Download For the course implementation in materials several different topic

IMPLEMENTATION IN MATERIALS
SEMICONDUCTOR COMPONENTS
122828
GROEP 2
JASPER DIEPHUIS S0090743
STEFAN VEENHOF S0139246
SUPERVISORS:
JOOST MELAI
GIULIA PICCOLO
TABLE OF CONTENTS
INTRODUCTION
For the course implementation in materials several different topic where available. Four diverse topics
were presented to choose from, we made the choice to investigate photodiodes. The first sessions was a
simply safety course through the MESA+ labs and after this several session in the MESA+ followed. In
these sessions in the clean room 3 wafers were made which contained several simply photodiodes and
some other structures. After the completion of the wafers the next phase of the course was involved
doing several measurements on the wafers. Several characteristics of the photodiodes and wafers were
measured and can be found in this report. In short the contents of this report are as follows: Firstly the
theory behind photodiodes and their production will be discussed. In this section the questions from the
manual will also be answered. Secondly the actual creation of the photodiodes and other structures is
described as we went through this process. Lastly the measurements will be discussed and some
conclusions will be drawn.
THEORY AND PROBLEM DEFINITION
THE PHOTODIODE
A photodiode consists of a couple of layers. Firstly there is a n-type layer and a p-type layer. In between
these two layer is the depletion layer. Because of the transportation of holes from the p-type to the ntype layer and the transportation of electrons from the n-type to the p-type layer a electric field is created
in the depletion layer. This field will oppose the transportation of the electrons and the holes and a state
of equilibrium will be created.
When light is absorbed by the photodiode, a photon and its energy can create a free electron-hole pair by
colliding with one of the silicon atoms. The electric field of the depletion layer will separate this pair (if the
pair is in the depletion layer). These separated carriers will than contribute to the photocurrent.
QUESTIONS
Q1 (p16): For the standard photodiode with oxide you can also use mask 1 and 3. Mask 2 on the other
hand has to be changed, in what way do you think it must be changed?
On the contrary of what is said in chapter 3 of the manual, the displayed masks will create a photodiode
with oxide layer. As can be seen, mask 2 in figure 3.1 of the manual ensures that the oxide layer of the
photodiode is not etched away. When a photodiode without an oxide is to be made, mask 2 can be
disregarded.
Q2 (p18): Explain for the structures in figures 3.4 and 3.5 where the channel is. What are the channel
width and length, and where are the source, drain and gate?
The channel can be seen clearly in the first masks of both figures as it is the n-type area between the two
p-type areas. The width of the channel of figure 3.4 is given as e1 and is 100 µm the length is equal to e2,
600 µm. For the second type of MOSFET as shown in figure 3.5 the width of the channel is given as f1 and
is 20 µm. The length of the second channel is f2, 4240µm. The aluminum strip that is placed over the
channel is called the gate with both MOSFETs. The other two aluminum strips (on the sides) with both
MOSFETs are the drain and source.
Q3 (p36): Describe the two physical processes that are responsible for the breakthrough of diodes.
The first process is called the avalanche breakdown. This process occurs when a high reverse voltage is
applied which creates a large electric field across the depletion layer. The carriers in the depletion layer
are accelerated by the field. When the carriers gain enough energy they are able to ionize silicon atom
they collide with. This creates a new hole-electron pair wish will cause even further collisions and
ionization. This works quite like an avalanche which is why it has acquired the name avalanche
breakdown.
The second process is called the Zener breakdown. This effect occurs with heavily doped junctions which
have narrow depletion layers. If a reverse voltage is applied and the depletion region is too narrow for
avalanche breakdown (because carriers can not reach high enough energies over the small distance
traveled) the electric field will grow. The electric field can eventually become strong enough to pull
electrons directly from the valance band to the conduction band.
Q4 (p36): Why is the breakthrough voltage an important parameter in the characterization of a pnjunction and how can it be increased?
