Influence of deposition temperature of thermal ALD deposited Al 2 O 3 films on silicon surface passivation

The effect of deposition temperature (Tdep) and subsequent annealing time (tanl) of atomic layer deposited aluminum oxide (Al2O3) films on silicon surface passivation (in terms of surface recombination velocity, SRV) is investigated. The pristine samples (as-deposited) show presence of positive fixed charges, QF. The interface defect density (Dit) decreases with increase in Tdep which further decreases with tanl up to 100s. An effective surface passivation (SRV<8 cm/s) is realized for Tdep ≥ 200 °C. The present investigation suggests that low thermal budget processing provides the same quality of passivation as realized by high thermal budget process (tanl between 10 to 30 min).

The effect of deposition temperature (T dep ) and subsequent annealing time (t anl ) of atomic layer deposited aluminum oxide (Al 2 O3) films on silicon surface passivation (in terms of surface recombination velocity, SRV) is investigated.The pristine samples (as-deposited) show presence of positive fixed charges, Q F .The interface defect density (D it ) decreases with increase in T dep which further decreases with t anl up to 100s.An effective surface passivation (SRV<8 cm/s) is realized for T dep ≥ 200 • C. The present investigation suggests that low thermal budget processing provides the same quality of passivation as realized by high thermal budget process (t anl between 10 to 30 min).C 2015 Author(s).All article content, except where otherwise noted, is licensed under a Creative Commons Attribution 3.0 Unported License. [http://dx.doi.org/10.1063/1.4922267]

I. INTRODUCTION
Silicon (Si) is the most widely used substrate for solar cell applications due to its well established technology. 1,2The solar cell efficiency is limited due to several losses such as lattice thermalisation loss, sub-band gap loss, looses associated with electrical contacts, recombination loss etc.The recombination losses occur both in the bulk region as well as at the semiconductor surface (due to dangling bonds). 3To make the solar cell economically viable, one way is to reduce the Si wafer thickness (in terms of gm/W usages and simultaneously increase in device efficiency).In thinner wafers, both surfaces are close to area of carrier generation and their collection and therefore, are prone to enhanced recombination losses which have detrimental effect on device performance.The losses at the semiconductor interfaces or surfaces can be suppressed by surface passivation for which two strategies exist and are well appraised in literature.][6][7][8][9][10] The interface defect density can be reduced by chemical route where atomic hydrogen or a halogen atom is, generally, attached with the unsaturated Si (dangling) bonds.Consequently surface is passivated 11 and this process is referred as chemical passivation.On the other hand, fixed charges (both in bulk and at interface) present in a dielectric layer repel the charges of same polarity at the Si interface deep into the bulk Si and consequently the recombination of the photo-generated carriers reduces.This is called field-effect passivation.An excellent surface passivation is a prerequisite for realization of next-generation high efficiency silicon solar cells.Minority carrier lifetime is the measure of surface passivation effectiveness and is generally quantified in terms of surface recombination velocity (SRV) which can be deduced from the measured minority carrier lifetime (τ eff ).
Thermally grown silicon di-oxide (SiO 2 ) is most widely used for silicon surface passivation which requires high temperature processing.In this situation, bulk lifetime degradation associated with high thermal budget cannot be ruled out. 12Besides SiO 2 , other dielectric layers such Si x N y :H, SiC and a-Si:H are also used for surface passivation. 13[7]9,24 ALD being a conformal deposition process provides high-quality thin films with precise control over the thickness. 9,255][6][7] However, these studies are scattered and are made on different type and quality materials with high thermal budget.However, literature in low thermal budget annealing process using RTP with less annealing time is very few. 9,26n this paper, a systematic study is made wherein Al 2 O3 films are deposited using thermal ALD process at different deposition temperature.The influence of T dep on silicon surface passivation properties is investigated.Low thermal budget rapid thermal process (RTP) is used to optimize the post deposition annealing time at a fixed annealing temperature (= 400 • C) and the minority carrier lifetime is measured to see the role of the process parameters on surface passivation.Capacitance-voltage (C-V) and X-ray photoelectron spectroscopy (XPS) measurements are performed to understand the mechanism of surface passivation.

