Search

Electronic defects in semiconductor devices play important roles in device performance and reliability. These defects are created during fabrication using plasma-based processing technology. The authors investigate the generation and annihilation of defects in amorphous silicon by measuring the photocurrent during processing with hydrogen and argon plasmas. Unfavorable species for defect generation and favorable species for defect recovery are distinguished. These results are beneficial for improving semiconductor plasma processing, which is used to make state-of-the-art devices such as FinFETs, computer memory, and solar cells.

Carrier trapping in hydrogenated amorphous silicon (a-Si:H) has been investigated by means of an optical pump-probe technique. The trapped carriers (electrons) at the conduction band tail are detected as an increment of the photocurrent, and their density is quantitatively determined under the assumption of carrier generation and recombination kinetics. We find that carrier trapping strongly depends on the band tail distribution in addition to the pump light intensity. Specifically, the trapped electron density increases with the Urbach energy that characterizes the valence band tail broadening. Under the condition of a pump light intensity of 10 mW/cm2 operated at 532 nm, the trapped electron density is determined to be ≈ 4 × 1017 cm− 3 for an intrinsic a-Si:H film with an Urbach energy of 45 mV. The effects of carrier trapping on the device performance are studied in single-junction a-Si:H p-i-n solar cells. The results suggest that carrier trapping causes a reduction in the fill factor of the solar cells.

Surface passivation of crystalline silicon (c-Si) is experimentally studied during the growth of a hydrogenated amorphous silicon (a-Si:H) and epitaxial silicon (epi-Si) passivation layer at a subnanometer to nanometer scale. The property of surface passivation is monitored in real time via in situ measurement of a photocurrent in c-Si under plasma-enhanced vapor deposition for the passivation layer growth. The measurement results suggest the following. Passivation is improved by the growth of an a-Si:H layer, where a large band offset is formed at the a-Si:H/c-Si interface, and the carrier recombination is suppressed. On the other hand, passivation is deteriorated with the growth of an ultrathin epi-Si layer ( d ≲ 2.5 ± 1.0 nm) because the band offset is not formed at the interface, and plasma-induced defects are created in c-Si. However, passivation is improved with a thick epi-Si layer ( d ≳ 2.5 ± 1.0 nm), where the band bending is formed near the epi-Si/c-Si interface, which partially suppresses the carrier recombination. The suppression of the plasma-induced defects as well as the formation of the band offset are important for surface passivation.

Defects in hydrogenated amorphous silicon (a-Si:H)/crystalline silicon (c-Si) heterojunction have been detected via in situ photocurrent measurement during growth of a-Si:H and consecutive postannealing. During growth, defects are generated not only in the a-Si:H layer but also in c-Si. The defects in the a-Si:H layer are completely recovered by postannealing, whereas those in c-Si are partially recovered; the residual defects are formed in c-Si. Interestingly, the postannealing induces the formation of the interface defects at a-Si:H/c-Si, which may deteriorate the performances of a-Si:H/c-Si heterojunction solar cells.

Details of light-induced annealing of hole trap state in undoped hydrogenated amorphous silicon (a-Si:H) have been studied; it has been found that prolonged illumination significantly reduces the density of hole trap states in the energy range deeper than 0.5 eV, and subsequent thermal annealing increases the density of hole trap states and restored the sample to the initial state before the illumination. We can speculate, from the experimental results and discussion in this work, that defect conversion processes are taking place during the long exposure to light; Si dangling bonds are generated from the precursors or latent sites which manifested as hole trap states located between 0.5 and 0.7 eV from the top of the valence band.

Internal photoemission (IPE) has been successfully applied to evaluate band offsets of heterojunctions (HJs) in crystalline silicon (c-Si)-based solar cells. Tunneling of carriers through the potential spike at HJ and the presence of a carrier conduction path in the wide-band-gap material of HJ can affect the IPE results. In other words, IPE measures the effective band discontinuity, including effects of the carrier conduction path. This feature of IPE is suited for the characterization of solar-cell structures. Results obtained for hydrogenated amorphous silicon/c-Si HJ and gallium phosphide/c-Si HJ are presented and discussed.

We have fabricated silicon nanocrystals with different dimensions by the thermal annealing and thermal oxidation of ultrathin hydrogenated amorphous silicon (a-Si:H) films (2–10nm) deposited by thermal chemical vapor deposition. Dimensions of silicon nanodots are the function of thickness of the ultrathin a-Si:H film. Therefore, we can change the dimensions of silicon nanodots (3–10nm) by varying the a-Si:H film thickness according to our requirements. From our experimental studies, we have drawn a calibration curve of required a-Si:H film thickness against the average dimension of fabricated crystalline grains.

