DEGREE PROJECT, IN MATERIALS SCIENCE AND ENGINEERING , FIRST LEVEL
STOCKHOLM, SWEDEN 2015
Plasma sprayed cylinder lining
coatings
FACTORS AFFECTING THE AMOUNT OF
CERAMICS IN PLASMA SPRAYED CYLINDER
LINING COATINGS
HERMINA ANTONSSON, TOVA JARNERUD
Plasma sprayed cylinder lining coatings
Factors affecting the amount of ceramics in plasma sprayed
cylinder lining coatings
Hermina Antonsson Nilsson
Tova Jarnerud
KTH Royal Institute of Technology
Materials Science and Engineering
Abstract
Scania uses powder plasma spray technology to coat their cylinder linings. The powder used contains stainless steel and ceramics. In order for the properties of the coating to reach the quality demands, it has to have a certain amount of pores and ceramics in it. Despite the process being strictly controlled the coating has proven to exhibit
differences in composition.
Process parameters for each coated lining are logged and as part of Scania’s quality control, samples of the coating are frequently tested. For the purpose of this work, three process parameters were mapped in search for correlations to the composition of the coating – electric current, voltage and power output. To investigate the homogeneity of the powder it was analyzed using Scanning Electron Microscopy.
The main findings were that there are no correlations found between the three process parameters and the composition of the coating. Moreover, in the powder analysis it was found that the composition of the powder itself varies to an extent that elements differ as much as twice the amount in between different batches and sites in the production. The results imply that it is difficult to prevent irregularities by managing process
parameters within the tolerance range. The variations in powder composition need to be further investigated in order to determine its impact on the quality of the coating.
Sammanfattning
Scania använder plasmabesprutning med pulver för beläggning av cylinderfoder. Pulvret som används består av rostfritt stål och keramer. För att beläggningens egenskaper ska uppnå kvalitetsmålen krävs en viss mängd porer och keramer. Trots att processen noga kontrolleras uppvisar beläggningen tendenser till skillnader i materialsammansättning. Som en del i Scanias kvalitetskontroll loggas processparametrar för varje belagt foder och stickprover av beläggningen testas. I det här arbetet har tre processparametrar valts ut i syfte att undersöka eventuella samband mellan dem och beläggningens
materialsammansättning. De valda parametrarna är elektrisk matning, spänning och effekt. Pulver har analyserats i svepelektronmikroskop i syfte att undersöka dess homogenitet.
Slutsatsen är att inget samband hittats mellan de tre processparametrarna och materialsammansättningen i beläggningen. Vidare kunde det i pulveranalysen ses att pulvrets sammansättning varierar så pass mycket att vissa ämnen förekommer i dubbelt så stor mängd på olika provtagningsområden.
Resultaten antyder att det är svårt att förhindra ojämnheter genom att styra
processparametrar inom de givna toleranserna. Variationen i pulvrets sammansättning bör undersökas vidare för att avgöra dess påverkan på beläggningens slutegenskaper och materialsammansättning.
Abstract Sammanfattning 1 Introduction ... 1 1.1 Background ... 1 1.2 Purpose ... 2 2 Method ... 2 2.1 Coating process ... 2 2.2 Process parameters ... 3
2.3 Quality testing at Scania ... 4
2.3.1 Samples for testing ... 4
2.3.2 Sample analysis ... 5
2.4 Collection of data ... 5
2.4.1 Powder analysis ... 5
2.4.2 Matching process parameters to test lab results ... 6
3 Results ... 6
3.1 Powder analysis ... 6
3.2 Matching process parameters to test lab results ... 7
4 Discussion ... 9 5 Conclusion ... 10 6 Acknowledgements ... 11 7 References ... 12 8 Appendices ... 13 8.1 Appendix 1 ... 13 8.2 Appendix 2 ... 16
1 Introduction
1.1 Background
Scania is a leading manufacturer of heavy trucks and busses, engines and services. It is a global company with a sales and service organization in more than 100 countries [1]. Part of the production includes coating of cylinder linings.