The breakdown voltage is mainly depend on the size of the depletion layer and so on the doping of the
pn-junction. To adjust the breakdown voltage the doping can be increased or decreased. The reason the
breakdown voltage is an important parameter is because, depending on the application, it is a very
desirable or undesirable effect. For instance some devices might not be able to handle the large current
that is created at the breakdown voltage and for these devices it is important to know the limit of the
reverse voltage that can be used. Other devices use the effect of the breakdown voltage for specific
purposes, for these devices the breakdown is exactly what is wanted.
Q5 (p36): Which factors influence the efficiency of the diode?
The most important factor is the lifetime of the electron-hole pairs. The longer the lifetime the higher the
chances of carriers, that were originally generated in the neutral region, of reaching the depletion layer. If
these carriers reach the depletion layer they will be separated by the electric field and contribute to the
photocurrent. Another factor is the recombination of holes at the surface, preventing this also increases
the efficiency. Lastly the reflection at the surface of the diode is an important factor. The lower the
reflection the more efficient the diode will be.
Q6 (p36): How can the efficiency be increased?
The lifetime of the electron-hole pairs depends on the impurities in the crystal and can be improved by
the application of special temperature steps during processing. To prevent recombination of holes at the
surface a layer of for instance SiO2 can be added on the surface and the dimensions of the metal contact
can be reduced. Minimizing the reflection can be done by added a special top layer (for instance S iO or
Al2O3) that reduces the reflection coefficient.
Q7 (p36): For which color of light is the photodiode most sensitive and why?
For this answer see the picture below, in which it is shown which wavelengths of light are reflected the
least. As can be easily seen light with a wavelength of around 0,6 µm is reflected the least. This
wavelength correspond with orange (perhaps even slightly red) light. The reasons the photodiode is the
most sensitive to this color are simply that if less light is reflectived more is absorbed and used.
Q8 (p36): In which way can the color sensitivity be shifted to the blue/ violet region and which problems
occur if you do this?
Applying top layers can reduce the reflection of a broader spectrum allowing the absorption of even blue
and violet light. A single top layer however only works only for a limited bandwidth. It is also possible to
use multiple top layer but applying these can be a difficult task.
Q9 (p36): Why is the curve factor always less than one?
Firstly the equation for the curve factor:
𝐶𝐹 =
𝑉𝑀𝑃 ∙ 𝐼𝑀𝑃
𝑉𝑂𝐶 ∙ 𝐼𝑆𝐶
This formula can also been seen as the maximum power divided by the ideal power. Now consider the
following figure:
In practice the VOC is always higher than Vmp and Isc always higher (more negative) than Imp. This result in
the fact that the product of VOC and Isc will always be larger than the product of Vmp and Imp. Now look
back at the formula for the curve factor and it will be easy to realize that the curve factor can never be
more than (or equal to) one.
Q10 (p38): What causes the contact resistance and why is it so important?
The contact resistance is caused by the resistance between the aluminum and the p +-layer and also by the
p+-layer itself. The contact resistance act as a series resistance in the model of the diode. The effect of this
resistance can be seen in the figure below:
As can be seen the contact resistance has a large influence on the V-I characteristic of the diode. The
higher the contact resistance the lower the maximum power will be. This makes the contact resistance
very important, since it has a direct influence on the maximum power and efficiency of the diode.
Q11(p38): Why is the voltage being measured with separate probes?
…… don’t know. Perhaps because the voltage meter has to have a very high resistance to ensure no
current goes through it.
Q12 (p38): Where does the accuracy of the calculated values of R□ and RC depend upon?
The accuracy depends on the contact resistance of the aluminum contacts with the probe and off course
the accuracy of the measurement device.
DETAILS OF THE PROCESSING
INTRODUCTION
The fabrication process of photodiodes is described in the following paragraphs. 3 wafers with several
photodiodes, Greek Cross structures, Long diffusion areas and MOSFETs were created in the cleanroom.
All these structures were realized on a n-type silicon wafer on which a layer of SiO3 with a high boron
concentration was deposited. When the wafer is heated the boron will diffuse into the silicon and create
a p-type region. This is the process which allows for the creation of the required structures. The entire
process will now be discussed in detail.
