Determining Grain Size

In order to determine grain size a combination of hand and computer analysis was done on the SEM image with the highest magnification for each sample. The program ImageJ⁶, a free image-processing program, was used to automatically determine the area of each grain. The SEM images did not have enough contrast for ImageJ to find the edges of the grains by itself even with extensive computer manipulation. The solution was to outline each particle by hand with dark ink on a picture that was printed to scale. An example of a hand outlined image is shown in Figure 4.4. The outlined pictures were then scanned and the software was able to be used for the grain size analysis. Only grains that were fully inside the area of view were used to avoid underestimating the grain size. Effective grain diameters were calculated assuming all the grains were perfectly circular so that grains with very different shapes could be compared with a standard measurement. The equation used to calculate grain diameters was

√ (4.1)

where d is diameter and A is the area of the grain. To confirm the computer analysis, the diameters of a few mostly circular grains were measured both by hand and by the computer. These measurements for the diameters were found to agree to within 1 nm.

The average grain size was calculated by adding up the grain sizes of all the particles and dividing by the number of particles in the SEM image.

4.4 References

1. Osborne, S. J. Electrocodeposition of Nanoparticle Composite Films Using an Impinging Jet Electrode, University of California, San Diego, 2006.

2. Sweet, W. S. Investigation of Electrocodeposition Using an Impinging Jet Electrode, University of California, San Diego, 2006.

3. Thiemig, D., Osborne, S. J., Sweet, W. S., Talbot, J. B. Electroplating of Copper-Alumina Nanocomposite Films with an Impinging Jet Electrode. ECS Transactions 2008, 11, 35–44.

4. Ghosh, S. K.; Grover, A. K.; Dey, G. K.; Totlani, M. K. Nanocrystalline Ni-Cu Alloy Plating by Pulse Electrolysis. 2000, 48–63.

Figure 4.5: Hand outlined SEM image of sample N1 taken 15 days post-deposition.

5. EG&G Princeton Applied Research Model 273 Potentiostat/Galvanostat Operating Manual 1985.

6. ImageJ, Image Processing and Analysis in Java. http://rsbweb.nih.gov/ij/, 10/18/2012.

59 5. Results and Discussion

This section discusses the morphology and grain sizes for 3 copper deposits made on platinum and 4 deposits made on nickel using the IJE system at the limits of operation of the IJE system. The purpose was to investigate if grain sizes of less than 100 nm could be achieved and the effect of flow rate and current density on grain size and morphology. Since self-annealing has been reported to occur in nanograin

deposits of copper,^1-4 representative SEM images were taken at 15, 38 and 230 days post-deposition to investigate if self-annealing occurred.

The process conditions that were used were flow rate of 2.5 or 7 L/min and current density of 10 or 200 mA/cm², which were at the low and high limits of operation of the IJE system. All deposits where made with a plating solution of 1 M CuSO₄ at a pH of 1.5 and were deposited to a thickness of about 30 μm based on the process parameters used in previous work by Thieming⁵ that showed grain sizes of less than 100 nm. Since the deposit thickness was not measured during deposition the deposition was stopped when 64 C of charge had passed through the system. A final charge of 64 C is equal to a deposited thickness of 30 μm assuming 100 % current efficiency, which was found previously⁶.

The surfaces of the rinsed platinum and nickel substrates were similar as shown in the SEM images in Figure 5.1. The platinum substrate was a fairly smooth surface with a few imperfections, while there is some evidence of ridges in the nickel substrate. On the macro-scale the nickel substrate looked smooth while the platinum substrate had ridges on it. However, the morphology of the copper films deposited on these substrates under the same plating conditions with the IJE were quite different as will be discussed.

(a)

(b)

Figure 5.1: SEM images of (a) platinum and (b) nickel substrate surfaces.

5.1 Copper Deposits on Platinum Substrates

Three deposits were done on platinum and analyzed by SEM. Table 5.1 shows the plating conditions for the three samples.

Table 5.1: Plating conditions for copper deposits on a platinum substrate.

