A RT liquid surface passivation technique to investigate the recombination activity of bulk silicon defects is described. For the technique to be successful, three critical steps are required: (i) chemical cleaning and etching of silicon, (ii) immersion of silicon in 15% hydrofluoric acid and (iii) illumination for 1 min.
A procedure to measure the bulk lifetime (>100 µsec) of silicon wafers by temporarily attaining a very high level of surface passivation when immersing the wafers in hydrofluoric acid (HF) is presented. By this procedure three critical steps are required to attain the bulk lifetime. Firstly, prior to immersing silicon wafers into HF, they are chemically cleaned and subsequently etched in 25% tetramethylammonium hydroxide. Secondly, the chemically treated wafers are then placed into a large plastic container filled with a mixture of HF and hydrochloric acid, and then centered over an inductive coil for photoconductance (PC) measurements. Thirdly, to inhibit surface recombination and measure the bulk lifetime, the wafers are illuminated at 0.2 suns for 1 min using a halogen lamp, the illumination is switched off, and a PC measurement is immediately taken. By this procedure, the characteristics of bulk silicon defects can be accurately determined. Furthermore, it is anticipated that a sensitive RT surface passivation technique will be imperative for examining bulk silicon defects when their concentration is low (<1012 cm-3).
High lifetime (>1 msec) monocrystalline silicon is becoming ever more important for high efficiency solar cells. Understanding the recombination characteristics of embedded impurities has been, and remains an important topic. One of the most widely used techniques to examine the recombination activity of grown-in defects is by a photoconductance method1. By this technique it is often difficult to completely separate surface from bulk recombination, thus making it difficult to examine the recombination characteristics of grown-in defects. Fortunately there exist several dielectric films which can achieve very low effective surface recombination velocities (Seff) of < 5 cm/sec, and thus effectively inhibit surface recombination. These are, silicon nitride (SiNx:H)2, aluminum oxide (Al2O3)3 and amorphous silicon (a-Si:H)4. The deposition and annealing temperatures (~400 °C) of these dielectric films are considered to be low enough not to permanently deactivate the recombination activity of the grown-in defects. Examples of this are the iron-boron5 and boron oxygen6 defects. However, recently it was found that vacancy-oxygen and vacancy-phosphorus defects in n-type Czochralski (Cz) silicon can be completely deactivated at temperatures of 250-350 °C7,8. Similarly a defect in float-zone (FZ) p-type silicon was found to deactivate at ~250 °C9. Therefore, conventional passivation techniques such as plasma enhanced chemical vapor deposition (PECVD) and atomic layer deposition (ALD) may not be suitable for inhibiting surface recombination to examine grown-in bulk defects. Furthermore, SiNx:H and a-Si:H films have been shown to deactivate bulk silicon defects through hydrogenation10,11. Therefore to examine the recombination activity of grown-in defects, a RT surface passivation technique would be ideal. Wet chemical surface passivation fulfils this requirement.
In the 1990s Horanyi et al. demonstrated that immersion of silicon wafers in iodine-ethanol (I-E) solutions provides a means to passivate silicon wafers, achieving Seff < 10 cm/sec12. In 2007 Meier et al. showed that iodine-methanol (I-M) solutions can reduce the surface recombination to 7 cm/sec13, while in 2009 Chhabra et al. demonstrated that Seff of 5 cm/sec can be attained by immersing silicon wafers in quinhydrone-methanol (Q-M) solutions14,15. Despite the excellent surface passivation achieved by I-E, I-M and Q-M solutions, they do not provide adequate surface passivation (Seff <5 cm/sec) to measure the bulk lifetime of high purity silicon wafers.
Another means to achieve a high level of surface passivation is by immersing silicon wafers in HF acid. The notion of using HF to passivate silicon wafers was first introduced by Yablonavitch et al. in 1986, who demonstrated a record low Seff of 0.25±0.5 cm/sec16. Although excellent surface passivation was attained on high resistivity wafers, we have found the technique to be non-repeatable, thus adding a large uncertainty to the lifetime measurement. Therefore to limit the uncertainty by consistently achieving a very low Seff (~1 cm/sec), we have developed a new HF passivation technique that incorporates three critical steps, (i) chemically cleaning and etching of silicon wafers, (ii) immersion in a 15% HF solution and (iii) illumination for 1 min17,18. This procedure is both simple and time efficient in comparison to the traditional PECVD and ALD deposition techniques listed above.
