Hydrogen is identified as a useful technique to passivate defects within crystalline silicon. However, the effect of hydrogen passivation for a silicon surface is normally characterized as a reduction in surface recombination velocity (SRV), which is not enough to reflect the detailed changes of electronic properties, such as defect density, defect energy levels, and capture cross section, of silicon surface states. In this paper, we utilized the transient capacitance measurement to characterize the detailed electronic properties of silicon surface states before and after hydrogenation. The differences, in terms of the effects of hydrogenation on silicon surface states, either in copper contaminated conditions or clean conditions, are presented and discussed.

Crystalline silicon still dominates the current photovoltaic market, of which the further development requires a comprehensive understanding of its electronic properties. Silicon surface passivation is one of main factors in influencing the electronic properties of crystalline silicon and hence the efficiency of silicon solar cells.1–5 So far, it has already developed a theory on characterizing silicon surface states using the parameters like surface recombination velocity, capture cross sections, surface defects’ energy levels, surface defects’ density and et al.6–9 SRV is a general parameter to describe the recombination rate of silicon surface,10 however, it is too general to give detailed information on defects’ type and the corresponding electronic properties of silicon surface states. In previous work, we have developed a technique11 using the transient capacitance measurement to characterize capture cross sections, defects’ energy levels and defects’ density of silicon surface states, with which we can look insight at the electronic properties of silicon surface states.

Hydrogen has been extensively confirmed to have the ability to passivate dangling bonds, metallic impurities, crystallographic defects and et al.12–18 DB Fenner et al.2 reported that hydrogen is able to terminate silicon dangling bonds, leading to reduced surface recombination velocity. NI Martı et al.19 found that hydrogen plasma treatment below 300°C for silicon wafers can improve silicon surface passivation resulting in higher effective minority carrier lifetime and cell efficiency. TTA Li et al.20 also claimed that hydrogen is capable of further enhancing the passivation ability of aluminum oxide on silicon surface via terminating silicon dangling bonds. Although numerous studies have been done on proving that hydrogen can passivate silicon surface states, few research have investigated the impact of hydrogen on the detailed properties of silicon surface states. Furthermore, copper decoration is able to dramatically change the electronic properties of silicon surface states like the transformation of localized defect energy levels to band-like defect energy levels.21,22 In addition, it is known that copper, decorated on the silicon surface with precipitates, causes a double side Schottky junction, whereas the copper precipitates/Si matrix interface induces defects’ electronic states in the silicon band gap.23,24 Therefore how hydrogen responds to the silicon surface states that contaminated with copper still needs more research.

In this paper, we used deep level transient spectroscopy (DLTS) equipment to characterize the transient capacitance of silicon surface states before and after hydrogen passivation on a metal-oxide-semiconductor (MOS) structure. The response of silicon surface states in either clean conditions or copper contaminated conditions to hydrogenation is investigated. A comprehensive understanding is proposed for explaining the mechanisms associated with the changes of the electronic properties of silicon surface states responding to hydrogenation.

Five 6 inch p-type 1Ω cm Cz wafers were firstly laser cleaved into 2 inch tokens for the subsequent experiments. Eighteen 2 inch samples were then cleaned with RCA solution and subsequently were oxidized at 1180 K temperature for 18 minutes to grow a 90 nm oxide film. Then we evaporated the aluminum contact onto the oxide film via the mask and the area is 1 mm2. Subsequently, eighteen samples were divided into two groups. One group kept clean and another group was contaminated with copper by cupric sulfate solution following a 300 °C anneal. DLTS tool was used to characterize the transient capacitance of the samples in these two groups at various temperatures. The DLTS equipment was a DL8000 Bio-Rad Fourier transform deep-level transient spectroscopy system (frequency=1MHz). The DLTS characterization was conducted at the temperature range of 80-300 K and the biased voltage at 2 V. After measurement, a hydrogen plasma treatment, with pressure at 120 mTorr, rf power at 0.28 W/cm2, flow rate at 12 sccm and a temperature of 300 °C for 20 min, was conducted on all the samples. After that, DLTS was used again to characterize the transient capacitance of the samples at various temperatures. The measured data was subsequently put into a model, presented below, to calculate the defects’ density, defects’ energy levels and capture cross sections of silicon surface states for all the samples.

The exact mechanisms of the transient capacitance measurement on characterizing silicon surface states is presented below. We first analyze the energy band diagrams of metal-oxide-semiconductor configuration under the different biased voltages. Figure 1 shows the energy band diagram of the aluminum-oxide-silicon configuration when the samples were applied with the biased voltages at VR and Vp, respectively.

FIG. 1.

