Electrical and Chemical Studies on Al2O3 Passivation Activation Process

Available online at www.sciencedirect.com

ScienceDirect Energy Procedia 60 (2014) 85 – 89

E-MRS Spring Meeting 2014 Symposium Y “Advanced materials and characterization techniques for solar cells II”, 26-30 May 2014, Lille, France

Electrical and Chemical studies on Al2O3 passivation activation process M. Pawlika,b*, J. P. Vilcota, M. Halbwaxa, D. Aureauc, A. Etcheberryc, A. Slaouid, T. SchutzKuchlyd, R. Cabale a

Institut d’Electronique, de Microélectronique et de Nanotechnologie (IEMN) UMR 8520, CS 60069, 59652 Villeneuve d’Ascq, France b Ecole Centrale de Lille, Cité scientifique, Villeneuve d'Ascq, Avenue Paul Langevin 59651, France c Institut Lavoisier de Versailles (ILV) UMR 8180, Université de Versailles-St-Quentin en Yvelines, 45 avenue des Etats Unis, 78000 Versailles, France d Laboratoire des sciences de l’Ingénieur, de l’Informatique et de l’Imagerie (ICube), UMR 7357, UdS/CNRS, 23 rue du Loess, BP 20 CR, 67037 Strasbourg Cedex2, France e CEA-INES 50 avenue du Lac Léman, 73375 Le Bourget du Lac, FRANCE

Abstract Al2O3 deposited using Atomic layer Deposition (ALD) technique is known as an excellent material for p-type silicon surface passivation. However a post-deposition annealing step is needed to make the passivation effective. Thanks to coupled electrical measurements (capacitance and photoconductance) and chemical analyses (X-ray Photoelectron Spectroscopy (XPS) and Secondary Ion Mass Spectrometry (SIMS)) carried out on the same p-type Cz silicon sample, a closer explanation of this activation process is given. The presence of hydrogen and oxygen is correlated to the evolution of the electrical parameters and the minority carrier lifetime for 0 to 60 min 450°C annealing.

© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2014 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Peer-review under responsibility of the scientific committee of the SiliconPV 2014 conference. Peer-review under responsibility of The European Materials Research Society (E-MRS) Keywords: Surface passivation; p-type silicon; interface defects ; effective charges ; Al2O3

1. Introduction Today, one major goal of photovoltaic research targets reducing the cost per watt ratio. Concerning crystalline silicon based cells, it consists in reducing raw material consumption by decreasing wafer thickness from 380μm toward 160μm or in using bifacial structure, but a higher impact of surface behavior is observed on cell performances. Surface passivation got then a renewed interest. First, SiO2 as natural passivation layer was tested using thermal, chemical or PECVD (Plasma Enhanced Chemical Vapor Deposition) deposition techniques [1]. Then, as it is used as anti-reflection coating layer (ARC), SiNx was also studied for solar cell passivation. Nevertheless, the best passivation layer depends on the type of silicon it is coated on. For p-type, the best suited film is a thin Al2O3 layer deposited by PE-ALD (Plasma Enhanced-Atomic Layer deposition) [2] which exhibits a very high value of minority carrier effective lifetime. This macroscopic parameter is used to evaluate the quality of a passivation layer. Indeed, once coated and annealed, aluminum oxide provides a high density of negative effective charges in combination with a very low

* Corresponding author. Tel.: +33 320197950; fax: +33320197878. E-mail address: [email protected]

1876-6102 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Peer-review under responsibility of The European Materials Research Society (E-MRS) doi:10.1016/j.egypro.2014.12.347

86

M. Pawlik et al. / Energy Procedia 60 (2014) 85 – 89

density of interface defects [3]. The presence of negative charges at the interface with silicon avoids the so-called parasitic shunting which is highlighted with SiNx [4]. In this study, we first show the potentialities of Al 2O3 passivation on Czochralski grown crystalline silicon (Cz c-Si) since it represents the main raw material used in solar industry. Then, an original approach is proposed, based on coupled electrical (C-V, Photoconductance (PCD)) and chemical (SIMS, XPS) characterizations, to discern the different mechanisms involved in the passivation process. In as-deposited state, aluminum oxide does not provide high effective lifetime value and requires an annealing treatment to be effective [5]. We follow the results obtained for different annealing times to give a closer explanation of what is happening at the interface from the atomic (chemical) level to the macroscopic (electrical) one.

