The Effect of Hydrogen Dilution on Optical Bandgap and Ge Atom

Fig 4.5 and Fig 4.6 show that the relationship between hydrogen dilution and the Ge content and also the bandgap. When we increase the hydrogen dilution the germanium content will also increase due to the more efficient dissociation of GeH₄ than SiH4. This can be seen from the three GeH4 concentrations. Also, the incorporation of Ge atom into the film will lower the bandgap of the material so that we can manipulate the bandgap of SiGe alloy by control the amount of hydrogen dilution, which is shown in Fig 4.5. There is another thing shown from Fig 4.5 that

Fig 4.5 The dependence of optical band gap of a-Si1-xGex:H alloys on RH2 at various RGeH4

the bandgap drops drastically when hydrogen dilution between 0 and 2 and then

0 2 4 6 8

30

reaches a saturation value for three different GeH4 concentrations. Fig 4.7 shows another effect that as hydrogen dilution increase the H content in the film decreased, which arise from the hydrogen facilitates the dissociation of GeH₄ and strengthens the hydrogen etching effect so that the Ge content is increased while H content is decreased.

31

Fig 4.7 Hydrogen content as a function of hydrogen dilution ratio at different germane concentrations.

4.1.4 The Effect of RF Power on Conductivity

From Fig 4.8, the increase of rf power makes the photo conductivity rise and dark conductivity decrease. Also, the figure shows the same trend as above mentioned that when hydrogen dilution increase the deposition rate and bandgap decrease and reach a saturation region. The figure at bottom shows that the influence of hydrogen dilution to SiH₂/SiH, GeH₂/GeH ratio and power. As hydrogen dilution increase the SiH2/SiH, GeH2/GeH ratio drop because hydrogen radicals will etch the growing surface and also feed energy to make atoms find a binding site with lowest potential energy. Moreover, increase power can lower the SiH2/SiH and GeH2/GeH ratio this can be explained by that it gives more energy to let atom diffuse on the growing surface and find a lowest energy binding site.

32

Fig 4.8 Dependence of the photo-conductivity (σph) and dark-conductivity (σd) with hydrogen dilution ratio

4.2 a-SiGe:H Single Junction Solar Cell Optimization

Although we have made certain progress in cell performance, there are still some aspects needed to be improved for higher efficiency. Therefore, in this section, the improvement of a-SiGe single junction solar cell from a baseline to the latest result will be discussed. Besides, some possible issues that can cause efficiency improvement will also be disclosed here.

4.2.1 Effect of Buffer Layer on Cell performance

In this section, we replace original a-Si p-layer with undoped a-Si buffer layer and hypothesize that this change will increase JSC and FF. Then further increasing germane concentration in gas phase from 11.1 % to 16.7 %, the characterized Ge content by XPS

75 80 85 90

33

shows that the film deposited at 16.7% R_GeH4 is composed of about 38 % Ge atom and bandgap is 1.45 eV by Tauc plot [35].

From the former structure, the thin a-Si p-layer between a-Si p+ and a-SiGe i-layer formed a graded bandgap profiling, but doping boron also induce a large amount of defects in a-Si:H material; furthermore, when depositing i-layer, the substrate heating will provide enough energy to let boron diffuse into a-SiGe i-layer and decrease film quality. Therefore, inserting a buffer layer into p/i interface can resist boron diffusion and can eliminate the defects induced by doping boron which in turn improve the interface quality. Furthermore, the a-Si n-layer between i-layer and n+ layer may hinder charge carrier transport as well. So, by considering above, we made following two changes:

1. Replace a-Si p-layer with a-Si buffer layer

2. Replace both a-Si p-layer and a-Si n-layer with a-Si buffer layer.

Fig 4.9 shows the I-V curve comparison between cells with one p-layer replaced by a buffer layer and two buffer layer replaced. Solar cells represent here were not post annealed. Cell structure with two buffer layers insertion is higher in efficiency than the cell with one buffer layer insertion due to the improved J_SC and FF. Moreover, the short circuit current density and therefore efficiency is higher than former structure, which can be explained by the improved interface quality by lowering the defect density and prevention of impurity diffusion. So we can conclude that cell with structure p+/b/i/b/n+ will present better cell performance than the former structure p+/p/i/n/n+.