The energy band diagrams of the MOS (aluminum-oxide-silicon) structure when the applied voltages were (a) Vr=2 V and (b) Vp=0 V, respectively. This figure is similar with the figure in Ref. 11 with the same authors. It is reproduced with permission from Song et al., Applied Physics Letters 111(15), 152103 (2017). Copyright 2017 AIP Publishing LLC.

FIG. 1.

The energy band diagrams of the MOS (aluminum-oxide-silicon) structure when the applied voltages were (a) Vr=2 V and (b) Vp=0 V, respectively. This figure is similar with the figure in Ref. 11 with the same authors. It is reproduced with permission from Song et al., Applied Physics Letters 111(15), 152103 (2017). Copyright 2017 AIP Publishing LLC.

Close modal

To calculate the transient surface charge densities, the defects’ densities and the defects’ energy levels of silicon surface states, we built the model as follows to simulate the physical processes during the transient capacitance measurement.

V1+V2=Vr
(1)
V2=qNadR22εsi
(2)
V1=q(Nfix+NadRNGB)d1εSiO2
(3)
C=AεSiO2εsid1εsi+dRεSiO2
(4)

Where V1 is the voltage drop across the oxide film, V2 is the voltage drop across the silicon, V is the applied voltage, q is the elemental charge, Na is the doping density of silicon, dR is the depletion region width, εsi is the permittivity of silicon, εSiO2 is the permittivity of the silicon oxide, Nfix is the number of fixed charges within the oxide film, NGB is the interface charge, d1 is the thickness of the oxide film, A is the area of the MOS structure, C is the measured transient capacitance.

From the equations (1)–(4), we can calculate the transient silicon surface charge as a function of the transient capacitance as illustrated in equation (5).

NGB=Nfix+Na(AεSiO2εSid1εSiC)CεSiO2+qNa(AεSiO2εSid1εSiCCεSiO2)2εSiO22εSiO2εSiVr2εSiqd1
(5)

We can further extract the defects’ energy levels and defects’ density from the equations (6)–(8).

dNGBdt=NTep=NTγσpT2exp(ETkT)
(6)
NGB=NGB0+NT(EF0ET)
(7)
dNGBdt=NTep=NTγσpT2exp(NGB0NT+EF0kT)exp(NGBNTkT)
(8)

Where NT is the defects’ density at each energy level, ep is the hole emission rate,γis a constant, σp is the hole capture cross sections, T is the absolute temperature in kelvin, ET is the defects’ energy level, k is the Boltzmann constant, NGB0 is the surface charge at zero bias, EF0 is the Fermi level at zero bias.

In conjunction with the measured transient capacitance data, we can calculate the transient signals of silicon surface charges as a function of time for all the groups of samples, as shown in Figure 2.

FIG. 2.

The transient signals of silicon surface charges as a function of time for both groups of samples at (a) clean or (b) copper contaminated conditions before and after hydrogen passivation.

FIG. 2.

The transient signals of silicon surface charges as a function of time for both groups of samples at (a) clean or (b) copper contaminated conditions before and after hydrogen passivation.

Close modal

In Figure 2(a), when the silicon surface states were clean, the surface charges, both before and after hydrogenation, decreased with the time. It was due to that the accumulated holes, filled at the pulsed voltage, emitted back to the valence band with the time. It can also be found that the surface charges after hydrogenation were only about one eighth of that before hydrogenation, which can be accounted for that most of surface defect states were passivated by hydrogen leading to a lower defects’ density and hence lower surface charges. However, when the silicon surface was contaminated with copper, the situation was a bit different as shown in Figure 2(b). In Figure 2(b), it exhibits a same trend that silicon surface charges, both before and after hydrogenation, decreased with the time. This was attributed to that the accumulated holes, filled at the pulsed voltage, needed to emit back to the valence band. Nevertheless, the surface charges after hydrogenation was about one third of that before hydrogenation, which was much smaller than the ratio mentioned above. The comparison of two ratios suggest that clean silicon surface states were apt to hydrogen passivation compared to copper contaminated silicon surface states. Please note that sulfate ion is not a common contaminant in silicon and will not affect the electronic properties of silicon.

Since the function of silicon surface charges with the time has been deduced, we can further calculate the hole emission rate at the different conditions, which is illustrated in Figure 3.

FIG. 3.

The transient surface charge decay rate as a function of surface charges when the silicon/silicon oxide interface was (a) clean before hydrogenation, (b) contaminated with copper before hydrogenation, (c) clean after hydrogenation, and (d) contaminated with copper after hydrogenation. Note that Tp is the voltage filling pulse time.

FIG. 3.

The transient surface charge decay rate as a function of surface charges when the silicon/silicon oxide interface was (a) clean before hydrogenation, (b) contaminated with copper before hydrogenation, (c) clean after hydrogenation, and (d) contaminated with copper after hydrogenation. Note that Tp is the voltage filling pulse time.