II. EXPERIMENTAL DETAILS
Al 2 O 3 films are grown on float zone (FZ) chemically mechanically polished (CMP) <100> oriented n-type single crystalline silicon of 5 ± 0.5 Ωcm resistivity having 325 ± 10 µm thickness.To ensure identical surface, electronic and optical properties; the samples are cut from the 100 mm diameter silicon wafer using laser scriber (YAG-50 series, M/s Argus, China).Each sample is a quarter of the larger wafer.The samples are cleaned in piranha solution (H 2 SO 4 :H 2 O2 :: 4:1 by volume) for 15 min followed by dip in 5% HF solution for 2 min.
The films are deposited in thermal ALD reactor (Model: R200, M/s Picosun, Finland) using trimethylaluminum [TMA, Al(CH 3 ) 3 ] (M/s SAFC, Hitech, UK) and H 2 O as precursors for aluminum and oxygen sources respectively.Each deposition cycle consists of two half cycles; one TMA pulse and one H 2 O pulse and each step is separated by a nitrogen purge to remove un-reacted excessive precursors.The desired layer thickness can be obtained simply by repeating the number of ALD cycles.The surface chemistry during the first and the second half cycles of thermal ALD process can be described by the following equations respectively. 27st half cycle: 2 nd half cycle: Al 2 O 3 films are deposited on one side (for C-V) as well as on both the sides (for lifetime) of the wafer for a fixed number of precursors cycles (= 300) with the deposition temperature variation (= 100 • C, 200 • C and 300 • C; and the samples are denoted as S 100 , S 200 and S 300 respectively).The films are annealed at 400 • C for different annealing time (t anl = 90 s, 105 s and 120 s) in nitrogen (N 2 ) ambient using rapid thermal process (RTP Model: AS-one 150, M/s Annealsys, France).
The thickness (d Al2O3 ) and refractive index (n Al2O3 ) of the films are measured by spectroscopic ellipsometer (Model: M2000, M/s J.A. Wollam Co. Inc., USA).Minority carrier lifetime is measured by two different techniques; Sinton's lifetime tester (Model: WCT-120, M/s Sinton Inc., USA) and microwave photoconductance decay technique (µ-PCD, Model: WT 2000 PV, M/s Semilab Zrt, Hungary).In the first technique, injection level dependence (∆n) of τ eff in a large range (10 13 < ∆n < 10 16 cm −3 ) is measured that gives an average value of the parameter whereas µ-PCD operates at a fixed injection level (∼ 1.2 × 10 13 cm −3 ).However, the advantage of µ-PCD is that it gives the τ eff map, from which the uniformity of lifetime across the sample surface could be seen.Both the systems give almost comparable τ eff provided the injection levels are matched. 10Fourier transform infrared spectroscopy (FTIR, Model 2000, M/s Perkin Elmer spectrometer, USA) is used to know the different bonding configurations in the film.The spectra are recorded in absorption mode.C-V measurements are carried out to find out the interface properties of Al 2 O 3 /Si and the data is used to evaluate flat-band voltage (V FB ), fixed charge density (Q F ) and interface defect density (D it ).For C-V measurements, the metal-insulator-semiconductor (MIS) structure is made by depositing aluminum dots (area ∼ 0.02cm 2 ) through a shadow mask on the front side and fully covered back side using an e-beam evaporation system (M/s Hind Hivac, India) and the measurements are done with an impedance/gain phase analyzer (Model: 1260, M/s Solartron, UK) at 1 MHz.XPS measurements (Model: 0520, M/s Omicron nanotechnology) are carried out to do surface analysis and to get information about the bonding that may provide vital information about the surface passivation mechanism.The system is operated at a base pressure of 10 −11 mbar, where spectra are recorded using Mg-K α as an X-ray source.

Film thickness and Refractive index
The film thickness and refractive index of as-deposited samples S 100 , S 200 and S 300 are given in Table I.To check the thickness homogeneity, d Al2O3 are measured at different locations on the samples.The variation in d Al2O3 across the sample (from edge to center) is less than ±1%.This shows that the films are highly uniform.Table I show the d Al2O3 and n Al2O3 of as-deposited samples deposited at different T dep for fixed 300 cycles along with the estimated errors.It can be seen that d Al2O3 increases with T dep .This is due to the dependence of the reaction kinetics on deposition temperature.At low T dep , growth rate is slow due to slower reaction kinetics primarily associated with thermal activation barrier. 28,29The refractive index also increases from 1.50 to 1.62 with increase in the deposition temperature.Increase in n Al2O3 may be due to the dense and compact films at higher temperature.The refractive index of S 100 is low in comparison to the films S 200 and S 300 .It is reported that at lower deposition temperature, the hydroxyl surface coverage on Al 2 O 3 surface is high which may results in the lower density of the film. 27The low refractive index of a film as compared to its bulk counterpart may be either due to small film thickness or low mass density or the combination of the two.At lower temperature, the decrease in n Al2O3 may be attributed to decreased density and to increased impurity levels in films. 30It is to be noted that in thinner Al 2 O 3 films (∼ 7 nm), deposited by ALD at 300 • C, n Al2O3 ∼ 1.5 is observed. 9This indicates that the lower n Al2O3 of S 100 vis-à-vis S 200 or S 300 is primarily due to lower mass density.