Sample Current Density

SEM images of the three samples 30 days post-deposition are shown in Figures 5.2-5.4. Samples P1 and P2 had small particles of approximately 100-250 nm in size on top of a smooth underlying deposit, while sample P3 had a very course granular structure with approximately 200-250 nm sized grains. X-ray diffraction was used to analyze the samples made on platinum and showed that the samples were pure copper, so the particles seen are not a result of a precipitate or an impurity.

Sample P1 shows small roundish particles that are on top of a much larger grained background. This morphology looks similar to SEM images of the nucleation step of copper deposition where the deposition time is so short that the substrate is not covered yet.⁷ It might be that layers are forming and the particles observed are from nucleation of a new layer.

Sample P2 looks similar to sample P1 in that there are larger grains with smaller particles on top. The small particles for sample P2 are larger (100-250 nm) than for P1 (100-150 nm) and more angular. The difference in size and shape of the smaller grains in P2 verses P1 could be due to the difference in current densities during deposition.

Figure 5.2: SEM image of sample P1 produced at 10 mA/cm² and 2.5 L/min.

Figure 5.3: SEM image of sample P2 produced with 200 mA/cm² and 2.5 L/min.

Sample P3 shows a very different morphology from samples P1 and P2 and is similar to deposits made with high overpotential observed by other methods of electrodeposition⁸ except that the grain sizes are much smaller. It has a very granular structure with most grains being clearly over 100 nm in diameter. While both P2 and P3 were made with a high current density, which is normally associated with the granular morphology seen in sample P3, the difference in morphologies may be due to higher flow rate used for P2. The morphologies seen for these deposits on platinum do not match the reported morphologies of Osborne,⁹ and Sweet¹⁰ for deposits made on copper or Thiemig⁵ on an unknown substrate.

5.2 Copper Deposits on Nickel Substrates

Since the copper deposits on platinum did not show the expected morphology or grain size reported in previous work by Thiemig⁵, Osborne⁹ or Sweet¹⁰ copper was plated onto nickel substrates. Table 5.2 shows the plating conditions for the four copper deposits on nickel substrates.

Figure 5.4: SEM image of sample P3 made at 200 mA/cm² and 7 L/min.

Table 5.2: Plating conditions for copper deposits on nickel substrates.

The morphologies of the four copper samples deposited on nickel showed grains of various shapes. As shown in Figure 5.5, at 15 days post-deposition samples N2, and N3 showed a mix of pyramidal and cuboidal grains that are not connected.

Pyramidal grains are most often seen in depositions with slow growth (low current density) and with few impurities.¹¹ While sample N2 was made with low current density, sample N3 was not. It is unclear why sample N2 was similar to N3 when they were made with different current densities; however the difference in flow rate at which they were deposited might be the reason. Sample N1 had oblong grains that while not touching were more closely grouped. The oblong grains were oriented randomly and did not line up with the fluid flow. Sample N4 had polygonal grains with some of them connecting to each other. Polygonal grains are more associated with higher current densities than pyramidal grains and sample N4 was plated with a higher current density. Figure 5.5 shows all four samples at 15 days post-deposition.

The copper deposits on nickel showed expected morphologies such as pyramidal,

cuboidal, and polygonal shaped grains¹¹ while also having grain sizes less than 100 nm.

5.2.2 Grain Size Distribution

The effects of plating conditions on the size and distribution of grains were examined. Table 5.3 shows the grain size data and Figure 5.6 shows the grain size distribution at 15 days post-deposition. At 15 days post-deposition the samples N1-N3

(a)

(c) (d)

(b)

Figure 5.5: SEM images of (a) N1, (b) N2, (c) N3 and (d) N4 at 15 days post-deposition.

had approximately the same average grain size, ~45 nm while sample N4 was quite a bit higher at 73 nm.

Table 5.3: Grain size data for samples 15 days post-deposition.

Sample

Figure 5.6: Grain size distribution for all samples at 15 days post-deposition.