Close modal

In Figure 3, we can find that the characteristics of the hole emission process were different for the clean and copper contaminated samples via comparing Figure 3(a) & (b) with Figure 3(c) & (d). It can be found that the hole emission rates increased with the tp value for the clean surface, whereas the hole emission rates coincides together at the initial stage for the different tp values when the silicon surface was contaminated with copper. The difference is due to that when the silicon surface was clean, the defects’ energy levels were discrete so that more filling pulse time (higher tp value) would lead to the higher occupation ratio of the shallow energy levels and hence resulted in faster hole emission rate. However, when the silicon surface was contaminated with copper, the defects’ energy levels were band-like and therefore the holes, emitted from the top of the defect band, exhibited an unique hole emission rate. It can also be found that hydrogen can reduce the hole emission rates of both samples in clean and copper contaminated conditions due to the decreased defects’ densities. Nevertheless, hydrogen is not able to change the electronic properties of defects’ energy levels, in other words, the clean silicon surface still had discrete energy levels, while the copper contaminated silicon surface had band-like energy levels. Copper contamination was able to precipitate at the silicon surface and formed a defect cluster with the surrounding silicon, of which the close defects would induce band-like electronic states in the silicon band gap. Hydrogen was able to passivate defect states, however, the radius of hydrogen atom was too small to passivate the most of defects’ states in the defect cluster, and thus the rest defect states in the defect cluster were still close enough to exchange electrons/holes leading to a band-like energy band. For clean silicon surface, the defect states were far away and therefore the defect energy levels were discrete. Hydrogen was able to passivate the most of defect states and the rest defects in clean silicon surface were still far away resulting in the discrete energy levels.

Then we can further extract the defects’ densities and defects’ energy levels according to the equations (6)–(8) for the clean and copper contaminated samples, as illustrated in Figure 4.

FIG. 4.

The extracted defects’ densities and defects’ energy levels of the silicon surface states when the silicon surfaces were (a) clean; (b) contaminated with copper before and after hydrogenation.

FIG. 4.

The extracted defects’ densities and defects’ energy levels of the silicon surface states when the silicon surfaces were (a) clean; (b) contaminated with copper before and after hydrogenation.

Close modal

In Figure 4, the defects’ densities were significantly reduced at a factor of ∼8 after hydrogenation when the silicon surface was clean, whereas the defects’ densities were only decreased at a factor of ∼2.5 after hydrogenation when the silicon surface was contaminated with copper. The difference is due to that it was easy for hydrogen to bond with the separated dangling bonds in clean silicon surface, nevertheless, when the silicon surface was contaminated with copper, it formed a defect cluster so that defects were close to each other and hydrogen has a certain radius so that it cannot fully passivate very close defects, which results in a lower passivation efficacy.

The capture cross sections can also be calculated as shown in Figure 5.

FIG. 5.

The capture cross sections of the silicon surface states when the silicon surface is clean or contaminated with copper before and after hydrogenation.

FIG. 5.

The capture cross sections of the silicon surface states when the silicon surface is clean or contaminated with copper before and after hydrogenation.

Close modal

In Figure 5, it can be found that the capture cross sections of the clean silicon surface states were reduced to ∼10-18 after hydrogenation, which was only one tenth of the capture cross sections before hydrogenation. Nevertheless, the capture cross sections of the silicon surface states contaminated with copper only decreased a bit after hydrogenation, which suggested a lower passivation efficacy. The reason was explained above that hydrogen cannot well passivate much closely distributed defect states in the defect cluster formed by the copper precipitate/silicon matrix.

In this paper, we find that hydrogen is able to passivate silicon surface states when the silicon surface was both clean and contaminated with copper. Nevertheless, the passivation efficacy of defects in clean silicon surface was much higher than that in copper contaminated silicon surface, which is attributed to that hydrogen can well passivate the separated defects, but was not able to well passivate much closely distributed defect states in the defect cluster. It is also found that hydrogen was not able to change the electronic properties of the defects’ energy levels, for which clean silicon surface still had discrete energy levels and copper contaminated silicon surface had band-like energy levels. Furthermore, the defects’ densities and capture cross sections were calculated before and after hydrogenation for both samples with clean and copper contaminated silicon surfaces.

The authors would like to thank National Natural Science Foundation of China (Nos. 51602085 and 51532007) for financial support.