DAY 1: CREATING P-TYPE REGIONS
The first thing done to the wafers was a thorough cleaning process to ensure there were no
contaminations of any kind on the wafer. This standard cleaning process will be described once, but was
in fact used many times. The wafers are first deposited in fuming pure HNO3. There are to beakers the
first is for the worst dirt and the second for the little that remains after the first. This ensures the second
beaker is always somewhat clean. The wafer spend 5 minutes in the first and than 5 minutes in the second
beaker to clean off all organic material. After this the wafer are rinsed in water. The wafer the proceed
into a beaker with a boiling (95°) 69% HNO3 solution. The wafer spend 10 minutes in this solution to clean
off any metals. After this the wafers a rinsed once again and dried using a spinner and some nitrogen to
remove the last drops.
The next step is to apply a photo resist to the wafer. Before applying the actual photo resist, a layer of
HDMS (hexamethyldisilazane) is applied to the wafer. This makes it easier for the resist to attach to the
wafer. The HDMS replaces the OH groups at the surface of the wafer with CH 3 groups. A few drops of
HDMS are applied and the wafer spins for 20 seconds to ensure an even distribution of the solution. After
that a few drops of resist 907/17 are applied and the wafer spins again to distribute it evenly. Now the
resist is applied and the wafer is baked for 5 minutes at 95° to remove the solvent.
Next is the actual lithography and the first mask was applied on the wafer. Since a positive resist was used
any areas touched by UV light will eventually be etched away. Only those areas that are not subjected to
UV light will retain their boron layer and only these areas will become p-type regions after heating. After
correctly positioning the mask the wafer is exposed to UV light for several seconds. Then the wafers are
baked for 60 second at 120°. The wafer are then put into a developing solution called TMAH
(tetramethylammoniumhydroxide) which removed the photo resist only in those areas exposed to UV
light. The wafer are baked once again after this step for 5-10 minutes at 120° (hard bake). The etching is
done with a HF solution which is buffered with NH4F.
𝐻𝐹 + 𝐻2 𝑂 ↔ 𝐻3 𝑂+ + 𝐹 −
𝑆𝑖 𝑂2 + 6𝐻𝐹 → [𝐻2 ][𝑆𝑖 𝐹6 ] + 2𝐻2 𝑂
Using the standard cleaning procedure described earlier the wafer are cleaned and the remaining resist
stripped from the wafers. After this the wafers are baked at 1100° in nitrogen for 30 minutes, during
which the p-type regions with boron doping are created.
2𝐵2 𝑂3 + 3𝑆𝑖 → 3𝑆𝑖 𝑂2 + 4𝐵
Followed by a bake of 1100° in oxygen for 30 minutes to oxidize then entire wafer.
DAY 2: DEFINING THE CONTACT POINTS
The day started by measuring the thickness of the oxide layer using a elipsometer. Using two lasers (one
of 632 nm and one of 1550 nm) the thickness is measured by looking at the change in polarization of the
laser when it is reflected.
𝑛 = 1 𝐴 𝑡632 = 88,67 ± 𝑛 ∙ 281,50𝑛𝑚
{
𝑛 = 0 𝐵 𝑡1550 = 370,22 ± 𝑛 ∙ 699,69𝑛𝑚
𝑡𝐴 = 370,17
𝑡𝐵 = 370,22
Thickness of the layer that was not etched.
𝑛 = 0 𝐴 𝑡632 = 117,94 ± 𝑛 ∙ 281,50𝑛𝑚
{
𝑛 = 0 𝐵 𝑡1550 = 118,81 ± 𝑛 ∙ 699,69𝑛𝑚
𝑡𝐴 = 117,94
𝑡𝐵 = 118,81
Thickness of the layer that was etched.
Once again a layer of photo resist is applied and then the second mask has be applied to the wafer. This
time careful alignment was needed, using a microscope and the arrows and crosses available on the
wafers and mask. Once the mask was properly aligned the wafers were exposed to UV light and baked
afterwards.
The etching process is once again done with a HF solution but in the dark. This is to prevent pitting and
staining. Light would generate charge in the photodiode (see the equation below) and this process
combined with the HF solution would create holes on the wafers.