The data for the deposits 38 days post-deposition are shown in Table 5.4 and Figure 5.7. After 38 days post-deposition sample N1, which was plated with both a low current density and a low flow rate, had the smallest grains, ~ 38 nm. Samples N2

0

0-15 15-30 30-45 45-60 60-75 75-90 90-105 105-120 120-135 135-150 150-165 165-180

% of Grains

and N3 had similar average grain sizes, ~60 nm while sample N4 had a much higher grain size average than the other samples, ~101 nm.

Table 5.4: Grain size data for samples 38 days post-deposition.

Sample

Figure 5.7: Grain size distribution of all samples measured 38 days post-deposition.

Table 5.5 and Figure 5.8 show the data for 230 days post-deposition.

0

0-15 15-30 30-45 45-60 60-75 75-90 90-105 105-120 120-135 135-150 150-165 165-180 180-195 195-210 210-225 225-240 240-255 255-270 270-285 285-300

% of Grains

After 230 days the images of samples N1 and N3, deposited with a low flow rate, have almost the same grain size, ~ 33 nm, which is smaller than the average grain sizes for the images taken at 15 and 38 days. This is most likely an anomaly because of the small sample size. Samples N2 and N4, deposited at high flow rate, also have nearly the same grain size, around 112 nm.

Table 5.5: Grain size data for samples 230 days post-deposition.

Sample Minimum (nm) Maximum (nm) Average (nm)

Standard

Figure 5.8: Grain size distribution of all samples measured 230 days post-deposition.

0-15 15-30 30-45 45-60 60-75 75-90 90-105 105-120 120-135 135-150 150-165 165-180 180-195 195-210 210-225 225-240 240-255 255-270 270-285 285-300

% of Grains

The following generalizations can be made. Generally, samples N1 and N3, plated with a low flow rate, had the smallest average grain size and smallest

distribution in grain size. When compared with the data from Thiemig the grain sizes from these experiments were smaller.⁵ Thiemig showed two samples with average grain sizes of 79 and 68 nm for plating conditions of 2.5 L/min with a current density of either 10 mA/cm² or 75 mA/cm², respectively.⁵ Sample N1 was made at with the same plating conditions as one of Thiemig’s samples, but shows an a grain size of almost half the size. A reason may be that the samples were stored some time before analysis and grain growth may have occurred. Sample N4, deposited with a high flow rate and a high current density, always had the highest average grain size of 73 -114 nm.

Overall these few samples were able to show that it is possible to obtain grain sizes less than 100 nm using an IJE system. It seems that a low flow rate gives the smallest grain sizes and high current density and high flow rate gives the largest grain sizes. While high current density is associated with smaller grain sizes⁵ this was not observed in these experiments. The flow rate seemed to have the most influence over grain size with a lower flow rate giving smaller grain sizes with the IJE system.

5.2.3 Time Dependence of Morphology

The morphology of the samples changes over time for all samples except N3.

These changes in morphology could be associated with self-annealing, the growth of grains over time at room temperature, observed for nanograin copper deposits.^12-14

Figure 5.9 shows the grain structure of sample N1 taken 15, 38 and 230 days post-deposition. Sample N1 has square and oblong grains that lay along the slopes of the larger mountains and valleys in the microstructure. There is a clear change in morphology between the SEM images taken at 15 and 38 days post-deposition. The grains that can be seen after 15 days were less distinct than the grains seen at 38 days post-deposition. There was no appreciable change in morphology observed between the SEM images taken at 38 and 230 days post-deposition.

The SEM images of sample N2 at 15, 38 and 230 days post-deposition are shown in Figure 5.10. For sample N2 the SEM images show a changing morphology where the grains change in shape and number as time progressed. At 15 days post-deposition there were only a few grains and most of them were square-like. At 38 days post-deposition there was a greater number of grains and they had become pyramidal in shape. After 230 days the grains were for the most part interconnected and

hexagonal in shape.