2. Experiments For this study, we used 200μm-thick p-type Cz c-Si 1-5Ωcm material. A first series of samples, called A, is composed of 3" wafers with a mechanical polishing on both faces. The second group of samples, called B, is 50mm square with chemical polishing. Group A is used for electrical and chemical measurements; those require a low surface roughness. Group B is typically used in industrial fabrication of solar cells and is used for carrier lifetime measurements. Al2O3 films are deposited using a PE-ALD process in a Beneq TFS 200 equipment. The deposition temperature is 200°C and the RF power is 50W. Group A is one-side coated with Al2O3 after an HF-Last clean. Two Al2O3 film thicknesses, 7 and 20nm, are deposited. Then, the samples are shared into several pieces; it insures the measurements being done on same material. One piece stays for reference in the “as deposited” state and the other ones are annealed from 5 to 60min at 450°C under N2 atmosphere (activation annealing). Once done, each sample is cut into three parts for the C-V, SIMS and XPS measurements. For the B group, the only difference consists in a double side deposition of a 20 nm-thick Al2O3 film. The carrier lifetime is measured with a Sinton WCT120 equipment using an IR pass filter. 3. Results 3.1. Lifetime measurement A comparison of the minority carrier effective lifetime is given in Fig. 1 for the different samples of Group B. As it is well known, in the as deposited state, Al2O3 does not provide any passivation effect; it needs a 450°C annealing for a significant effective minority carrier lifetime to be observed (Fig. 1a). At 5min, the passivation is fully effective, and a longer annealing time tends to slightly decrease the effective lifetime of carriers (Fig. 1b). As measurements are done on p-type Cz-Si, the results shown in Fig. 1 are obtained after stabilization (12 hours under solar illumination provided by a solar simulator). a

b

100

120

80

100 80 60

As deposited 5min 450°C 30min 450°C 60min 450°C

40 20 0 1E13

1E14

1E15

1E16

1E17

Effective lifetime [µs]

Effective lifetime [µs]

140

60 40 20 15

-3

Effective lifetime @ 10 cm

0 0

10

20

30

40

50

60

-3

Minority carrier density [cm ]

Annealing time [min]

Fig. 1. (a) Effective lifetime versus minority carrier density during different annealing times for p-type Cz c-Si 1-5Ω.cm coated both sides with 20 nm of PEALD deposited Al2O3 film. (b) Evolution of the effective lifetime at the injection level of 1015cm-3

3.2. C-V analysis A C-V measurement conducted on a MOS structure is an efficient way to get information about the insulator-semiconductor interface. Obtaining a MOS structure is done by coating, front- and back-side, with a Ti/Au metallization on Group A samples. Front side metallization is patterned using a physical mask that avoids any additional chemistry linked to photolithography processes. When Al2O3 is coated on silicon, a thin intermediate layer of SiO2 is induced [6]. The different electrical charges that are present in this structure are depicted in Fig. 2. Surface charges depend on the doping and the band bending, which is related to interface charges quantity. These charges arise from the silicon oxide layer induced during deposition process, in our case Al2O3

87

M. Pawlik et al. / Energy Procedia 60 (2014) 85 – 89

coated by ALD technique. Fixed, mobile and oxide trap charges located from the surface to the oxide bulk are also linked to the deposition technique [7]. The total charge is defined as the sum of all charges present. Qs: Surface charge Qit: Interface charge Ec

ψs

Ei EF Ev

Al2O3 SSiO2

Air

Si

Qf: Fixed charge Qot: Oxide trap charge Qm: mobile charge

Fig. 2. Charges distribution through the Si-SiO2-Al2O3 stack

Several algorithms can be used to retrieve the effective charge (Qeff) and the density of interface states (Dit); let us consider the Terman [8] and the High-Low frequency (HF-LF) also called Castagne-Vapaille [9] methods. In the high frequency domain (1MHz) C-V measurements, interface charge cannot follow the applied alternative voltage [10]. So if we consider the particular applied voltage where the band bending is equal to 0 (flatband voltage, Vfb), Qit gives no contribution. Similarly, as surface charge depends on band bending, and in case of uniform doping, Qs can be set to 0. So, measurements obtained from C-V curves at 1MHz reflects a behavior linked to an effective charge, Qeff, which is the sum of Qf, Qot and Qm [11]. Finally, Qeff as well as the interface defect density Dit can be retrieved from modelling, using the well know formula [12]:

) MS 

V fb

Qeff Cox

where )MS is the workfunction difference and Cox the oxide capacitance. The evolution of these two parameters is given in Fig. 3. However, during the experimentation, we failed to get a true low frequency C-V curve for as deposited samples. Since HF-LF method needs low frequency C-V measurements, the Dit point for this method is not available.