𝑆 + 2𝐻2 𝑂 → 𝑆𝑖 𝑂2 + 4𝑒 − + 4𝐻 +
𝐶𝑜𝑛𝑡𝑖𝑛𝑒𝑢𝑠 𝑐𝑦𝑐𝑙𝑒 { 𝑖
𝑂2 + 4𝑒 − + 4𝐻 + → 2𝐻2 𝑂
After cleaning the sheet resistance of the wafers is measured using a 4-point probe. This method is
explained in more detail in the measurement section and will not be discussed now. The results of the
measurement were a bit inconsistent between the two machines that were used. However the sheet
resistance was around 10 Ohms.
DAY 3: ETCHING THE ALUMINUM
A layer of aluminum was applied to the wafers and as a result the third step of the cleaning process was
omitted. The boiling HNO3 solution would have cleaned off all the aluminum, which is why this was not
done when cleaning the wafers. Using the MAXIMUS machine the resist was applied (without HDMS,
because the resist doesn’t need to attach to silicon but aluminum) and the wafers were baked. After that
the third mask was aligned and the wafers exposed to UV light. The MAXIMUS machine was once again
used to develop the resist after which the wafers were hard baked. To etch the aluminum PES is used.
After etching the wafers are cleaned again (without boiling HNO3 off course).
After this a aluminum layer is deposited on the backside of the wafers, this is done to simplify contact
between the wafers and ground during measurements. The last step is called sintering, in this step the
wafers are baked at 400° inside a wet nitrogen (N + a little H2O) atmosphere for 10 minutes. In this time
free H+ attach to free Si connections, this is done to ensure that these free connections do not interfere in
later stages.
MEASUREMENTS
INTRODUCTION
The second phase of the course was measuring and testing. In the following section it is described how
the photodiodes and other structures where measured and tested. The results have partly been
processed in tables but another part has been added as an appendix because some of the graphs are not
digital. Further more the results be discussed in a short conclusion.
MEASUREMENT EQUIPMENT
Before the measurements and tests are described, the specification of the used equipment is discussed,
so it doesn’t repeated elsewhere in the report. Two devices were used, a semiconductor parameter
analyzer and a sort of DC analyzer.
HP-4145 Semiconductor Parameter Analyzer (SPA):
The most import specifications are given in the manual and have been copied into this report, see table
x.x.x.
7.1
QUALITATIVE MEASUREMENTS
In the first measurement (M-1) 40 photodiodes divided over 3 wafers were measured and tested. From
each diode the basic characteristics were measured: the short circuit current (𝐼𝑠𝑐 ), the open circuit voltage
(𝑉𝑜𝑐 ) and the breakdown voltage (𝑉𝑏 ). Beside these parameters, the photodiode is also classified in three
states (M-2). Good, bad and defect.
METHOD
With a I-V curve that includes the breakdown voltage and a forward conduction is it easy to classify the
photodiode. A bad photodiode shows the basic characteristics of a diode but with high a 𝑅𝑠 or/and low
𝑅𝑝 . A defect photodiode doesn’t show a characteristic that looks like a diode. Within this I-V curve the
basic characteristics are also presented and so it can be used for both measurements. The photodiodes
are measured evenly with and without oxide on top of the layer.
The measurement was done using the following construction. From the SPA there are two connections,
one ground and one signal. In the figure x.x.x. a simplified version of this construction is shown.
Figure x.x.x.
RESULTS
Measurement one (M-1) and two (M-2) are both shown in a table in appendix x.x.x. With this table the
difference between oxide and non-oxide involving the short-circuit current (𝐼𝑠𝑐 ) can be placed in a
histogram, see graph x.x.x. The numbers on the horizontal axis are the column number on the wafer. So
the couples that can be seen, are on the wafer, right next together.