Minority carrier lifetime
As mentioned earlier that the minority carrier lifetime is the key parameter to know the quality of passivation.The measured effective carrier lifetime (τ eff ) value has contributions from the bulk (τ bulk ) and the two wafer surfaces (τ s ) of the silicon sample and these two parameters are related with τ eff as 31,32 1   the wafer, respectively.S eff is surface recombination velocity (SRV).Amongst the various modes associated with Eq. ( 3), the fundamental mode (m= 1) is the most significant and the contribution of the higher modes is significantly small and could not be detected experimentally.In this case the above equation reduces to where τ s = W/2S eff .The factor 2 accounts for the contribution of both surfaces.For very large bulk lifetime, τ eff τ s and in this case the upper limit of SRV (S eff max ) can be estimated using the following relation: For τ eff measurements, symmetrically passivated samples are prepared by depositing Al 2 O 3 films of the same thickness on both sides of the substrates.In Table II, measured τ eff values and corresponding S eff,max of as-deposited and annealed samples of S 100 , S 200 and S 300 at ∆n = 1 × 10 15 cm −3 are given.In case of annealed samples, annealing time is varied (t anl = 90 s, 105 s and 120 s) whereas the annealing temperature is fixed at 400 • C. It is observed that the τ eff values of as-deposited samples of S 100 and S 200 is almost the same (584 µs and 588 µs respectively) within the measurement uncertainty but in S 300 , τ eff is increased to 824 µs with increase in T dep .A systematic decrease in τ eff values with the rise in t anl is observed in S 100 .This decrease in τ eff may be due to the lower density (T dep = 100 • C) structure of the as-deposited film that may not be an effective passivation layer and therefore, results in low value of τ eff after annealing.This observation can be corroborated with the C-V data of the annealed film discussed later.On the other hand, τ eff increases with t anl from 90 s to 105 s in S 200 and S 300 and with further increase in t anl (∼120 s), τ eff decreases from its maximum.It can be seen from Table II that maximum τ eff (∼2274 µs) is observed in S 200 with t anl = 105 s.This corresponds to SRV ∼7 cm/s which is close to S 300 sample within measurement error.4][35][36][37][38] Figure 1 shows the SRV (at ∆n = 1 × 10 15 cm −3 ) as a function of t anl of as-deposited and annealed (t anl = 90 s, 105 s, 120 s) samples of S 100 , S 200 and S 300 .5][6][7] Though, similar annealing time is required to achieve SRV ∼7 cm/s for S 200 and S 300 but the thermal effective budget of S 300 is more than S 200 due to higher T dep (for the fixed t anl and T anl ).
Figure 2 shows the injection level dependent τ eff for as-deposited and annealed samples of S 200 for different t anl .The minority carrier lifetime results illustrate quality surface passivation in the entire range of ∆n. Figure 3 shows the τ eff map of the as-deposited and annealed samples (S 200 ) where the four samples (quarter pieces) are put together for the mapping using µ-PCD in order to provide an idea about the uniformity of passivation in the samples.for both AlO 4 and AlO 6 . 39The absorption peak at 513 cm −1 is attributed to Al-O stretching mode of condensed AlO 6 . 40Absorption peaks at 653 cm −1 and 700 cm −1 are also dominated by O-Al-O bending and Al-O stretching for AlO 6 , respectively. 39Katamreddy et al. 41 have reported that the absence of the peak at 530 cm −1 due to Al-O stretching in condensed AlO 6 octahedra indicates the amorphous structure of the film.Similar observations are made in our samples also.A small hump observed in the range 1075-1150 cm −1 may be attributed to Si-O-Si symmetric stretching mode. 40,41This indicates towards the formation of ultra thin layer of SiO 2 at the Al 2 O 3 /Si interface.The appearance of absorption peak at ∼ 1395 cm −1 owe their existence to Al=O stretch bands. 42The absorption bands seen in the range 2340-2360 cm −1 are due to carbonate species in the films.The band at ∼ 2360 cm −1 is assigned to CO 2 , bonded to strong Lewis sites of alumina. 43igure 4(b) shows the FTIR spectra of as-deposited and annealed samples of S 200 at different t anl in the range of 400-2500 cm −1 .The absorption peak at 513 cm −1 in the as-deposited film broadens after annealing at different temperatures (t anl = 90 s, 105 s and 120 s).The peak observed at 675 cm −1 in as-deposited sample is shifted to 710 cm −1 in case of annealed samples which is attributed to Al-O stretching for AlO 6 .The annealed films also show absorption peak at 1395 cm −1 .In the annealed films, the peaks at 2340 cm −1 and 2360 cm −1 is less intense as compared to as-deposited films.Similar results are also obtained for S 100 and S 300 .