Experimental studies and numerical analysis have been carried out to clarify the relationship between carrier diffusion lengths and defect density in undoped a-Si:H. It has been confirmed that in device quality plasma-deposited a-Si:H, the diffusion lengths of both electrons and holes under steady-state illumination of the intensity equivalent to normal solar cell operating conditions are determined by the density of Si dangling-bond defects ranging between 3×1015 and 8×1016 cm−3. This rather trivial result, however, has not been obtained in previous studies in which the carrier transport data obtained by the steady-state photocarrier grating method were treated incorrectly. The ratio of the drift mobility of electrons to that of holes and the ratio of electron lifetime to the hole lifetime in a-Si:H under illumination have been determined and their implications discussed.

We have fabricated nanocrystalline dots by thermal annealing (TA) of thermal chemical-vapor-deposited amorphous silicon (a-Si:H) films. In S. Hazra, I. Sakata, M. Yamanaka, and E. Suzuki, Appl. Phys. Lett. 80, 1159 (2002), we observed that ultrathin a-Si:H films (2–10 nm) are stressed because of the presence of deformed crystallites or paracrystallites. With the increase of thickness, volume fraction of paracrystallites decreased in the films and stress in the films gradually reduced. Therefore, by changing the thickness, we can control the stress in the a-Si:H films and thereby the dimensions (in the range of 5 to 10 nm) as well as volume fractions of nanocrystallites formed by TA. On the other hand, it has been found that relaxed a-Si:H networks form polycrystalline films by TA.

To obtain the morphology of ultrathin hydrogenated amorphous silicon (a-Si:H) films with thicknesses of 2–10 nm grown by thermal chemical-vapor deposition, we have utilized spectroscopic ellipsometry and introduced a model, called the fine crystalline grain model. This model resolves that the growth of the amorphous silicon matrix starts with the formation of dense deformed crystallites or paracrystallites. A better fit to the experimental data has been obtained with this model when the a-Si:H film thickness is below 20 nm. The fine grain component gradually decreases with the film growth. It has been clarified that the presence of deformed crystallites in the starting a-Si:H films acts as a constraint for their crystallization by rapid thermal annealing.

Spectroscopic ellipsometry identified E1 transitions at a lower energy than that for c-Si (3.38 eV). These transitions are generated from the Si paracrystallites or disordered crystallites in the ultrathin silicon films (2–10 nm) deposited by thermal chemical vapor deposition. During the growth of the film, paracrystallites expand gradually; disorder in the paracrystallites increases. Finally, a completely disordered Si network, i.e., the amorphous network, is generated. The presence of disorder crystallites in the ultrathin Si films acts as a constraint for the crystallization of the ultrathin films by rapid thermal annealing.

Experimental studies by means of a constant photocurrent method (CPM) have been carried out on the deep defect states in undoped hydrogenated amorphous silicon (a‐Si:H). Assuming Gaussian energy distributions, two types of defect states (ST1 and ST2) have been found from careful analysis of CPM spectra; one of these states (ST1) is a neutral Si dangling‐bond (Si D0) state, and the other (ST2) is a negatively charged dangling‐bond‐like defect state located in the lower gap. ST2 lies deeper in energy by 0.1 eV, and has a narrower full width at half‐maximum (FWHM) when compared with ST1. Possible candidates for ST2 have been discussed by referring to the previous models of defects in a‐Si:H. The FWHM of ST1 (the Si D0 state) does not always depend on the overall structural disorder estimated from Raman spectra. The energy position of the Si D0 state measured from the valence‐band edge is almost independent of the optical band gap. Long exposure to light increases the density of ST1 (the Si D0 state) and decr...

Properties of hydrogenated amorphous silicon (a‐Si:H) prepared by glow discharge decomposition of silane with visible light illumination on a growing surface have been described. The visible light illumination during growth has resulted in smaller light‐induced changes, slightly lower photoconductivity before the prolonged illumination, lower content of bonded hydrogen, and lower ratio of the number of silicon dihydrides to that of total silicon‐hydrogen bonds. Formation of Si network and consequent structural disorder of SiSi bonds, on the other hand, has not been affected by the illumination during the growth. The improved stability against the long exposure to light can be ascribed to the lower bonded hydrogen content and the reduced density of SiH2 bonds, while the structural disorder of the Si network does not have a direct relationship with the light‐induced changes.