Isc (uA)
70
60
50
Oxide
40
Non-oxide
30
20
10
0
1
2
3
4
5
6
7
8
9 10 1
3
5
7
9
1
3
5
7
9
# diode
In this graph it is easy to see the difference between the oxide and the non-oxide photodiodes. This
difference can be explained with the advantage of a top layer of SiO2. The most important advantage is
that the lifetime of a electron hole pair will be extended so that the chance that the pair will reach the
depletion region is higher. The lifetime is extended because the SiO 2 on top of the upper neutral region
prevents the recombination of the electron hole pair at the surface. The more pairs from the neutral
regions reach the depletion region, the more reverse current will flow during closed circuit. Another
reason is that the layer of SiO2 reduces the reflection of the photodiode. SiO2 is not the most useful
substance, but it is better the silicon surface. More successful chemicals are: SiO, Si3N4 and Al2O2.
7.2
PLOTTING THE I-V AND P-V CURVE AND DETERMINING R S AND R P
In this measurement the I-V and P-V curves of different photodiodes were measured and as a reference a
standard industry diode was also measured. In both measurements a controlled source of light will shine
on the photodiode, so the amount of reverse current (𝐼𝑟 ) can be manipulated.
METHOD
The same construction as the previous measurement was used, only this time the light intensity will be
the variable. In the first measurement (M-3) the I-V and P-V curves of the reference diode are measured
with a 𝐼𝑟 of 50 µA and 100µA. This in combination with a constant voltage of -0,25 V across the diode.
In the second measurement (M-4) the I-V and P-V curves of the diodes on the wafer are to be measured.
This time only with a 𝐼𝑟 of 100µA
In the third (M-5) and the fourth measurement (M-6) the resistances of two bad photodiodes with
respectively a bad serial resistance or bad parallel resistance have to be determined. The parallel
resistance can be found with one measurement with a reverse current of 100µA. The serial resistance can
be found by doing two measurements using a different reverse current in each measurement. In the table
x.x.x (grote tabel in 7.1) from the previous paragraph the photodiodes are listed with a classification, and
based on these results diode number 20 from wafer 1 is chosen for the 𝑅𝑝 measurement and diode
number 5 from wafer 2 is chosen for the 𝑅𝑠 measurement.
RESULTS
All the plotted curves are added to the appendix.
-M-3 appendix x.x.x.
-M-4 appendix x.x.x.
-M-5 appendix x.x.x.
-M-6 appendix x.x.x.
The results of measurement M-3 and M-4 are extracted from the graphs, 𝐼𝑠𝑐 , 𝑉𝑜𝑐 , 𝑃𝑚𝑎𝑥 and 𝐶𝐹(Curve
Factor) are given in de following table x.x.x
Task Wafer #
Diode #
M-3
NA
VTB8440B
-48,32
0,39
-13,6
0,722
NA
VTB8440B
-99,25
0,41
-29,2
0,718
1
3
-98,31
0,48
-37,3
0,790
M-4
Isc (uA)
Voc (V)
Pmax (uW)
CF
1
4
-102,8
0,48
-39,4
0,798
1
9
-99,12
0,48
-37,5
0,788
1
10
-100,7
0,48
-38,1
0,788
3
1
-98,66
0,48
-37,3
0,788
3
2
-95,84
0,48
-36,1
0,785
3
9
-102,5
0,48
-39,6
0,805
3
10
-98,15
0,48
-37,8
0,802
𝑅𝑝 from a bad diode
To determine the 𝑅𝑝 from a I-V curve the following method, shown in figure x.x.x from the manual, can be
used.
With the printed graph in the appendix (appendix x.x.x.) the 𝑅𝑝 is determined using a pencil and a ruler.
The following parameters were found:
∆𝑉 = 600𝑚𝑣
∆𝐼 = 6.25µ𝐴
These parameters entered into the formula give:
𝑅𝑝 =
∆𝑉
= 96𝑘𝑂ℎ𝑚
∆𝐼
Because the measurement is done with a pencil and a ruler the accuracy is relatively low. The accuracy of
the SPA can be neglected in comparison with that of the ruler. A approximation is that the maximum error
is half a millimeter on the grid of the graph. With these numbers a maximum and minimum resistance can
be calculated. The grid has a resolution of 3mV per half millimeter on the horizontal axis and roughly
0.5µA per half millimeter on the vertical axis. This gives a maximum and a minimum of the real 𝑅𝑝
𝑅𝑝 𝑚𝑎𝑥 = 105𝑘𝑂ℎ𝑚
𝑅𝑝 𝑚𝑖𝑛 = 88𝑘𝑂ℎ𝑚
𝑅𝑠 from a bad diode
From the manual, figure x.x.x gives the method needed to determine the 𝑅𝑠 .