Figure 5.11 shows SEM images of sample N3 taken 15, 38 and 230 days post-deposition. For sample N3 the grains are pyramidal and cubic in shape and are not interconnected. The SEM images of sample N3 show little evidence of a changing morphology.

The SEM images of sample N4 taken 15, 38 and 230 days post-deposition are shown in Figure 5.12. Sample N4 shows similar morphology and growth to sample N3. There is not a significant difference in number of grains between the SEM images taken at 15 and 38 days, but the particles are flatter at 15 days post-deposition

compared to 38 days. At 230 days post-deposition the grains are interconnected and on top of each other.

(a)

(b)

(c)

Figure 5.9: SEM of sample N1 produced at 10 mA/cm² and 2.5 L/min at (a) 15, (b) 38 and (c) 230 days post-deposition.

74

(a)

(b)

(c)

Figure 5.10: SEM of sample N2 produced at 10 mA/cm² and 7 L/min at (a) 15, (b) 38 and (c) 230 days post-deposition.

Figure 5.11: SEM of sample N3 produced at 200 mA/cm² and 2.5 L/min (a) 15, (b) 38 and (c) 230 days post-deposition.

(b)

(c) (a)

(a)

(b)

(c)

Figure 5.12: SEM of sample N4 produced at 200 mA/cm² and 7 L/min at (a) 15, (b) 38 and (c) 230 days post-deposition.

5.2.4 Time Dependence of Grain Size

It was difficult to analyze grain growth, as only one SEM image of the deposit was used to calculate the grain size. However, it is clear that at least for the samples plated at a high flow rate, N2 and N4, there was substantial growth that occurred between 38 and 230 days post-deposition.

Table 5.6 shows the grain size data and Figure 5.13 shows the grain size distribution for sample N1. Sample N1 showed the grain size becoming smaller as time progressed. But since the standard deviations are large, the variation in grain size is most likely little or none. From Figure 5.13 the general size and distributions are similar for each time frame. This seems to agree with the lack of changes in

morphology observed.

Table 5.6: Grain size data of sample N1 over time.

Minimum (nm) Maximum (nm) Average (nm)

Standard Deviation (nm)

15 Days 19 100 44 22

38 Days 10 78 38 17

230 Days 11 106 34 19

Figure 5.13: Grain size distribution of sample N1 at 15, 38 and 230 days post-deposition.

In agreement with the change in morphology seen, sample N2 shows grain growth over time as shown in Table 5.7. Figure 5.14 shows the grain size distribution over time shifting to higher values as one would expect if the small grains are being cannibalized by the bigger ones as in self-annealing.¹ By taking the measurements at 15 days post-deposition as t = 0 it was possible to fit the data to equation 2.1 for the grain growth of sample N2 as shown in Figure 5.15. For n =1 a linear regression gave k = 45 and an R² value of 0.99.

Table 5.7: Grain size data for sample N2 over time.

Minimum (nm) Maximum (nm) Average (nm)

Standard

0-15 15-30 30-45 45-60 60-75 75-90 90-105 105-120 120-135 135-150 150-165 165-180

% of Grains

Size Range (nm)

15 Days

38 Days 230 Days

Figure 5.14: Grain size distribution of sample N2 at 15, 38 and 230 days post-deposition.

Figure 5.15: Fit of Equation 2.1 for sample N2 with t = 0 at 15 days post-deposition.

0-15 15-30 30-45 45-60 60-75 75-90 90-105 105-120 120-135 135-150 150-165 165-180 180-195 195-210 210-225 225-240 240-255 255-270 270-285 285-300

% of Grains

The grain size data for sample N3 is shown in Table 5.8. For sample N3 the average grain size increases, then decreases with time, but like sample N1 the values are close together with a large standard deviation, so it is likely that little or no grain growth occurred. Figure 5.16 shows the grain size distribution over time and it also shows little if any growth. While the change in morphology of sample N3 suggested self-annealing might have occurred between 15 and 38 days, this is not supported by the grain size data.

Table 5.8: Grain size data for sample N3 over time.

Figure 5.16: Grain size distribution of sample N3 at 15, 38 and 230 days post deposition.