We propose a low temperature back-surface field (BSF) technology for crystalline silicon (c-Si) thin film solar cells. The BSF structure has a heterojunction between a p-type c-Si substrate and a boron (B)-doped p-type hydrogenated amorphous silicon (a-Si:H) layer deposited on the back surface of the c-Si substrate at 200degC. The back-surface recombination velocity, Sb, of minority carriers can be reduced to less than 1000 cm/s in this structure, while the value of S b is 106 cm/s when a B-doped p-type layer is epitaxially grown on the c-Si substrate. We have clarified, from internal photoemission (IPE) and attenuated-total-reflection Fourier transform infrared (ATR-FTIR) spectroscopy measurements, that possible reasons for the observed low value of Sb are (1) the band lineup of the heterojunction (HJ) between B-doped p-type a-Si:H and c-Si and (2) the hydrogen passivation of defects in B-doped a-Si:H

A common concept is that crystalline silicon ($c\text{\ensuremath{-}}\text{Si}$) has short-range order (SRO) and long-range order (LRO), while amorphous silicon has only SRO consistent with the continuous random network model. Spectroscopic ellipsometry measurements show that in the initial stage of growth, ultrathin $a\text{\ensuremath{-}}\text{Si}:\mathrm{H}$ films deposited by thermal chemical-vapor deposition at $460\phantom{\rule{0.2em}{0ex}}\ifmmode^\circ\else\textdegree\fi{}\mathrm{C}$ contains deformed crystallites or paracrystallites. Paracrystallites have SRO and medium-range order (MRO). Hence the ultrathin $(4--10\phantom{\rule{0.3em}{0ex}}\text{nm})\phantom{\rule{0.3em}{0ex}}a\text{\ensuremath{-}}\text{Si}:\mathrm{H}$ network possesses MRO with SRO. During the growth, paracrystallites gradually change into a continuous random network.

The subband-gap absorption current in a hydrogenated amorphous silicon film has been measured during plasma-enhanced chemical vapor deposition. The current is probed by a near-infrared laser while photoexcited carriers are generated under visible laser illumination. The trapped charge density is determined from the magnitude of current under the assumption of carrier generation and recombination kinetics. The result indicates that trapped charges are distributed uniformly in the film during growth, and they are reduced after the growth. The trapped charge density is minimized at a growth temperature of ≈ 473 K.

Time evolutions of photocurrents in hydrogenated amorphous silicon films under plasma-enhanced chemical vapor deposition have been measured. The photocurrent is gradually increased with growth time, and the corresponding photoconductivity is improved. The improvement is dependent on the temperature, and more pronounced during postgrowth annealing. The postgrowth annealing plays an important role in improving the transport of photoexcited carriers.

Ultra-thin hydrogenated amorphous silicon thin films have been deposited by thermal chemical vapor deposition (CVD) to prepare smooth top surface of the films avoiding the ion bombardment. Rapid thermal oxidation of thermal CVD a-Si:H results in nanocrystalline dots in the ultra-thin silicon films. Spectroscopic ellipsometry (SE) and high resolution transmission electron microscopy (TEM) have been used to investigate the optical and structural properties of both ultra-thin a-Si:H and nanocrystalline silicon films. To analyze the ellipsometric data of ultra-thin a-Si:H films, a new parameterization i.e., the combination of Sellmeier law and four Lorentz peaks, has been successfully introduced. Width of the Lorentz peaks are directly related with the change of optical functions with the thickness of a-Si:H films. It has been certified that the dense Si matrix with smaller degree of disorder is formed when the thickness exceeds 8nm and the films with the thickness of less than 3.8 nm becomes voided. To interpret the ellipsometric data for nanocrystalline silicon films, three layer model (SiO2, poly-Si+a-Si+void and SiO2) has been adapted. It is inferred from SE and TEM analyses that the size and the density of nanocrystalline dots can be controlled by the morphology of initial ultra-thin a-Si:H films and RTO conditions.

We sometimes observe poor internal quantum efficiency (IQE) in the long-wavelength region of p-type single-crystalline silicon (c-Si) solar cells having back-surface-field (BSF) layers made of p-type hydrogenated amorphous silicon (a-Si:H). We found that this poor IQE is caused by the appearance of a polycrystalline phase (PP) in a-Si:H. We showed that two-step growth of BSF structures with hydrogen plasma treatment (HPT) is effective for achieving a high performance of BSF with improved reproducibility. First, a-Si:H layers were prepared on c-Si substrates at a high growth rate to suppress the appearance of PP, and consequently, the defects at the a-Si:H/c-Si interface induced by the higher growth rate were effectively passivated by HPT. Finally, a-Si:H was deposited as a capping layer on the first a-Si:H layer after HPT. We will discuss the passivation mechanism of interface defects with HPT using the data of the hydrogen profile in the a-Si:H layers treated by hydrogen plasma.