The printed graph in the appendix (appendix x.x.x.) was also edited with a pencil and the data is acquired
with a ruler. The following parameters were found:
∆𝑉 = 25.5𝑚𝑣
∆𝐼 = 4.62µ𝐴
These parameters entered into the formula give:
𝑅𝑠 =
∆𝑉
= 5.5𝑘𝑂ℎ𝑚
∆𝐼
The possible error involved in this measurement can be compared with that of the 𝑅𝑝 measurement but
the scale is a bit different. The horizontal axis is the same, but the vertical axis is 1/3 µA per half
millimeter and that gives a minimum and maximum as follows:
𝑅𝑠 𝑚𝑎𝑥 = 6.6𝑘𝑂ℎ𝑚
𝑅𝑠 𝑚𝑖𝑛 = 4.5𝑘𝑂ℎ𝑚
7.3
MEASURING R  USING THE FOUR–POINT METHOD
In this paragraph the sheet resistance of the large p+ diffusion area is measured. This is done with the
four-point measurement method which will be explained in this paragraph.
METHOD
The four-point method is constructed as follows, see figure x.x.x.
A current is forced between the two outer contact points. With the two inner contact points the voltage is
measured. The current and the voltage that can be measured give the sheet resistance. The current does
not flow in a straight way but spreads over the sheet, therefore a factor necessary to compensate for this
behavior. This factor is given in the manual as 4,53. With this the formula for the sheet resistance is as
follows:
𝑅□ = 4,53 ∙
𝑉
𝐼
For this measurement the current is set on 100µA. A HP mulitmeter is used for this measurement, the
specification can be found be above. This multimeter also has a current source, so there is no other
equipment needed for this measurement.
RESULTS
Six diodes were tested two from each of the three wafers, on each wafer the two big non-oxide diodes
were tested. The results can be seen in table x.x.x. The results do not seem very reliable because the
sheet resistance is very high. During the measurements it was noticeable that the more pressure on the
wafer with the four point needles, the less the measured voltage was. During the measurement we
thought that the needles were pushed against the wafer strongly enough but perhaps this was not the
case. Luckily this is not the only sheet resistance measurement, in the following measurements the sheet
resistance will be measured again, this will allow us to check these results.
Wafer
#
Voltage
1
1
0,32
100
14496
1
2
0,41
100
18573
2
1
0,35
100
15855
2
2
0,095
100
4303,5
3
1
0,24
100
10872
3
2
0,38
100
17214
7.4
Set current (µA)
R□(Ohm)
THE SHEET- AND CONTACT RESISTANCE OF THE DIFFUSION AREA
When devices are created on a wafer there is always a point where it has to be connected to another
device. For this connection little space is available and so the contact resistance ( 𝑅𝐜 ) might be significant.
The contact resistance is the resistance from the aluminum contact holes to the diffusion area. On the
wafers a special structure was implemented to check this resistance. With this structure the sheet
resistance can also be measured.
METHOD
On the wafer the following structure was implemented:
The different lengths of diffusion areas (𝐿𝟏 and 𝐿𝟐 ) make it possible to determine 𝑅𝐜 and also the 𝑅□ .
The formulas are from the manual and are as follows:
𝑅□ =
𝐵
∙ (𝑅1 − 𝑅2 )
𝐿2 ∙ (𝑐 − 1)
𝑅𝑐 =
𝑐 ∙ 𝑅2 − 𝑅1
2 ∙ (𝑐 − 1)
With
𝐿1
𝐿2
The parameter 𝐵 is the width/length of the contact holes. The parameters 𝐿𝟏 , 𝐿𝟐 and 𝐵 are determined
during the fabrication and are as follows:
𝐿𝟏 = 200µ𝑚
𝑐=
𝐿𝟐 = 2000µ𝑚
𝐵 = 60µ𝑚
For this measurement the SPA is used but the way of measuring is a bit like the point-measurement. Two
probes drive a current through the diffusion area and two other probes measure the voltage caused by
the current. The measurement setup is shown in figure x.x.x.