Scania uses plasma spray technology on their cylinder linings to coat it with a mixture of stainless steel and ceramics. The coating, in combination with lubricating oil, reduces the friction between the moving piston ring and the cylinder lining. This reduction of energy loss in turn reduces the fuel consumption. Moreover, the coating is highly enduring which prolongs the lifetime of the cylinder lining. In order for the lining to reach the quality demands, a certain amount of ceramics and cavities in the plasma layer is required.
The plasma state is one of the four states of matter in addition to solid, fluid and gas. Plasma is a mixture of neutral, positive and negative particles and cannot be stocked like the other states of matter [2]. When a molecule is heated to the plasma state, its
constituents are separated, meaning the electrons move freely from the nucleus. It requires high-‐energy input and can be reached only at very high temperatures in order to obtain ionization. Plasma technology is mainly used for technical applications.
Figure 1 -‐ Illustration of plasma spray technology using plasma torch and powder
Assuming they can be powdered, the plasma spray technology can be used with almost all materials having suitable melting temperatures. As seen in Figure 1, a constricted arc is used for the ionization of atomic gases (helium, argon) or molecular gases (nitrogen,
hydrogen) as well as dissociation of them. A powdered filler material is injected into the high-‐energetic plasma arc via a plasma torch with the help of a carrier gas [3]. The powder is not molten on to the surface, but mechanically attached to it. The quality and properties of the coating depend strongly on how the molten particles spread as they solidify on the surface formed by the previous layer [4]. This in turn depends on the particles’ surface temperature, degree of melting and their velocity. Only the hot and molten particles with the sufficient impinging velocity will form the coating since cold and unmelted ones tend to bounce off the surface.
At Scania, a powder consisting of 35% ceramics (Al2O3, ZrO2) and 65% stainless steel, is
received in a barrel from a supplier in Switzerland together with a chemical analysis specifying the composition of elements in it. The powder is then distributed through a silo to the pipes that lead it to a feeder by gaseous pressure. From the feeder it is then fed to a scale to ensure that the correct amount of powder is distributed to each lining in the plasma spraying process. From the scale, the powder is once again fed through pipes by argon gas to the plasma-‐spraying booth. Scania has two production lines with two booths respectively. In each booth there are two torches.
1.2 Purpose
The purpose of this work is to find out if there is any correlation between the process parameters and the amount of ceramics in the plasma coating of the cylinder lining. Scania’s goal is to be able to prevent irregularities in the coating by regulating process parameters. In search for correlations, relevant data is collected and analyzed.
2 Method
2.1 Coating process
The process of coating the cylinder linings, including preparation, consists of a number of steps. The linings, made of cast iron, are washed from grease and antirust residues. The inside is then blasted with aluminum oxide to achieve a rougher surface. Once the lining is in the plasma-‐spraying booth, powder is sprayed on with 75 grams per minute onto the inner surface in ten thin layers, making a total thickness of 250 𝜇m. The torch moves down and up five times inside the rotating lining to achieve an even distribution of the coating. Finally, the coated lining is marked with an identity code by an inkjet. The electric arc used has an energy input of 40V and 340A. In order to avoid differences in the voltage during the plasma-‐spraying process, the torch needs maintenance check every 45 running hours.
2.2 Process parameters
Each lining receives a log of all the individual process parameters including highest, lowest and mean value as well as standard deviation. These parameters are compiled and those considered interesting are marked in Table 1. Electric current, voltage and power output were the process parameters considered most relevant or likely to affect the quality of the coating due to exhibiting the largest deviations from their target values. The logs can be used for troubleshooting parameters by comparing them to their target values and tolerances. For example, a too high or too low voltage might result in a curved arc that distributes the powder unevenly and leaves unmelted particles. Thus the electric arc needs to be precisely adjusted.
Table 1 -‐ General process parameters and those selected for this work
*Unit of flow rate.