Capacitance-Voltage (C-V) measurements
Capacitance-Voltage (C-V) measurement is generally used to assess the dielectric property of a material.C-V measurements are performed on MIS structure at 1 MHz frequency.The applied bias voltage (V G ) is varied from negative to positive which shifts the C-V curve from inversion to accumulation region (forward sweep).These measurements are performed to extract the fixed charge polarity and density (Q F ), which is responsible for field effect passivation.Figure 5(a) shows the normalized C-V curves for as-deposited samples.The C-V curves for the as-deposited samples shift towards negative voltages with decrease in deposition temperature.The magnitude of shifting increases with decrease in deposition temperature which suggests that fixed charges also increases with reducing T dep .The fixed charge density is calculated using eq.( 5) where ϕ ms is the work function difference between metal and semiconductor, V FB is flat band voltage which is extracted from C-V curve and C OX is oxide capacitance of the film.The Q F of as-deposited films is positive which are listed in Table III.Rafi et al. 36 also observed positive fixed charges in the Al 2 O 3 films deposited at 100 • C. Figure 5(b) shows the normalized C-V curves for as-deposited and annealed samples (at t anl = 90 s, 105 s, 120 s) of S 200 .A shift towards less negative voltages in C-V curves with respect to its as-deposited curve is observed with annealing.The calculated values of Q F are negative (Table III) which confirms that there is an activation of negative fixed charges in these films.The density and polarity of fixed charges are related to the deposition process of oxide films and is associated with local non-stoichiometry or structural defects. 4,9,24,44It can be seen that with increase in t anl from 90 s to 105 s, value of Q F decreases and thereafter for t anl = 120 s, the values start increasing.Inset of Fig. 5(b) shows a representative C-V curve of annealed (t anl = 90 s) film of S 100 .The drop in capacitance value in accumulation region and stretching of the curve indicates towards degradation in the film quality after annealing.This observation is rather surprising and needs further probing.The interface defect density (D it ) at Al 2 O 3 /Si interface can be extracted from the conductance measurements (C-f) and the calculated values (for all as-deposited and annealed samples of S 200 ) are listed in Table III.It is observed that the value of D it decreases with the increase in T dep .The lower value of D it for as-deposited sample S 300 is primarily responsible for the high value of τ eff .It is well established that in thermal ALD, chemical passivation is dominant which is quantized in terms of D it . 45Further, the D it value decreases with annealing from 4.2 × 10 12 to 1.3 × 10 12 eV −1 cm −2 for t anl from 90 s to 105 s for S 200 .Henceforth for an increase in annealing duration to 120 s, D it increases.The lowest D it value is observed for sample S 200 annealed at t anl = 105 s.It is known that the surface passivation is governed primarily by both Q F and D it .The effective surface passivation requires high Q F and low D it values.As the two parameters have opposite trend, a trade-off between the two decodes the lowest SRV.In the present case, the maximum τ eff or minimum SRV are realized at t anl = 105 s (S 200 ).C-V measurements reveal that the Al 2 O 3 /Si  yields a high level of chemical passivation after annealing that can be attributed to moderate D it (= 1.3 × 10 12 states eV −1 cm −2 ).

