other hand, under ambient conditions these planes seem to be characterized by poor passivation. As a result, the non-radiative recombination of the free carriers via surface states becomes as strong as to compete with the radiative recombination processes. This explains qualitatively the decrease in the luminescence intensity obtained from porous layers at low current densities (Figure 4.9, curve 3).


In contrast, pore walls at high anodic current densities exhibit weak passivation in solution and efficient passivation under ambient conditions. However, the saturation of the dangling bonds does not explain the strong increase (relative to the bulk material) in the luminescence intensity induced by porosity. It is well known that electrochemical dissolution of n-type semiconductor materials removes preferentially the dislocations and other lattice imperfections that may play the role of non-radiative recombination centers. The decrease in the density of non-radiative recombination centers accompanied by in-situ surface passivation can explain qualitatively the observed increase in CL intensity induced by porosity. One should also take into account that GaP, being an indirect gap semiconductor, in the porous form exhibits surface related vibrations [75]. Porous structures expose larger surface, thus more surface vibrational modes can be excited and participate in the process of free-carrier radiative recombination and, consequently, increase its probability considerably.


The degradation of luminescence in bulk GaP and GaP-based light emitting diodes under neutron irradiation, intense laser excitation, etc., has been studied extensively for many years [76, 77, 78]. Particle irradiation usually introduces non-radiative recombination centers, attenuating the luminescence. However, for beam currents Ib larger than 50 nA, the emission in the porous specimens is much more susceptible to beam damage than in as-grown GaP. It is therefore possible that the pronounced CL attenuation in porous GaP is due to irradiation-stimulated out-diffusion of radiative centers. In porous layers possessing a high surface-to-volume ratio stimulated out-diffusion of impurities should obviously be much more significant than in bulk material.


Ultraviolet luminescence from porous GaP has been reported in the literature previously [27, 26]. Broad, very low intensity emission at energies higher than 2.75 eV was occasionally observed in the porous samples presented in Figure 4.7. This emission is sensitive to irradiation, i.e. is very rapidly attenuated by radiation (see the insert in Figure 4.7). Currently, there are two possible reasons for the occurrence of UV emission. According to the SEM analysis, the dimensions of the main structural entities of the porous layers are of the order of 100 nm or higher, and therefore they cannot provide conditions for quantum confinement of free carriers. However, it is feasible that the micro-porous skeleton may be covered by a thin nano-porous film in which the occurrence of quantum size effects is possible. For example, in GaAs electrochemical etching processes have been shown to result in a wide distribution of porosity, varying from micrometer to nanometer range features [79]. If this is the case for porous GaP samples, then the relatively rapid attenuation of the UV emission under the action of the electron beam (see the insert in Figure 4.7) can be attributed to local heating due to the reduced thermal conductivity of the film.


The second possible reason for the observation of UV luminescence is the formation of a thin oxide film covering the surface of the pores during anodization [80]. Electron beam induced dissociation/damage of the oxide would account for the attenuation of the emission during irradiation. Further studies are necessary to elucidate the origin of UV emission in porous gallium phosphide.



Home | Chapter 1 | Chapter 2 | Chapter 3 | Chapter 4 | Chapter 5 | Chapter 6 | Appendixes | Papers | Top
© Copyright www.porous-35.com , all rights reserved.
Quick, Simple, Free SEO CMS website builder and design software