Parameter Unit Target value Tolerance Considered interesting for this work
Electric current A 340 +/-‐ 25 √
Voltage V 36 +/-‐ 6 √
Power output kW -‐ -‐ √
Argon NLPM* 40 +/-‐ 3 -‐
Hydrogen gas NLPM 3.2 +/-‐ 0.6 -‐ Shroud Ar NLPM 16 +/-‐ 1.5 -‐ AirJet (Air) bar 6 +/-‐ 0.4 -‐ Carrier gas (Ar) NLPM 3.2 +/-‐ 0.4 -‐ Feeder velocity g/min 75 +/-‐ 3 -‐
Stirrer % 80 +/-‐1.3 -‐
Pressure mbar -‐ -‐ -‐
Weight kg -‐ -‐ -‐
Cooling water flow l/min -‐ -‐ -‐ Cooling water inflow °C -‐ -‐ -‐ Cooling water outflow °C -‐ -‐ -‐
Conductivity 𝜇S -‐ -‐ -‐
Airflow P1 m/s -‐ -‐ -‐
2.3 Quality testing at Scania
2.3.1 Samples for testing
Samples of the coated cylinder linings are tested on a daily basis to check the quality of the coating. Eight samples are collected out of the daily production, usually around one thousand linings. In order to examine all the layers of the coating it is analyzed in the cross-‐section by destructive testing. In the lab, three coins are drilled out from each lining. Two from where the lining is exposed to the highest stress due to the piston changing direction – Bottom Dead Center (BDC) and Top Dead Center (TDC) -‐ and one from the middle part of it. BDC is where the ignition of the fuel air mixture occurs. The lab has 24 hours to approve the samples in order for the linings to be sent to the next step in the production. Figure 2 illustrates the positions in the cylinder.
2.3.2 Sample analysis
There are desirable pores in the coating that make sure the thin layer of oil is kept on the surface and fulfills its purpose of reducing friction by lubricating between the piston and the cylinder lining. An approved sample has 1.25-‐4% pores. However if a sample has 1-‐1.25% pores it can be approved given that the amount of ceramics is high enough. The amount of ceramics and pores, hereby referred to as CAP, has to be a total of 20% at minimum.
To analyze the samples, the Zeiss microscope software AxioVision SE64, is used to determine the amount of pores and CAP. In preparation for the microscope the coins are split in half to expose the cross-‐section, molded in Bakelite and then polished in six steps using diamond coated polishing wheels. An operator takes 12 pictures per coin and does an ocular inspection to detect if there are pores or ceramics not found by the software. In that case they can be manually registered. The software calculates a mean value based on the pictures taken. The operator then transfers these into a test lab Excel document together with the identity code of the cylinder lining.
2.4 Collection of data
For the purpose of this work, powder was collected and analyzed and process
parameters were extracted and interconnected with data from the test lab. The period from which the data was collected extends from September 17th to December 22nd 2014
and is limited to production line 2. The software AxioVision SE64 was tested in order to be familiar with the sample analysis procedure.
2.4.1 Powder analysis
The following test was performed. Small amounts of powder were collected from five different sites in the production line. It was analyzed to determine if there were any differences in its composition from that in the barrel in which it is delivered to that in the plasma-‐spraying booth, or in between different batches. The powder samples are hereby given the following designations -‐ numbered from 1 to 5.
1. A full barrel, on the top surface, in the middle.
2. The same full barrel, on the top surface, along the edge of the barrel. 3. An almost empty barrel.
4. The silo where the powder is distributed from the barrel to the feeders. 5. The booth, under the torch (what is left after the spraying process).
The powder was mounted on to double sided adhesive conductive carbon tape and coated with a thin layer of gold in a vacuum chamber to make nonconductive elements conductive. It was then imaged (Appendix 1) in SEM (Scanning Electron Microscopy Hitachi –S 3700N). For measurement of the elemental composition, the powder was analyzed using EDS (Energy Dispersive X-‐ray Spectroscopy). The SEM uses a focused beam of high-‐energy electrons that interact with the atoms in the sample, revealing information about its topography and chemical composition [6]. Each sample was
analyzed in three different areas: A, B and C (Appendix 2), and mean values for each sample was calculated and compiled in tables.
2.4.2 Matching process parameters to test lab results
A special designed MATLAB program was used to interconnect data from the test lab Excel document with data from the process parameter log. The MATLAB program extracts the desired data and stores it in columns in a new Excel document. Part of the new document consists of columns with the amount of pores and CAP of each sample from the test lab. The other part of it consists of columns with the title and unit of the process parameter and its highest, lowest, target and mean value respectively for each sample. The process parameters chosen for analysis in this work were electric current, voltage and power output as seen in Table 1. These data were plotted against each other to find correlations, if any.