Another interesting observation can be seen from Fig 4.10 is that quantum efficiency had been improved compared with the former one by the increased absorption in the whole spectrum. However, the absorption range still has no significant shift toward long wavelength region; despite that germanium content in the

34

film is about 38 %. The results conflict the general understanding that elevated germanium content will directly increase the absorption of long-wavelength region.

The reason for below expectation absorption in long wavelength region is still not clear, and the possible reason maybe the diffusion length is shortened by the increased germanium content, so that the charge carrier excited by long wavelength photons will have larger chance to be recombined before collected by external circuit. So, if the hypothesis is correct, then further optimization of film thickness is needed for improvement of quantum efficiency, and other possible solutions are now under consideration.

Fig 4.9 Illuminated J-V characteristics of 216-nm-thick a-Si_1-xGe_x:H single junction solar cells prepared with one or two buffer layer.

0.0 0.2 0.4 0.6 0.8

35

Fig 4.10 QE spectra of a 216-nm-thick a-Si1-xGex:H single junction solar cell under zero bias and -0.5 V bias.

4.2.2 RF Power Effect on cell performance

Remind that post-annealing will provide some energy to vibrate molecule and ease tension within material, increasing RF power also provide energy for molecule rearrangement. When depositing thin film, the dynamic balance is very important, for material relax itself while atom continuously deposit on growing surface. This process is more efficient compared with post-annealing. Therefore, elevate RF power will facilitate molecule diffusion on the growing surface and speed the deposition rate.

Theoretically, higher RF power will improve film structure, but it also increase deposition rate so that the molecule might have no chance to diffuse due to the fast growth on growing surface. So, the preferable power is the result of dynamic balance between molecule relaxation and deposition rate.

Recall from previous result in Fig 4.8, when we elevate power from 20 W to 30 W, both photo and dark conductivity were improved nearly one order, as expected

36

with pervious discussion. But we notice one interesting thing from Fig 4.11 that cell with a-SiGe i-layer deposited at 20W has higher efficiency than cell deposited at 30 W.

This observation can be understood by that although increasing RF power will facilitate film relaxation, but the film deposited at 30 W will suffer from stronger ion bombardment than film deposited at 20 W, thus create more defect throughout the material. And in turn lower the FF at higher power. So, we can conclude that the film quality is result of compromise of dynamic situation. Further optimization between 20 W and 30 W should be done in the future.

Fig 4.11 Illuminated J-V characteristics of a-Si_1-xGe_x:H single junction solar cells with i layer prepared with 20 W and 30 W.

4.2.3 Thickness Variation for p+/b/i/b/n+ Structure

In this section we want to further optimize i-layer thickness because we elevate germane concentration from previous 11.1 % to now 14.3 %, so that the proper film

37

thickness will change. Recall that in section 3.2.2 we have mentioned that the optimized thickness will change with Ge content in the film and also found that i-layer thickness of 200 nm will represent best cell performance for germanium concentration 11.11% in gas phase. Another difference from sec. 3.2.2 is structure, the former baseline is p+/p/i/n/n+, while after a serious experiments we found p+/b/i/b/n+ will represent better cell performance. Hence the proper i-layer thickness should be find out again.

The comparison of I-V curve between different i-layer thickness 180 nm and 226 nm are shown in Fig 4.12, and cell with thinner i-layer thickness present lower efficiency than cell with 226 nm thickness. Although shorten the i-layer thickness makes J_SC become larger, but FF drops and compensate the increase of J_SC, so total efficiency is lower than thicker i-layer thickness. The reason of this result is not so clear because when thickness is thinned, charge carriers has higher possibility to be separated from the material before being recombined, which in turn cause larger JSC. However, the drop of FF is unexpected and might be an experimental error. Therefore, further examine of film thickness is needed to find out an optimal thickness.