RESULTS
Sheet resistance
In table x.x.x the result of this measurement are given.
Wafer #
#
R1 (Ω)
R2 (Ω)
R□(Ω)
Rc(Ω)
1
3
116
1170
35,13
-0,56
1
4
112
1080
32,27
2,22
1
6
110
1030
30,67
3,89
1
8
110
1010
30,00
5,00
2
1
115
1020
30,17
7,22
2
3
115
1040
30,83
6,11
2
5
109
1030
30,70
3,33
2
7
102
1030
30,93
-0,56
The sheet resistance seems this time more reliable because the measured values are more stable and the
technique used is also more reliable than the previous four point measurement. Looking at these results
the previous results seem very unlikely. Putting the results of both 𝑅□ measurements in a histogram is not
very helpful, since the results differ so greatly.
Contact resistance
A histogram of the contact resistance (𝑅𝑐 ) can be found in graph x.x.x.
Contact Resistance
Rc (Ω)
8.00
6.00
4.00
R_c
2.00
0.00
-2.00
1--3
1--4
1--6
1--8
2--1
2--3
2--5
2--7
wafer # -- diode #
The most remarkable observation is the fact that there are negative results. Negative resistances are not
considered valid results and might be caused by a broken device on the wafer. Why these faults are so
clear when the contact resistance is calculated is shown by the formula that is used. The formula used to
determine the contact resistance contains the following factor:
𝑐 ∙ 𝑅2 − 𝑅1
𝑐 = 0,1
The parameter should be 𝑅2 > 10 ∙ 𝑅1
but it is very close to
𝑅2 ≈ 10 ∙ 𝑅1
And with a little error in the measurement a negative result is possible. There could be different reason
for these errors. One could be that de measurement error is too large and the probe needles are perhaps
not stable enough. Because the contact area is very small, a wrong contact is easily possible. This kind of
error is however not likely, because the SPA has an accuracy that is much higher. Another reason could be
that the devices are broken, not that a broken device can have a negative resistance, but it could lead to
negative results.
Because the results were processed after the completion of the measurements , there possibility to
perform the measurement again. Performing the measurement again could determine whether the
device was broken or the measurement method had flaws.
7.5
MEASURING THE SHEET RESISTANCE USING THE GREEK CROSS STRUCTURE
Another test structure that can be used for easy determination of the sheet resistance is the Greek Cross
structure, see figure x.x.x.
METHOD
Once again the SPA is used and just like in the previous measurement, four probes are used. Two probes
are to set up a current and the other two probes measure the voltage.
With this special structure there is also a formula given which can determine the sheet resistance when
the voltage and current are given. The formula is as follows:
𝑅□ =
𝜋 𝑉
∙
ln 2 𝐼
RESULTS
In table x.x.x. are the results given
Wafer
#
Measured Voltage (mV)
R□ (Ohm)
1
1
67,6
3064
1
5
62,6
2837
1
8
62
2810
2
1
60,9
2760
2
5
59,9
2715
2
8
59,5
2697
3
1
56,9
2579
3
5
53,8
2438
3
8
50,4
2284
This is the third measurement were the sheet resistance is measured, but it is also the third time that the
result is completely different from the other results. A average of the results is roughly 2,5kOhm, this is
lower than the measurement of 7.3 but it is still to high to be a sheet resistance. So this measurement
most likely gone wrong somewhere. The error that has occurred must be a structural error because the
results of the measurement are relatively close to each other. That leaves two possibilities, structural
error in the measurement or a structural error in the design of the device. The last one is a bit unreal
because that would stand for that all wafers ever made with this pattern would be faulty. On the other
hand no real structural measurement error can be thought of , so that makes it difficult. A more
unrealistic error would be that the given formula is wrong, but we look into in and we couldn’t find
something.
A summarize of the three measurements a the great difference is given in this graph (x.x.x.) The fact that
the vertical axis is logarithmic says enough over the abnormality.