Minimum (nm) Maximum (nm) Average (nm)

Standard Deviation (nm)

15 Days 24 114 47 22

38 Days 26 155 58 28

230 Days 14 131 33 23

Similar to sample N2, sample N4 showed grain growth over time as shown in Table 5.9. Figure 5.17 shows the same shifting of grain size distribution as sample N2 indicating self-annealing, which matches with the morphology changes seen for sample N4. Unlike for sample N2, the grain growth for sample N4 did not fit equation 2.1 (k =38, R² = 0.44) as shown in Figure 5.18. The average grain size at 230 days has a very large standard deviation which might account for the deviation from the

standard growth curve.

Table 5.9: Grain size data for sample N4 over time.

Minimum

Figure 5.17: Grain size distribution of sample N4 at 15, 38 and 230 days post-deposition.

0-15 15-30 30-45 45-60 60-75 75-90 90-105 105-120 120-135 135-150 150-165 165-180 180-195 195-210 210-225 225-240 240-255 255-270 270-285 285-300

% of Grains

Size Range (nm)

15 Days 38 Days 230 Days

Figure 5.18: Fit of Equation 2.1 of sample N4 with t = 0 at 15 days post-deposition.

According to Harper the time it takes for grain growth to occur is roughly equal for all deposits with a deposit thickness greater than 1 µm.¹² This was not seen with this data as all samples were the same thickness yet had different grain growth times. Using the 15 days as the time to stop growing for samples N1 and N3 and 230 days for samples N2 and N4 the k value was computed according to equation 2.2 using Harper’s given values of 0.23 µm for d0 and 5.5 hr for t₀.¹² Samples N1 and N3 had a k value of 6 while samples N2 and N4 had a k value of 92 compared to Harper’s value of 2.5.¹² While some of the error comes from not knowing the precise time that grain growth stopped, it is also clear with such different k values for deposits with the same thickness that these samples do not follow Harper’s grain growth model.

0

It is likely the changes in morphology and grain size observed are associated with self-annealing. Since all samples were deposited with the same plating bath and are the same thickness, the only variable studied that affects the self-annealing time was plating rate.^1,12,13

Sample N1 shows morphology changes only between 15 and 38 days post-deposition, which suggests self-annealing ended within that time frame. This does not agree with the grain size data, but this could be because of the small sampling of grain size. Sample N2 followed the basic grain growth model, equation 2.1, which agrees with pervious findings on self-annealing when no additives were in the plating solution.¹⁴ Since sample N3 shows little evidence of morphology change or grain growth that indicates that any self-annealing was completed before 15 days or that no self-annealing took place. Samples N2 and N4, which showed morphology and grain size changes after 230 days after deposition, could either have stopped self-annealing between 38 and 230 days or self-annealing could have continued past 230 days. Both of these scenarios are consistent with observations that grain growth from self-annealing can last for months.¹

Overall, the changes in morphology and grain size showing self-annealing do not match the predictions from literature that samples made at higher current densities self-anneal faster.¹ The samples made with the high current density, N3 and N4, do not show the same behavior and N4 self-annealed for a longer time than N2 which was made at a lower current density. The IJE system is different from other systems, including DC and pulse plating, used to create nanograin copper in that the fluid flow

is controlled and the fluid impinges on the substrate. This could account for the differences between these findings and literature.

5.3 References

1. Lagrange, S.; Brongersma, S. .; Judelewicz, M.; Saerens, a; Vervoort, I.;

Richard, E.; Palmans, R.; Maex, K. Self-annealing Characterization of

Electroplated Copper Films. Microelectronic Engineering 2000, 50, 449–457.

2. Dong, W.; Zhang, J.; Zheng, J.; Sheng, J. Self-annealing of Electrodeposited

2. Dong, W.; Zhang, J.; Zheng, J.; Sheng, J. Self-annealing of Electrodeposited

In document UC San Diego UC San Diego Electronic Theses and Dissertations (Page 71-0)