E-mail address:debarge@phase.c-strasbourg.fr(L.Debarge).
0927-0248/02/$-see front matter r2002Elsevier Science B.V.All rights reserved.
PII:S0927-0248(01)00171-4
used to passivate a-Si:H.The thin-film transistor (TFT)technology has also utilised the properties of the ECR-H 2plasma to reduce the carrier trap density of poly-crystalline silicon and to improve the leakage current in very short hydrogenation time [4,5].In comparison with RF plasma,the ECR technique offers a high ionisation rate together with low energy ions,resulting in a less damaged surface.
Besides,a selective emitter structure offers several advantages in comparison with a homogeneous emitter.It consists in tailoring the emitter in two parts:the metal contacts are deposited on a highly doped region,and a low doped emitter is diffused in between.This structure gives a good collection yield and good passivation properties between the contacts (due to the low surface concentration),and low contact resistivity is achieved.Previous works based on a single high temperature step have shown the efficiency of such structures,either using an emitter etch back technique (plasma assisted [6]or chemically [7]),or a selective diffusion in open-furnace [8].
An emitter etch backusing the metal grid as a maskhas been previously investigated [6],using the reactive ion etching (RIE)technique,leading to a self-aligned selective emitter (SASE).RIE is an interesting method to texture surfaces,usually done with fluorine-based gases.However sheet resistan
ce control on highly n +-doped emitters is very delicate with RIE based on fluorine gases because of a too high etch backrate.
In this paper,we present a selective emitter process,which utilise the simultaneous etching and hydrogenation properties of the ECR-H 2plasma.Firstly,after a short description of our ECR-plasma chamber (Section 2),we will study in detail the plasma etch effects on a phosphorous-diffused emitter (Section 3)with sheet resistance mapping and SIMS phosphorous profiles.Then,by mean of lifetime measurement,correlated with the previous observations,we discuss hydrogen diffusion and the bulkpassivation efficiency of the H 2-plasma (Section 4).Finally,photovoltaic results on screen-printed contacted solar cells are presented (Section 5).
2.ECR-plasma equipment
The reactor used is an FUV-4,from J.I.P.ELEC (Grenoble,France).The sample is held on a quartz handler,in the centre of a water-cooled stainless steel chamber.In this technique,the hydrogen radicals are formed in a microwave plasma (2,45GHz).The electric field is brought into the chamber by four antennas,placed in the upper part of the chamber.A magnetic field of 875Gauss enables the plasma excitation at the electron cyclotron resonance.The plasma is created in the upper part of the chamber
directly in contact with the sample.The sample is heated from the bottom with 12halogen–tungsten IR lamps through a double quartz window,with circulating water in-between.Initial pressure in the chamber before H 2enters is 10À5mbar.Working conditions investigated were in the following range:tempera-ture [350–4001C],incident power [300–800W],pressure [7–40mTorr]and duration
[5–60min].
L.Debarge et al./Solar Energy Materials &Solar Cells 72(2002)247–254
248
L.Debarge et al./Solar Energy Materials&Solar Cells72(2002)247–254249 3.Etch back analysis
This section and the following one(Section4)are dealing with use mirror polished 5cmÂ5cm mc-Si Polix s samples.A phosphorous emitter has been diffused(POCl3, 33O/sq),in order to characterise the effects of the plasma treatment as well as the hydrogen behaviour in regards to the phosphorous present in the emitter and the multicrystalline silicon grains.
In order to study the plasma etching effect on the emitter,we performed sheet resistance(r S)measurements on the surface before and after the plasma treatment. Although the initial r S is homo
geneous after the POCl3diffusion(3371O/sq),the differences in r S values after the H2-plasma are quite high,ranging between60and 160O/sq.On the sheet resistance mappings,one can clearly see that the etch rate on some grains is higher than on others(Fig.1).From X-ray measurements on these specific grains,one can conclude that(111)crystalline planes are less sensitive to the hydrogen plasma than the(100)planes,this due to higher number of backbonds for the silicon atoms in the case of the(111)planes[9].
It has been previously shown that H increases the resistivity of n-type materials [10,11].The sheet resistance measurements are integrating both emitter etch back effect and phosphorous passivation.In order to verify that the emitter is etched off, we performed SIMS measurements on different grains on the same multicrystalline sample.Three profiles are shown on Fig.2:two profiles are from the reference grain (Grain A),(100)oriented,before and after plasma treatment,and the third profile is from a neighbouring grain,(111)oriented,which also underwent the plasma(Grain B).The reference surface concentration after diffusion is2Â1020at/cm3.On
the Array Fig.1.Sheet resistance mapping revealing some grains orientation after the plasma step.
same grain,after the plasma treatment,the surface concentration drops to 8Â1017at/cm 3,having remo
ved 200nm of silicon.Nevertheless,the surface concentration of the Grain B remains almost unchanged,with only 20nm etched off.These results confirm the high selectivity of the plasma etch,depending on grain orientation,as shown with sheet resistance measurements.
Raman spectra done before and after the plasma step are showing no trace of amorphous silicon on the surface after the H 2-plasma.This is due to the soft plasma-surface interaction featuring the ECR technique.Hydrogen performs a chemical etch rather than a physical etch,because the ions impinging on the surface have a low energy.The Raman peakof silicon at 520cm À1is shifted to 522,6cm À1after the plasma step,which shows that the silicon matrix is densified due to the hydrogen incorporation [12].
4.Hydrogenation
Our purpose is to find out the plasma conditions for the best passivation properties on mulcrystalline wafers.We are using the photo-conductivity decay technique (PCD)to measure the minority carriers lifetime (mapping of 580points).The samples come from the same mc-Si ingot,are adjacent and processed in the same batch during the phosphorous diffusion.After the plasma treatment,the emitter is removed in a CP133solution.During the whole measurement process,both surfaces are passivated in a methanol/iodine solution.
On Fig.3,we map the bulklifetime of a sample before and after the ECR-H 2plasma step.
Before the treatment,the average lifetime is 25m s,and quite homogeneous all over the sample (the standard deviation of the lifetime value is only 5m s).With the best plasma conditions,the mean bulklifetime reaches the value of 44m
s.
Fig.2.Phosphorous profiles for different grain orientation.
L.Debarge et al./Solar Energy Materials &Solar Cells 72(2002)247–254
250
In some grains,the bulklifetime even reaches 60m s and more (see the bottom left corner of Fig.3b).On the contrary,minority carriers lifetime in some grains is not improved at all (centre right of the mapping).As a consequence,one can clearly see that there is a grain orientation effect on the minority carrier lifetime improvement,exactly as previously shown for the surface etch back.In addition,one can easily correlate the previous sheet resistance measurements with the actual bulklifetime mappings:lifetime improvement can be directly link ed to the surface etching.Indeed,the more the emitter is etched off,the better the bulk passivation is.
In order to explain this fact,we performed hydrogen concentration measurements (with SIMS characterisation)on the same grains where the phosphorous profiles were measured.
These profiles are shown on Fig.4and are to be compared with the phosphorous profiles (Fig.2)to understand the hydrogen behaviour.The full line on the graph is the hydrogen reference profile.The hydrogen concentration decreases quickly
characterisefrom Fig.3.Minority carrier lifetime mappings comparing bulklifetime before and after
hydrogenation.
Fig.4.Hydrogen profiles for different grain orientation.
L.Debarge et al./Solar Energy Materials &Solar Cells 72(2002)247–254251

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。