-4,0x10

12

-6,0x10

12

-8,0x10

12

-1,0x10

13

-1,2x10

13

-1,4x10

13

10

14

10

13

10

12

10

11

Terman HFLF

-2

-1

12

-2

-2,0x10

Dit [cm .eV ]

b

Qeff [cm ]

a

0

10

20

30

40

50

Annealing Time [min]

60

0

10

20

30

40

50

60

Annealing Time [min]

Fig. 3. (a) Effective charge density (Qeff) versus annealing time and (b) average interface defect density (Dit) between 0.4eV and 0.2eV under the midgap. Lines are guides to the eyes.

It can be noticed that the density of effective charge is already high just after the deposition of Al2O3, an annealing time shorter than 5min increases it a little bit more. For higher annealing times, the charge density decreases and stays almost constant over the 15min to 60min interval. Concerning interface defect density, according to the Terman method, a clear decrease of one order of magnitude can also be observed. Since HF-LF method, is more accurate for low defect densities, longer annealing times just keep constant the interface defect density considering measurement deviations.

88

M. Pawlik et al. / Energy Procedia 60 (2014) 85 – 89

3.3. Chemical behavior

intensity [u.a] Intensity [u.a]

Oxygen Normalized intensity [u.a]

To get a better understanding of the phenomena occurring during the different steps of the passivation, a chemical analysis has been done to link the above reported macroscopic observations to microscopic behaviors at atomic level. As an example, we report hereby SIMS profiles (Fig. 4). Since oxygen improves the ionization efficiency, hydrogen and SiO2 levels are normalized by its value. It can be obviously remarked that the activation annealing has a particular effect on the hydrogen content within the Al2O3 film: hydrogen is less present in bulk Al2O3 and on Al203-air interface and more present at the silicon interface. Also, more Si-O bonds seem to be created by the activation annealing since a slight increase in SiO2 peak is observed. The lower Dit value which is recorded (see Fig. 3) for annealed samples can be then linked to a drastic increase of the chemical passivation. Furthermore, the SiO2 layer thickness is unchanged. As deposited 5min 450°C 30min 450°C 60min 450°C

6,9 4,6

Hydrogen

2,3 1500 1000 500 0

0

50

100

150

200

250

300

50

100

150

200

250

300

50

100

150

200

250

300

Oxygen 0

10500 7000 3500 0

SiO2 0

Sputter time [s]

Fig. 4. Hydrogen, oxygen and SiO2 profiles obtained by SIMS measurement (20nm -thick Al2O3) for different annealing times.

An XPS analysis has been performed on a 7nm-thick Al2O3 layer. Spectra of silicon, oxygen and aluminium are given in Fig. 5. A strong shift of Al, Si and O peaks toward higher energy levels can be observed for the annealed samples tending to show that some Si-O and Al-O-Si bonds have been created during the thermal process. The strengthening of these bonds along the thermal process time can be perceived by comparing the respective position of the as-deposited, 5min and 30 min spectra; 5 min spectrum is always in between the two other ones. 100k

a

90k

O1s

b

1400

Al2p

c

Si2p

1200

80k

12k

70k Count/s

14k

1000

10k

60k

800

8k

50k 40k 30k

6k

600

4k

400

2k

200

As deposited 5min 450°C 30min 450°C 60min 450°C

20k 10k 0 536

534

532

530

528

0 80

78 76 74 72 Binding Energy [eV]

70

0 106

104

102

100

98

Fig. 5. Binding energy of (a) Oxygen, (b) Aluminium and (c) Silicon for different annealing times.

4. Discussion Following the Al2O3 film deposition, a high amount of negative fixed charges can be measured (Fig. 3) owing to C-V measurements and related data post-processing. As an under layer of SiO2 is formed during the ALD deposition process (XPS measurements), these charges are attributed to oxygen interstitials between SiO2 and Al2O3. At this stage, no passivation effect can be detected since a high density of defects (also obtained via from C-V measurement) masks the field effect which is present at the interface. These defects can be correlated to the presence of oxygen interstitial which creates intermediate energy levels in the band gap with boron, and with silicon dangling bonds.