X-ray photoelectron Spectroscopy (XPS)
XPS being a surface sensitive technique is used to get both qualitative and quantitative information of the stoichiometry and chemical states at the surface.The results of representative asdeposited and annealed (t anl = 105 s, T anl = 400 • C) samples of S 200 is given here.The survey spectra depict mainly Al and O contributions except carbon contamination (spectra not shown).It is attuned to the C 1s peak position to 285 eV for correction.Figures 6(a) and 6(b) show the Al 2p and O 1s core level spectra respectively of as-deposited and annealed films.These core level spectra are de-convoluted into their Gaussian components to extract the bonding information.In Fig. 6(a), peaks at 74.47 eV and 75.55 eV confirm the presence of Al-O and Al-OH bonds, respectively.The absence of peak at 73 eV which otherwise represents the Al-Al bond suggests that the entire Al is oxidized by the surface-saturation reaction. 46n Fig. 6(b), peaks at 531.40 eV and 532.68 eV correspond to Al-O and OH/COO bonds respectively.The existence of OH/COO bonds in thermal Al 2 O 3 films are due to the usages of H 2 O as oxidant. 47The elemental ratio of oxygen to aluminum is determined from the peak areas of the Al-O contributions within O 1s and Al 2p core levels.In this case, the contributions from COO and OH groups are not considered.For thicker films, a sensitivity factor is required to calibrate the O/Al ratio from XPS. 13 The resulting O1s Al-O /Al2p Al-O ratio for the as-deposited and annealed film (t anl =105 s) are 1.15 and 1.28 respectively.This shows that the annealed film becomes more oxygen rich than the as-deposited state.The deconvoluted peak areas of Al-O in O 1s and Al 2p spectra of as-deposited and annealed (t anl = 105 s) film of S 200 are given in Table IV.These values are used to calculate the O1s Al-O /Al2p Al-O ratio.Inset of Fig. 6(b) shows the C 1s core level spectra of as-deposited and annealed (t anl = 105 s) sample of S 200 .This spectra show that the intensity of carbon is less for annealed sample w.r.t.its as-deposited state.This observation is also supported by the decrease in FTIR peak intensity at 2340 cm −1 observed in annealed samples w.r.t.it's as-deposited state (S 200 ; Fig. 4(b)).The O-H bonds which are present in neutral charge state 48 decreases in annealed film in O 1s spectra which signifies that there is an increase in oxygen dangling bonds (O db ) in annealed state than as-deposited state.Al-OH bond contribution in Al 2p spectra decreases with the annealing which is also support this observation.The increase in O db may suppress the positive fixed charges and activate negative fixed charges in the annealed film.This observation is also supported by the presence of negative fixed charges found in the annealed films estimated using C-V data.

IV. CONCLUSIONS
Silicon surface passivation is studied by thermal ALD deposited Al 2 O 3 films deposited at different temperature and annealed for varying time duration.S 100 film show low refractive index value than S 200 and S 300 .For as-deposited films, the lifetime value is comparable for S 100 and S 200 whereas τ eff increases for S 300 .The τ eff value of S 100 decreases with annealing whereas τ eff increases for S 200 and S 300 .The lowest SRV value is realized for films S 200 and S 300 which are annealed at 400 • C for 105 s using RTP.However, the total thermal budget of S 300 is more as compared to S 200 for the same annealing time and temperature due to higher deposition temperature.The best optimized low thermal budget conditions in terms of deposition temperature & annealing conditions are obtained for S 200 .C-V measurements show lowest Dit under these annealing conditions.C-V measurement of annealed films of the sample (S 200 ) deposited at 200 • C show that there is an activation of negative fixed charges in these films.XPS study shows that there is a decrease in OH bonds in the O 1s spectra after annealing.This suggests an increase in the oxygen dangling bonds which is responsible for negative charge activation.

1
Academy of Scientific & Innovative Research (AcSIR), CSIR-National Physical Laboratory (CSIR-NPL) Campus, New Delhi-110012, India 2 Silicon Solar Cell Group (Network of Institutes for Solar Energy) CSIR-NPL, New Delhi-110012, India (Received 3 February 2015; accepted 24 May 2015; published online 5 June 2015) where, τ s = W 2 /(Dχ m ).χ m are the roots of the transcendental equation χ m tan(χ m ) = S eff W/D where the subscript m represents the m th root and; D and W are the diffusion constant and thickness of

Figure 4 (
Figure 4(a) shows the FTIR spectra of as-deposited samples (S 100 , S 200 and S 300 ) in absorption mode.All the absorption bands are marked with arrows.The broad absorption band in the range of 400-750 cm −1 is attributed to the mixed contributions of Al-O bending and stretching modes

FIG. 6 .
FIG. 6.(a) Deconvoluted Al 2p core level spectra of as-deposited and annealed sample of S 200 .(t anl = 105 s, T anl = 400 • C).The peak decomposition into Al-O (blue) and Al-OH (red) are indicated.The measured data is shown by black (hollow) circles and the resulting fitted curve is shown by black line.(b) Deconvoluted O 1s core level spectra of as-deposited and annealed sample of S 200 .(t anl = 105 s, T anl = 400 • C).The peak decomposition into Al-O (blue) and OH/COO (red) are indicated.The measured data and the resulting fitted curve are shown by black (hollow) circles and by black line respectively.Inset shows the C 1s core level spectra of as-deposited and annealed (t anl =105 s, T anl = 400 • C) sample of S 200 .

TABLE I .
Thickness (d Al2O3 ) and refractive index (n Al2O3 ) of as deposited samples: S 100 , S 200 and S 300 .These values have ∼1% measurement uncertainty.

TABLE IV .
Deconvoluted peak area of Al-O in O 1s and Al 2p spectra of as-deposited and annealed (t anl = 105 s, T anl = 400 • C) sample of S 200 .