38

Fig 4.12 Illuminated J-V characteristics of a-Si1-xGex:H single junction solar cells with different i layer thickness (180 nm and 226 nm).

4.2.4 Layer Thickness Variation with Elevated Ge Incorporation

In order to increase the ability to absorb more long-wavelength photons, the germane concentration in gas phase increased from 8.3% to 11.1%, and then evaluated the effect of i-layer thickness on cell performance. As shown in Fig 4.13 and Fig 4.14, cell with 200 nm i-layer thickness presents better cell performance compared with 150 nm and 300 nm ones. The main difference between these cells is J_SC, which can be explained by increased absorption in infrared region and diffusion length. As thickness vary from 150 nm to 200 nm, the current density increased due to the improved absorption of infrared photons. However, when film thickness vary from 200 nm to 300 nm, despite that the further increase in infrared photon absorption, FF decrease due to film thickness exceed the diffusion length of a-SiGe material and result in poor charge carrier transport. Therefore, the proper i-layer thickness in this experiment should be

39

200 nm and the proper thickness will change with the germanium content in the film.

In the following experiment, we will fix i-layer thickness around 200 nm for further increased in Ge content and then find tune i-layer thickness again to find out a more suitable thickness.

0.00 0.25 0.50 0.75

0 5 10 15

200 nm

300 nm

C u rr en t D en si ty ( mA /c m

2

)

Voltage ( V ) 150 nm

Fig 4.13 Illuminated J-V characteristics of a-Si1-xGex:H single junction solar cells prepared with different i layer thickness

40 junction solar cells on i layer thickness.

4.2.5 Back Contact Comparison

Back contact act as an electrode and also a light trapping structure, proper surface microstructure together with front TCO light trapping structure can increase effective optical path length by several times. TCO back contact can be sputtered on the top of n-layer and provide resistive ability to post-oxidation.

Thus, we examine three kinds of back contact: directly coat Ag onto semiconductor, ITO coated with Ag electrode and ZnO coated with Ag electrode, the

41

result is shown in Fig 4.15. As we can see from the figure, cell with n-layer coated with TCO and Ag electrode will represent higher JSC and also FF than Ag electrode only, due to the better charge carrier collection ability by TCO. Another thing shown in the figure

Fig 4.15 Illuminated J-V characteristics of 216-nm-thick a-Si_1-xGe_x:H single junction solar cells with different combination of back contact material.

Table 4.1 Illuminated J-V parameters of a-Si1-xGex:H single junction solar cells with different combination of back contact material.

V

OC

is that ZnO back contact is better than ITO in our experiment, so that we will choose ZnO as back contact for current research. Further improvement of TCO is now under

0.0 0.2 0.4 0.6 0.8

42

investigation.

4.2.6 Anneal Temperature Optimization

In the cell fabrication, post-annealing is also an essential part of standard cell fabrication process. But in the thin film characterization, material suffer from post-oxidation will certainly decrease film quality especially in microcrystalline based material [36], because the incorporated impurities such as oxygen atom served as a dopant that will contribute electron to rise dark conductivity and also create more defect lead to deteriorate film quality. Hence, post-annealing is not preferable for thin film characterization.

Apart from thin film characterization, post-annealing of cell covered by back contact will not deteriorate the cell performance; on the contrary, under proper post-annealing condition, cell performance will be greatly improved by providing extra energy to relax tension within not adequately bonded atoms and improve atoms arrangement. Finally, post-annealing provides a better interface contact between cell and material. Furthermore, the annealing condition will differ slightly from material to material. Therefore, we should test annealing condition before further investigation of cell performance.