M. Pawlik et al. / Energy Procedia 60 (2014) 85 – 89

Considering the activation annealing, it is clear from SIMS measurements that it favors hydrogen to diffuse towards the interface, through the thin silicon oxide film, and to decrease Si dangling bond density. As a consequence, the Dit drops by 1 order of magnitude. This decrease is also supported by bonds creation between silicon and oxygen. On the other hand, the density of effective charges only concerns the SiO2/Al2O3 interface and bulk aluminum oxide. We assume the fixed charge is the sum of charges due to positive aluminum interstitial and negative oxygen interstitial close to the interface. But XPS analyses show an O/Al ratio drop from 1.85 to 1.32 going from SiO2/Al2O3 interface to bulk aluminum oxide. So the excess of oxygen interstitial explains the negative value of effective charge. Then, for a short annealing time (5 min), aluminum interstitials create bonds first; this decreases the amount of positive charges leading to an increase of effective charges toward negatives values. Its slight decrease with annealing can be related to the general organization of elements within the oxide layers implying that less aluminum and oxygen interstitials are present. The cumulative effect of low Dit and negative Qeff allows then to increase significantly the effective lifetime of minority carrier.

5. Conclusion We show that complementary results can be obtained using electrical and chemical characterizations to explain the passivation process of Al2O3 on p-type Cz Si which is of interest in our case. Among other effects, the passivation anneal fosters the diffusion of hydrogen towards the interface that drastically reduces the interface defect density by one order of magnitude. It is also observed that the negative effective charge density is already present for as deposited samples and is slightly affected by the thermal treatments. Acknowledgement This work is partly supported by the French ANR under the ANR-11-PRGE-0004 "BIFASOL" project, the Ecole Centrale de Lille and the French RENATECH network. Authors want to thanks A. Beaurain and N. Nuns from the Regional Platform of Surface Analysis for valuable discussions about XPS and SIMS experiments.

References [1]

Schmidt J, Moschner JD, Henze J, Dauwe S, Hezel R. Recent progress in the surface passivation of silicon solar cells using silicon nitride. 19th European Photovoltaic Solar Energy Conference, Paris, France; 2004. [2] Repo P, Talvitie H, Li S, Skarp J, Savin H. Silicon Surface Passivation by Al2O3: Effect of ALD Reactants. Energy Procedia 2011;8:681-7. [3] Benick J, Richter A, Li T-TA, Grant NE, McIntosh KR, Ren Y, Hermle M, Glunz, SW. Effect of a post-deposition anneal on Al2O3/Si interface properties. In: 35th IEEE Photovoltaic Specialists Conference (PVSC); 2010. p. 891–6. [4] Dauwe S, Mittelstädt L, Metz A, Hezel R. Experimental evidence of parasitic shunting in silicon nitride rear surface passivated solar cells. Progress in Photovoltaics: Research and Applications 2002;10:271-8. [5] Schmidt J, Veith B, Brendel R. Effective surface passivation of crystalline silicon using ultrathin Al2O3 films and Al2O3/SiNx stacks. Physica Status Solidi Rapid Research Letters; 2009;3(9):287-89. [6] Werner F, Veith B, Zielke D, Kühnemund L, Tegenkamp C, Seibt M, Brendel R, Schmidt J. Electronic and chemical properties of the c-Si/Al2O3 interface. Journal of Applied Physics 2011;109:113701. [7] Schroder DK. Semiconductor material and device characterization. New York: Wiley; 2006. [8] Terman LM. An investigation of surface states at a silicon/silicon oxide interface employing metal-oxide-silicon diodes. Solid-State Electronics 1962;5: 285-99. [9] Castagné R, Vapaille A. Description of the SiO2/Si interface properties by means of very low frequency MOS capacitance measurements. Surface Science 1971;28:157-93. [10] Altındal Ş, Tataroğlu A, Dökme İ. Density of interface states, excess capacitance and series resistance in the metal–insulator–semiconductor (MIS) solar cells. Solar Energy Materials and Solar Cells 2005;85(3):345-58. [11] Nicollian EH, Brews JR. MOS (Metal Oxide Semiconductor) Physics and Technology. New York: Wiley; 1982. [12] Suh D, Liang WS. Electrical properties of atomic layer deposited Al2O3 with anneal temperature for surface passivation. Thin Solid Films 2013;539:30916.

89