1. Experimental Setup
2. Preparing the 15% HF Solution
Note: HF is a dangerous chemical and must be treated with care. It causes slow, sustained, and deep damage to the body following exposure. HF does not readily burn the skin like other acids – rather it absorbs quickly into the skin and causes deep blistering and damage to bones. This means that bones become brittle and blistered as the fluorine reacts with calcium. HF also binds with free calcium which is used in nerve regulation and osmotic cell balance, so binding of free calcium in the body can be fatal. It is paramount that the user follows laboratory safety protocols when using HF, and ensures they know the location of the HF first aid kit and hexafluorine (or calcium gluconate gel).
3. Calibration of the Lifetime Tester
4. Wet Chemical Treatment of Silicon Wafers Prior to Measuring
5. Measurement Procedure
6. Chemical Cleaning of Beakers and Containers
Figure 1a shows a schematic and Figure 1b shows a photograph of the experimental setup. When a silicon wafer is immersed into the HF solution, subsequently placed onto the lifetime tester stage and a measurement is performed (before illumination), a lifetime curve which is limited by surface recombination will result, as shown by the blue triangles in Figure 2. However, when the sample is illuminated for 1 min (while immersed in HF), as shown in Figure 1, and a measurement is performed immediately after illumination, an increase in the lifetime will occur, as shown by the red circles in Figure 2. This increase in lifetime after illumination results due to a reduction in surface recombination, and thus the red circles in Figure 2 represent the lifetime which is now limited by bulk recombination and not surface. The increase in lifetime post illumination will vary from sample to sample, however if the technique is working correctly, an increase in lifetime should always occur provided the bulk lifetime is not low (<100 µsec), whereby any reduction in surface recombination by illumination will not improve the lifetime because bulk recombination becomes dominant.
Although the bulk lifetime is attained after illuminating the silicon wafer for 1 min19, the surface passivation is temporary and will start to degrade within seconds of the halogen lamp being switched off. Thus it is important for the measurement to be performed directly after the illumination period to achieve the lowest surface recombination, as demonstrated in Figure 3. The red circles in Figure 3 correspond to the lifetime when the sample is measured directly after the lamp is switched off, and the blue circles correspond to the lifetime when the sample is measured 1 min after the illumination period. From the figure, it is evident that the high level of surface passivation is temporary and degrades within seconds of the illumination source being terminated. Therefore it is imperative that a measurement is performed directly after illumination to attain the bulk lifetime of the silicon wafer. In contrast, Figure 3 also demonstrates that even when the lifetime degrades (blue circles), it can be completely recovered by illuminating the silicon wafer once again. This process can be repeated many times without any permanent increase in surface recombination as shown in Figure 3.
To ensure the technique is working correctly each time a measurement is conducted, control silicon wafers should be used. The control silicon wafers represent samples that have been measured by the technique multiple times and have produced the same lifetime each time. The control samples should always undergo the same wet chemical pretreatment as the samples to be measured. Prior to measuring the bulk lifetime of silicon wafers, the control samples should be measured first. Thus if their lifetime is not within ±20% of their previous lifetime measurement, a problem has occurred and the measurement should be aborted until the issue is resolved. On the contrary, if the control samples produce lifetimes within ±20% of their previously measured lifetime, the measurements can proceed. In some cases, the bulk lifetime of the samples will be low, and thus the measured lifetime before and after illumination will be the same, contrary to Figure 2. In this case, even though illumination will still reduce surface recombination, no improvement in the lifetime post illumination will be observed because bulk recombination is much higher than surface. When measuring a sample such as this, comparison to a control wafer can elucidate whether there is a problem with the measurement setup or the measured wafer simply has a very high bulk recombination.
To demonstrate that HF passivation does achieve measurements of the bulk lifetime, FZ 1 Ω-cm n– and p-type silicon wafers were passivated with ALD Al2O3 and PECVD SiNx, as shown in Figure 4. Figure 4 shows that for both doping types, HF passivation can attain the same lifetime as achieved with Al2O3 and SiNx films. The lower lifetime achieved by the SiNx film on the p-type sample is due to depletion region recombination caused by the positive charge contained within the SiNx film. In contrast, depletion region recombination, if present, does not appear to significantly affect the lifetime measurement of either n– or p-type silicon when using the HF passivation technique9,17,18. This also makes the technique desirable for analyzing bulk defects, because any injection dependence observed from the lifetime measurement can be attributed to bulk recombination and not surface.