100000
R□ (Ohm)
10000
1000
7.5
7.4
100
7.3
10
1
7.6
MEASURING THE JUNCTION DEPTH 𝒙𝒋 OF THE DIFFUSION AREA
The objective of last measurement is to determine the junction depth(𝑥𝑗 ) of the diffusion area. This can
be done with a simple yet destructive method and for that reason the measurement was already done for
us with another wafer.
METHOD
The method used for this measurement is the ball-grooving method. This method makes use of a steel
ball (diameter of 4 cm) to grind a pit into the
wafer. Once this pit is made, it has to be treated
with a few drops of stain etch (200ml of HF plus a
few drops of HNO3). Because the 𝑃 + -type silicon
oxidizes faster than the 𝑁 − -type region the
resulting pit can be seen as a circle with a dark
edge under the microscope, see figure x.x.x.
When distance a and b are measured, the
junction depth 𝑥𝑗 can be calculated.
𝑥𝑗 =
With r being the radius of the steel ball.
𝑎∙𝑏
2∙𝑟
RESULTS
The measurement that was made for us had the following result, see figure x.x.x.
The parameters are already given in the picture:
𝑎 = 184µ𝑚
𝑏 = 566.67µ𝑚
Using these parameters in the formula gives:
𝑥𝑗 =
𝑎∙𝑏
= 2.61µ𝑚
2∙𝑅
FINAL CONCLUSIONS
APPENDICES
Appendix x.x.x
Measure results of 7.1
Wafer #
Photodiode #
Isc (µA)
Voc
Vb
Classification
Reason
1
1
-31,58
0,4-0,6
-77
+
-
1
2
-21,51
0,4-0,6
-77,4
+
-
1
3
-29,35
0,4-0,6
-78,6
+
-
1
4
-21,32
0,4-0,6
-78,2
+
-
1
5
-30,42
0,4-0,6
-79,8
+
-
1
6
-22,1
0,4-0,6
-80,4
+
-
1
7
-31,68
0,4-0,6
-80,6
+
-
1
8
-23,72
0,4-0,6
-79
+
-
1
9
-35,78
0,4-0,6
-80
+
-
1
10
-27,43
0,4-0,6
-80
+
-
1
11
-37,12
0,4-0,6
-80,6
+
-
1
12
-30,88
0,4-0,6
-81
+
-
1
13
-38,07
0,4-0,6
-80
+
-
1
14
-30,6
0,4-0,6
-80
+
-
1
15
-38,72
0,4-0,6
-80,8
+
-
1
16
32,18
0,2-0,4
-80,2
-
Rp
1
17
-57,28
0,4-0,6
-80,6
-
Rp
1
18
-44,87
0,4-0,6
-81,2
-
Rp
1
19
-65,38
0,4-0,6
-82
-
Rp
1
20
-59,12
0,4-0,6
-81
-
Rp/Rs
2
1
-30,03
0,4-0,6
-80,2
--
Rs
2
2
-22,85
0,4-0,6
-79,4
--
Rs
2
5
-30,48
0,4-0,6
-79,6
--
Rs
2
6
-23,07
0,4-0,6
-79,4
--
Rs
2
9
-30,89
0,4-0,6
-80
--
Rs
2
10
-22,82
0,4-0,6
-80,2
--
Rs
2
13
-30,12
0,4-0,6
-78,4
--
Rs
2
14
-22,01
0,4-0,6
-80,6
--
Rs
2
17
-32,12
0,4-0,6
-79,2
--
Rs
2
18
-23,89
0,4-0,6
-79,2
--
Rs
3
1
-28,69
0,4-0,6
-79,4
+
-
3
2
-20,19
0,4-0,6
-78,6
+
-
3
5
-29,41
0,4-0,6
-80
+
-
3
6
-20,95
0,4-0,6
-80
+
-
3
9
-31,16
0,4-0,6
-81,6
+
-
3
10
-20,87
0,4-0,6
-82
+
-
3
13
-29,21
0,4-0,6
79,8
+
-
3
14
-21,41
0,4-0,6
81,2
+
-
3
17
-30,32
0,4-0,6
80,6
+
-
3
18
-22,43
0,4-0,6
81
+
-