Fig 4.16 and Fig 4.17 show the relationship between different annealing temperatures on cell performance lasting for 30 minutes with pump fully open. We can observe that when annealing at 155 ^OC (actual value), the cell reaches its maximum efficiency; however, further increase annealing temperature will deteriorate cell efficiency drastically. So the proper annealing temperature should be around 150 ^OC to

43

Fig 4.16 Illuminated J-V characteristics of 216-nm-thick a-Si1-xGex:H single junction solar cells post-annealed at various temperature (TS= 145 ^OC, 150 ^OC, 155 ^OC, 160 ^OC).

a-Si1-xGex:H single junction solar cell on post-annealing temperature.

44

achieve maximum cell efficiency and avoid suddenly dropping in cell efficiency. This temperature is also examined by other cells and cell maximum efficiency shows agreement with annealing temperature about 150 ^OC. So, in the following experiment we will set anneal temperature at 150 ^OC to ensure cell efficiency and maintain consistency.

4.2.7 Bandgap Engineering

From previous discussion, our cell structure and deposition condition had been changed a lot. The current structure is p+/b/i/b/n+ and p+-layer is a-Si p-layer. Intrinsic layer thickness is 226 nm. According to literature about a-SiGe […],bandgap profiling in intrinsic layer is an essential part to achieve higher cell efficiency and better stability (thickness can be thinner).

Due to the doping of germanium atom, the bandgap of a-SiGe material can be easily tuned by control germanium content in film, which provide a chanceto smooth the band gap. As we know, bandgap discontinuity is a problem is narrow bandgap material such as a-SiGe, charge carrier will be hard to transport through bandgap discontinuity site thus lower short circuit current. Therefore, if we dynamically change germane concentration (or stepwise change) in gas phase when deposition, we will have a gradually narrow and then become larger bandgap. Smooth bandgap will help charge carrier transport easily and also increase V_OC. Moreover, a gradual change of bandgap near n-layer can facilitate hole transport due to the smaller mobility of holes.

In pratical situation, there are a lot of possible bandgap profiling, and by the help of computer modeling, we conclude that the “V” shape will have best efficiency. The bandgap profiling used in our device is decreasing from about 1.75 eV to 1.53 eV near

45

p-layer, and then fix at 1.53 eV, finally gradually magnify the bandgap. Also note that the slope of bandgap changing of two sides are different, the slope is sharp near p-layer while the slope is smooth near n-layer. This is because the hole needs gradual change in bandgap to help them transport, while the sharp change near p-layer will give larger bandgap and V_OC, too.

Fig 4.18 illustrates the improvement of FF by bandgap grading. We change germanium content in film by stepwise change germane concentration in gas phase.

Although the step is not so smooth but the result still coincident with our expectation, graded bandgap increased cell efficiency by enhancing FF and V_OC. This is because the improved charge carrier transport by bandgap profiling and also cause larger VOC. After knowing that, we will further optimize the bandgap grading structure.

Fig 4.18 Illuminated J-V characteristics of a-Si_1-xGe_x:H single junction solar cells prepared with constant and graded GeH₄ concentration.

0.0 0.2 0.4 0.6 0.8

46

4.2.8 n-layer Thickness Variation

The last part we examine the influence of n-layer thickness to device quality. As we can see from the Fig 4.19, the n-layer thickness is changed from 30 nm to 20 nm, and the cell efficiency has been greatly improved due mainly to elevated FF. Thus, the possible reason might be the decrease of resistance by decreasing n-layer, and we can get an idea from this figure that n-layer thickness will have impact on the device quality. Therefore, more tests should be done to examine the optimal thickness of n-layer.

Fig 4.19 Illuminated J-V characteristics of a-Si_1-xGe_x:H single junction solar cells prepared with constant and different graded GeH₄ concentrations.