Figure 1. HF passivation setup. (A) schematic of the light-enhanced HF passivation and measurement setup17. Reproduced with permission from J. Solid State Sci. Technol.,1(2), P55 (2012). Copyright 2012, The Electrochemical Society. (B) photograph of the setup. Please click here to view a larger version of this figure.
Figure 2. Enhancement of the surface passivation by illumination. Effective lifetime of a high resistivity silicon wafer immersed in 15% HF, before (blue triangles) and directly after (red circles) illumination. Please click here to view a larger version of this figure.
Figure 3. Degradation of the surface passivation post illumination. Effective lifetime of a 5 Ω-cm n-type silicon wafer immersed in HF directly after illumination (red circles) and 1 min after illumination (blue circles). The figure demonstrates how the passivation can be recovered by subsequent illumination steps. Please click here to view a larger version of this figure.
Figure 4. Comparison with other dielectric films. Effective/bulk lifetime of FZ 1 Ω-cm n– and p-type silicon wafers passivated with HF (orange circles), PECVD SiNx (green squares) and ALD Al2O3 (blue diamonds). Please click here to view a larger version of this figure.
The successful implementation of the bulk silicon lifetime measurement technique described above is based on three critical steps, (i) chemically cleaning and etching the silicon wafers, (ii) immersion in a 15% HF solution and (iii) illumination for 1 min17,18,19. Without these steps, the bulk lifetime cannot be measured with any certainty.
As the measurement technique is conducted at RT, the surface passivation quality is highly susceptible to surface contamination (metals, organic films). Thus to effectively remove surface contaminants, an SC 1 solution is used (H2O : NH4OH : H2O2)20. When the silicon wafers are immersed in SC 1, the solution removes organic surface films by oxidative breakdown and dissolution, along with metal contaminants such as, gold, silver, copper, nickel, cadmium, zinc, cobalt and chromium20. Post SC 1 cleaning, it is possible that some trace elements remain trapped in the hydrous oxide film resulting from the clean, and therefore an HF dip is required to remove the film. Following an HF dip, the silicon surfaces are then cleaned in SC 2 (H2O : HCl : H2O2)20. While SC 1 effectively removes most impurities, SC 2 is designed to remove alkali ions and cations such as aluminum, iron and magnesium. Furthermore, SC 2 will also remove any other metallic contaminants that weren't removed in SC 1. Following an SC 2 clean, the wafers can be HF dipped to remove the hydrous oxide film. Once the silicon wafers have been cleaned of surface contaminants by SC 1 and SC 2, they require a short surface etch in TMAH. TMAH is an anisotropic etch solution, meaning it only etches along (111) crystal plains. Therefore during the chemical etch, small silicon pyramids are formed on the surfaces, exposing (111) plains, which roughens the surfaces and helps improve the hydrogen coverage when immersed in HF21,22. Therefore, with an optimized surface condition, surface recombination can be inhibited when the treated silicon wafers are immersed in HF and subsequently illuminated.
Optimization of the HF solution was examined in our previous publication17. It was found that when silicon wafers are immersed in 15%-30% HF, the lowest surface recombination is attained. This occurs because the HF concentration is high enough to passivate most of the silicon dangling bonds with hydrogen, and provides a field effect passivation mechanism by retaining a high surface charge caused by a difference in the silicon Fermi level and reduction potential of the HF solution23. The choice of 15% HF was for safety reasons. Another important addition to the HF solution was the inclusion of HCl. By adding a small amount of HCl in the 15% HF solution, the hydrogen concentration in the HF solution is increased, which in turn increases the amount of hydrogen available for the surface passivation of silicon wafers, allowing the bulk silicon lifetime to be obtained post illumination23.
Illumination of silicon wafers immersed in HF can significantly improve the surface passivation by creating additional bonds with chemical species in the solution through electron and hole transfer across the silicon/HF interface23,24. There are a number of chemical bonds that can form at the silicon/HF interface, such as Si-H, Si-OH and Si-F23–27. Passivation of silicon immersed in HF has been shown to primarily come from the creation of Si-H bonds, which is considered to be one of the most stable bonds when silicon is immersed in HF26,27. However, while the surface passivation is enhanced post illumination, as shown in Figure 3, the passivation is known to degrade within seconds after the illumination source has been terminated. Therefore it is unlikely the enhanced passivation post illumination is primarily due to the creation of stable Si-H bonds, as the passivation should not degrade if this were the case. Rather it is hypothesized that the enhanced surface passivation comes from the creation of unstable bonds with hydroxyl groups (Si-OH) and fluorine (Si-F)26.