We have found that the back-surface recombination velocity, S b, of minority carriers in crystalline silicon (c-Si) thin film solar cells can be reduced to less than 103 cm/s when a boron (B)-doped p-type hydrogenated amorphous silicon (a-Si:H) layer is deposited on the back surface of a p-type c-Si substrate at 200°C, while the value of S b is 106 cm/s when a B-doped p-type layer is epitaxially grown on the c-Si substrate. We have clarified, from internal photoemission (IPE) and attenuated-total-reflection Fourier transform infrared (ATR-FTIR) spectroscopy measurements, that the band lineup of the heterojunction between B-doped p-type a-Si:H and c-Si and the hydrogen passivation of defects in B-doped a-Si:H are possible reasons for the observed low value of S b.

We have measured for the first time the band lineup at the interface between boron-doped p-type hydrogenated amorphous silicon (a-Si:H) and crystalline silicon (c-Si) by internal photoemission and have found that the band discontinuity mainly exists on conduction-band (CB) side in this type of heterojuctions. This band lineup is related with the low back-surface recombination velocity observed in n+(c-Si) - p(c-Si) - p+(a-Si:H) solar cells.

In this reply, we address the issues raised regarding our article [J. Appl. Phys. 90, 1067 (2001)] and show that spectroscopic ellipsometry is a tool to identify the paracrystallinity of the ultathin hydrogenated amorphous silicon (a-Si:H) films.

It has been proposed and experimentally confirmed that band-engineered hydrogenated amorphous silicon (a-Si:H)/hydrogenated amorphous silicon carbide (a-SiC:H) heterojunctions on c-Si can be applied to electrically programmable and erasable memory devices. A test diode with the structure c-Si/graded a-SiC:H/a-Si:H/uniform a-SiC:H/Al exhibits a large hysteresis in the C-V characteristic with a retention time of 0.8 s at room temperature.

Ultrathin amorphous silicon films prepared by the thermal chemical vapor deposition (CVD) have been investigated using a spectroscopic ellipsometer. The analysis of ultraviolet-visible (UV-VIS) spectroscopic ellipsometric data reveals the morphology of the ultrathin films. To determine the optical functions of such films from ellipsometric data, a new parameterization, i.e., the Sellmeier law with four Lorentz oscillators, has been successfully introduced. A direct correlation has been made between the new parameters and the change of optical functions with the thickness of ultrathin a-Si:H films. By the analysis of ellipsometric data, it has been shown that the formation of dense Si matrices with low structural disorder is possible when the thickness of a-Si:H is more than about 8 nm, and the film with a thickness of less than around 3.5 nm develops voids.

In-situ hydrogenation which is performed in a thermal chemical-vapor-deposition (CVD) chamber by interrupting film growth and by introducing atomic hydrogen is known to be effective for decreasing the defect density in thermal CVD hydrogenated amorphous silicon (a-Si:H). In this study we have experimentally examined the validity of a previous implicit assumption that the film properties after in-situ hydrogenation process could be insensitive to the quality of starting CVD films. This assumption has been proved to be invalid and it has been clarified that one must start with high quality CVD samples in order to obtain high quality samples after in-situ hydrogenation. We explain the present results by assuming that the film structure of the thermal CVD a-Si:H, which affects the hydrogen penetration into the lattice during hydrogen plasma treatment and consequently the defect passivation, can vary with CVD growth conditions.

We find no correlation between a significant variation of the disorder induced band-tail density-of-states distribution and the density of either annealed or light-induced metastable defects in hydrogenated amorphous silicon, which contradicts some currently prevalent notions about the origin of these defects.

It has been shown, by taking account of the variation of experimental conditions of steady-state photocarrier grating (SSPG) measurements with long exposure to light, that light-induced Si dangling bond states control hole diffusion lengths under illumination in plasma-deposited hydrogenated amorphous silicon (a-Si:H) with defect density less than 1016 cm-3. This conclusion is simple but different from previous reports where SSPG data were treated incorrectly. On the other hand, the effects of defect states on diffusion lengths are small in chemical-vapor-deposited a-Si:H treated with hydrogen plasma.