0.0 0.2 0.4 0.6 0.8

0 5 10 15 20

V

_OC

J

_SC

FF η (mV) (mA/cm

²

) (%) (%) 20 nm 748 16.31 70.38 8.59 30 nm 752 16.46 65.90 8.15

C u rr en t D en si ty (mA /c m

2

)

Voltage (V)

20 nm n-layer

30 nm n-layer

47

4.3 Deposition of μc-SiGe:H Thin-Film and Performance of Solar Cells

From literature we know that microcrystalline silicon material grows near the amorphous to microcrystalline transition will have better film quality. To grow film near the transition we have some options to do, such as changing RF power, changing hydrogen dilution. Also, the amount of atomic germanium in the film plays an important role to film quality. In the following paragraph we discuss three factors mentioned above that will influence the crystallinity respectively.

4.3.1 Effect of RF Power on Film Quality

From Fig 4.20 we can see that as the power increases the silicon crystalline peak appears and then reaches saturation region after 200 W. We also notice that when power varying from 100 W to 200 W, the transition from amorphous to microcrystalline appears. We can compare the Raman spectra to conductivity (Fig 4.21). When varying power from 200 W to 600 W, the crystallinity reaching a saturation region so that the conductivity become almost the same. Also, the dark conductivity is about 10^-6 order which means that the film is highly crystallized. But when power varying from 100 W to 200 W (more specifically from 100 W to 130 W), a distinct change appear so we can conclude that the film deposited at 130 W is suitable for a microcrystalline silicon germanium film. Note that when depositing film at 130 W and 160 W there are some powder formed, which is not favorable for deposition. Some group observes the same phenomenon that powder formation when film depositing nearing the transition region. So we will fix power at 200 W and try to

48

changing hydrogen dilution to let the film deposit closer to the transition region.

100 200 300 400 500 600 700 600 W

300 W

200 W

100 W

S cat ter in g i n ten si ty ( a. u .)

Raman Shift (cm

^-1

)

Fig 4.20 Raman scattering spectra for μc-Si1-xGe_x films deposited on bare glass at fixed germane concentration (RGeH⁴) with various power ranging from 100 W to 600 W.

49

100 200 300 400 500 600

10

^-11

10

^-9

10

^-7

10

^-5

10

^-3

σ

_ph

σ ( S /c m)

Power (W) σ

_d

Fig 4.21 Dark conductivities (solid symbols) and photo conductivities (open symbols) as a function of discharge power.

Also, we can observe from Fig 4.22, the (220) direction is more pronounced as film deposited closer to the transition region so as to grain size. Note that the (220) crystallization direction is perpendicular to the growing surface, and the charge carrier in solar cell also transport in the direction perpendicular to the growing surface. So the (220) direction will be helpful to charge carrier transport. That is one of the reason that when film deposit at transition region will get better film quality.

50

Fig 4.22 XRD spectra of μc-Si1-xGex films deposited on bare glass at fixed

germane concentration (RGeH⁴) with various power ranging from 100 W to 600 W.

4.3.2 Effect of Hydrogen Dilution on Conductivity

Here we can see from Fig 4.23 that when hydrogen dilution increased the crystallinity increased and then reached a saturation region. This is because when hydrogen dilution ratio increased, the amount of atomic hydrogen at growing surface is also increased. So the hydrogen etching effect is more pronounced and leading to higher crystallinity. Once the mount of atomic hydrogen reaches a threshold value at the growing surface, the ability to crystallize the film is the same for all hydrogen dilution. So film deposited at hydrogen dilution ratio of 95.3 % is close to the transition region. But there are also some powders formed. Therefore the best condition would be hydrogen dilution at 96.3 %, and we also need further improvement.

51

Compare Fig 4.23 and Fig 4.24, the highly crystallized film reflect that there is no significant change in conductivities. The photo response is about one order for all the film, but the dark conductivity is a little bit too high Hence further improvement will be decreasing the crystallinity by lowing pressure.

200 400 600

Fig 4.23 Raman scattering spectra of μc-Si1-xGex films deposited on bare glass at

Fig 4.23 Raman scattering spectra of μc-Si1-xGex films deposited on bare glass at

在文檔中 非晶與微晶矽鍺合金作為薄膜太陽能電池吸收層之特性分析與最佳化 (頁 43-0)