1.
B.
Hoex
 et al, “
Silicon surface passivation by atomic layer deposited Al2O3
,”
J. Appl. Phys.
104
(
4
),
044903-1
044903-12
(
2008
).
2.
D. B.
Fenner
,
D. K.
Biegelsen
, and
R. D.
Bringans
, “
Silicon surface passivation by hydrogen termination: A comparative study of preparation methods
,”
J. Appl. Phys.
66
(
1
),
419
424
(
1989
).
3.
J.
Schmidt
 et al, “
Surface passivation of high-efficiency silicon solar cells by atomic-layer-deposited Al2O3
,”
Progress in Photovoltaics Research & Applications
16
(
6
),
461
466
(
2008
).
4.
A. G.
Aberle
, “
Surface passivation of crystalline silicon solar cells: A review
,”
Progress in Photovoltaics Research & Applications
8
(
5
),
473
487
(
2000
).
5.
H.
Mackel
and
R.
Ludemann
, “
Detailed study of the composition of hydrogenated SiNx layers for high-quality silicon surface passivation
,”
J. Appl. Phys.
92
(
5
),
2602
2609
(
2002
).
6.
M.
Giesen
 et al, “
Characterization of p-n junctions and surface-states on silicon devices by photoemission electron microscopy
,”
Applied Physics A
64
(
5
),
423
430
(
1997
).
7.
G.
Declerck
,
Characterization of Surface States at the Si-SiO 2 Interface
(
Springer US
,
1979
), pp.
105
148
.
8.
A.
Chiabrera
, “
Semiconductors’ characterization: Kinetics of one energy-level recombination centers and surface states
,”
Solid State Electronics
15
(
3
),
277
284
(
1972
).
9.
E. A.
Fogels
and
C. A. T.
Salama
, “
Characterization of surface states at the Si-SiO[sub 2] interface using the quasi-static technique
,”
Journal of the Electrochemical Society
118
(
12
),
2002
(
1971
).
10.
P.
Saintcast
 et al, “
Very low surface recombination velocity on p-type c-Si by high-rate plasma-deposited aluminum oxide
,”
Applied Physics Letters
95
(
15
),
151502
(
2009
).
11.
L.
Song
 et al, “
Characterization of silicon surface states at clean and copper contaminated condition via transient capacitance measurement
,”
Applied Physics Letters
111
(
15
),
152103
(
2017
).
12.
L.
Song
,
L.
Mai
, and
S.
Wenham
, “
Laser induced localised hydrogen passivation
,”
Solar Energy
122
,
341
346
(
2015
).
13.
L.
Song
,
A.
Wenham
, and
S.
Wenham
, “
Low temperature diffusion and its impact on hydrogenation
,”
Solar Energy Materials & Solar Cells
149
,
221
225
(
2016
).
14.
L.
Song
 et al, “
Controlled cooling process for efficient hydrogenation
,”
Journal of Alloys & Compounds
698
,
892
897
(
2017
).
15.
L.
Song
, “
Electric field-enhanced hydrogenation
,”
Applied Physics A
122
(
10
),
930
(
2016
).
16.
K. J.
Chang
and
D. J.
Chadi
, “
Theory of hydrogen passivation of shallow-level dopants in crystalline silicon
,”
Physical Review Letters
60
(
14
),
1422
1425
(
1988
).
17.
L.
Song
,
J.
Wilson
, and
J.
Lee
, “
Laser recrystallization for efficient multi-crystalline silicon solar cell
,”
Journal of Physics D Applied Physics
49
(
31
),
315601
(
2016
).
18.
L.
Song
 et al, “
Laser enhanced hydrogen passivation of silicon wafers
,”
International Journal of Photoenergy
2015
,
1
13
.
19.
I.
Martín
 et al, “
Improvement of crystalline silicon surface passivation by hydrogen plasma treatment
,”
Applied Physics Letters
84
(
9
),
1474
1476
(
2004
).
20.
T. T. A.
Li
and
A.
Cuevas
, “
Role of hydrogen in the surface passivation of crystalline silicon by sputtered aluminum oxide
,”
Progress in Photovoltaics Research & Applications
19
(
3
),
320
325
(
2011
).
21.
T.
Heiser
 et al, “
Transient ion drift detection of low level copper contamination in silicon
,”
Applied Physics Letters
70
(
26
),
3576
3578
(
1997
).
22.
J. S.
Jeon
, “
Electrochemical investigation of copper contamination on silicon wafers from HF solutions
,”
Contemporary Economic Policy
143
(
9
),
2870
(
1996
).
23.
E.
Nes
and
J.
Washburn
, “
Transmission electron microscope investigation of the growth of copper precipitate colonies in silicon
,”
J. Appl. Phys.
44
(
1
),
486
487
(
1973
).
24.
M.
Seibt
 et al, “
Formation and properties of copper silicide precipitates in silicon
,”
Physica Status Solidi
166
(
1
),
171
182
(
1998
).