While the three critical steps listed above are designed to give the best results, there are circumstances where the measurement can produce erroneous results. In most cases, the likely source of error is surface contamination, which can come from contaminated chemical solutions or a poorly filtered DI system. Under these conditions, it is best to test the DI water system using a conductance meter. If the DI water is not filtered correctly, the system requires changing before any measurements are to be performed. This will also affect other lab processes and thus a compromised DI water system will affect everyone. On the contrary, if the DI water is giving a sensible reading on the conductivity meter, the possible sources of contamination are the TMAH solution or the 15% HF solution (SC 1 and SC 2 will not be compromised). In this case, it is best to decant the solutions, chemically clean (SC 1 and 2) the containers and prepare new solutions. Furthermore, if the silicon wafers have been contaminated by a solution, they will require SC 1 and SC 2 cleaning multiple times before the surfaces are clean. To avoid solution contamination and thus erroneous lifetime measurements, it is best to prepare TMAH and HF solutions which are only used for this technique (and not by other processes in the laboratory). Another source of surface contamination could come from dielectric or metal films previously deposited on the silicon wafer. Thus if the wafers have undergone a dielectric or metal deposition, the surfaces require chemical cleaning and silicon etching prior to the three step process of the bulk lifetime measurement technique.
Although the technique is both simple and time efficient, the use of HF acid restricts the technique to a fume hood. Irrespective of this, the technique provides equivalent surface passivation to the best passivating dielectric films in the world (SiNx:H, Al2O3 and a-Si:H), furthermore this technique does not require any complex machinery, nor does it require elevated temperatures. As the purity of silicon wafers improves, in a drive to improve solar cell efficiencies, defect concentrations will decline and thus their recombination activity will become difficult to measure using techniques such as deep level transient spectroscopy and Fourier transform infrared spectroscopy. Therefore it is anticipated that minority carrier lifetime measurements which incorporate a RT liquid surface passivation technique will be imperative for examining bulk silicon defects when their concentration is low (<1012 cm-3).
The authors have nothing to disclose.
This program has been supported by the Australian Government through the Australian Renewable Energy Agency (ARENA). Responsibility for the views, information or advice expressed herein is not accepted by the Australian Government.
Hydrofluoric acid (48%) | Merck Millipore, http://www.merckmillipore.com/AU/en/product/Hydrofluoric-acid-48%25,MDA_CHEM-100334 | 1003340500 | Harmful and toxic. Any supplier could be used provided the chemical is Analytical Reagent (AR) grade. |
Hydrochloric acid 32%, AR | ACI Labscan, http://www.rcilabscan.com/modules/productview.php?product_id=1985 | 107209 | Harmful and toxic. Any supplier could be used provided the chemical is Analytical Reagent (AR) grade. |
Ammonia (30%) Solution AR | Chem-supply, https://www.chemsupply.com.au/aa005-500m | AA005 | Harmful and toxic. Any supplier could be used provided the chemical is Analytical Reagent (AR) grade. |
Hydrogen Peroxide (30%) | Merck Millipore, http://www.merckmillipore.com/AU/en/product/Hydrogen-peroxide-30%25,MDA_CHEM-107209 | 1072092500 | Harmful and toxic. Any supplier could be used provided the chemical is Analytical Reagent (AR) grade. |
Tetramethylammonium hydroxide (25% in H2O) | J.T Baker, https://us.vwr.com/store/catalog/product.jsp?product_id=4562992 | 5879-03 | Harmful and toxic. Any supplier could be used provided the chemical is Analytical Reagent (AR) grade. |
640 mL round plastic container | Sistema, http://sistemaplastics.com/products/klip-it-round/640ml-round | N/A | This is a good container for storing the 15% HF solution in. |
WCT-120 lifetime tester | Sinton Instruments, http://www.sintoninstruments.com/Sinton-Instruments-WCT-120.html | N/A | |
Dell workstation with Microsoft Office Pro, Data acquisition card and software including Sinton Software under existing license. | Sinton Instruments, http://www.sintoninstruments.com | N/A | |
Halogen optical lamp, ELH 300W, 120V | OSRAM Sylvania, http://www.sylvania.com/en-us/products/halogen/Pages/default.aspx | 54776 | Any equivalent lamp could be used. |
Voltage power source | Home made power supply | N/A | Any power supply could be used provided it can produce up to 90 Volts and 1-5 Amps. |
Conductivity meter | WTW, http://www.wtw.de/uploads/media/US_L_07_Cond_038_049_I_02.pdf | LF330 |