Thin hydrogenated amorphous silicon (a-Si:H) layers with thicknesses of 90-600 Å grown by thermal decomposition of disilane were treated with rf hydrogen plasma just after deposition at the same temperature (430-440°C). During this process (referred to as in situ hydrogenation), atomic hydrogen passivates defects, and the effective thickness of this passivated layer is estimated to be 220 Å. Atomic hydrogen also induces structural relaxation of the Si network even in a-Si:H deposited at high temperature (>400°C) while the change of bonded hydrogen content is rather small in in situ hydrogenation.

Experimental studies on the light-induced changes in plasma-deposited hydrogenated amorphous silicon prepared under visible-light illumination have clarified that (1) the type 2 defect state (ST2), previously found by the present authors (ref. 5), is not a latent site for the light-induced changes in this material; (2) the illumination on the growing surface can modify the surroundings of ST2; and (3) the light-induced increase of the density of neutral Si dangling bonds is small in the illuminated GD sample, especially in the initial stage of long exposure to light.

A constant photocurrent method (CPM) has been adopted to study detailed mechanisms of light-induced changes in undoped hydrogenated amorphous silicon (a-Si:H). Defects other than neutral Si dangling bonds are generated in undoped a-Si:H by illumination for less than 20 hours, and these states control the photoconductivity of samples. Another type of defect state exists in the annealed state of a-Si:H whose density decreases with prolonged illumination, in contrast to a neutral Si dangling bond state. Possible origins of these defect states are discussed.

Experimental studies have been carried out to clarify the origin of the reduction of light-induced changes in plasma-deposited hydrogenated amorphous silicon prepared under visible-light illumination (referred to as illuminated GD a-Si:H in this paper). The changes are smaller in the illuminated GD sample than in the a-Si:H sample deposited in the same run without the illumination even when the content and bonding mode of hydrogen are almost the same in both samples. This implies that the possible reduction of latent sites for the changes by the illumination is an essential origin of the observed reduction.

A small-signal open-circuit voltage decay (OCVD) method has been applied for the first time to measure carrier lifetime in hydrogenated amorphous silicon solar cells under dc bias illumination. The observed decay has multiexponential time dependence with the time constants between 0.7 µs and 20 µs. In the n-i-p type cells investigated in this paper, these time constants correspond to deep trapping hole lifetimes near the n-layer/i-layer interface of the cell. The light-induced changes in the OCVD signal have been observed.

Experimental studies have been made on the electrical properties of hydrogenated amorphous silicon (a‐Si:H) using liquid Schottky barriers. We have found that the quasi‐static capacitance‐voltage (C‐V) method can be applied to the a‐Si:H/quinone‐hydroquinone (Q‐HQ) liquid Schottky junction. This method enables us to determine the net density of positive space charge due to ionized traps and impurities in a‐Si:H (Ne), the built‐in potential (Vb), and the width of the surface space‐charge layer (W), of this liquid junction. The barrier height of an undoped a‐Si:H/Q‐HQ junction has been estimated to be more than 1 eV from the value of Vb thus obtained. By C‐V and surface photovoltage (SPV) measurements on the same samples, we have studied the changes in the properties of a‐Si:H with doping and with prolonged illumination. It has been found that phosphorous (P) doping drastically decreases the hole diffusion length measured by the SPV method and increases the value of Ne. Slight boron (B) doping increases the...

We have found that B (the square of the gradient of (αh/ω)1/2 versus h/ω plot) is dependent on the preparation conditions of a‐Si:H films and has a distinct correlation with photoconductivity. B includes information on the tail states and the band edge of extended states of a‐Si:H. B is proposed as one of important characterization parameters for a‐Si:H as well as photoconductivity.

Light-soaked states in hydrogenated amorphous silicon (a-Si:H) are not recovered by near-infrared or visible light illumination. Simple trapping of photogenerated carriers in the gap states of a-Si:H can, thus, be ruled out as the main process which initiates the light-induced changes. The carrier trapping accompanying the change in the atomic configuration, on the other hand, cannot be excluded as the initial process of the changes. The illumination on the light-soaked states may induce the simultaneous bond reconstruction with detrapping. The efficiency of this process has been estimated to be less than 10-5.

We have made Raman scattering studies on hydrogenated amorphous silicon (a-Si:H) prepared at high deposition rates ranging from 10–105 A/s. The full width of half maximum of the TO-like band, ΔωTO, increases with the increase in the deposition rate, r g, which implies an increase in the structural disorder of a-Si:H with the increase in r g. However, the degree of such increase is small when compared with those observed in a-Si:H alloy materials such as amorphous silicon carbide and amorphous silicon nitride. We have found that the value of ΔωTO is correlated with the B value (the square of the slope of √αh ν versus h ν plot).