3 Results
The results from the collected and analyzed data are presented below.
3.1 Powder analysis
In the EDS analysis, the distribution of elements of each of the five samples was analyzed in three different areas. Table 2 illustrates a compilation of the elements in them. In each respective sample, mean values of every element were calculated and transferred into Table 2. Values in highlighted cells are divergent as maximum (green) and minimum (yellow) for elements with larger variations. Notable is that there is carbon in the
powder, however as it is a light and small element it is not detectable in the EDS analysis and therefore not presented in the table.
Table 2 -‐ Mean atomic distribution of elements in five different samples of powder from EDS, maximum values highlighted in green, minimum values highlighted in yellow
EDS analysis Sample
Element [at. %] 1 [at. %] 2 [at. %] 3 [at. %] 4 [at. %] 5
Iron 22.57 19.93 11.69 11.28 21.12 Aluminum 27.52 27.95 33.89 22.94 27.11 Oxygen 42.59 44.80 49.38 49.69 44.34 Zirconium 2.61 3.08 2.43 2.51 2.97 Chromium 4.21 3.76 2.22 2.20 3.99 Molybdenum 0.24 0.26 0.14 0.13 0.23 Manganese 0.26 0.22 0.14 0.13 0.23 Silicon 0 0 0.11 0.12 0.01
3.2 Matching process parameters to test lab results
The correlation between the process parameters during the plasma-‐spraying process, of each torch, and the CAP content of the coating was analyzed. The process parameters chosen was electric current, voltage and power output. The scatter plots in Figures 3, 4 and 5 illustrate these parameters’ distribution of data. In Figure 3 the CAP content is plotted against electric current, in Figure 4 against voltage and in Figure 5 against power output.
Figure 3 -‐ Scatter plot of CAP vs. electric current in Torch 1 and 2 for each Booth in production line 2
20 22 24 26 28 30 32 34 36 336 337 338 339 340 341 342 343 344 CAP c on te nt [%]
Electric current [A]
CAP content vs. electric current
Booth 1 Torch 1 Booth 1 Torch 2 Booth 2 Torch 1 Booth 2 Torch 2
Figure 4 -‐ Scatter plot of CAP content vs. voltage in Torch 1 and 2 for each Booth in production line 2
Figure 5 -‐ Scatter plot of CAP content vs. power output in Torch 1 and 2 for each Booth in production line 2
20 22 24 26 28 30 32 34 36 35 36 37 38 39 40 CAP c on te nt [%] Voltage [V]
CAP content vs. voltage
Booth 1 Torch 1 Booth 1 Torch 2 Booth 2 Torch 1 Booth 2 Torch 2 20 22 24 26 28 30 32 34 36 11 11,5 12 12,5 13 13,5 14 CAP c on te nt [%] Power output [kW]
CAP content vs. power output
Booth 1 Torch 1 Booth 1 Torch 2 Booth 2 Torch 1 Booth 2 Torch 2
4 Discussion
In the test lab Excel document there is room for data from three positions from each lining tested, however there is only data from one position presented -‐ BDC (Figure 2) -‐ why the results in this report are based on these.
The tolerance range for the amount of pores is in between 1.25% and 4%. The software AxioVision SE64 used to determine the amount of pores and CAP was tested and the same sample and area was analyzed multiple times with different results each time. These test results differ with a margin of error of 1%. This makes it difficult to evaluate the results since this margin of error represents 36% of the tolerance range. Thus with the tolerance shifted to ranging in between 0.25% and 5%, it is wider than the ideal. It appears that the test results are dependent on the operator. When samples exhibit results below the required amount of pores or CAP in the test lab, the samples are analyzed over again until an approved result it obtained. The unqualified results are not being transferred into the test lab Excel document, and are therefore not traceable, why the process parameters cannot be interconnected with those. In the test lab Excel
document, data from 220 samples from the period chosen for this work was collected. It was matched to the process parameter logs with the data extraction program. It would be of interest to study the samples that did not fulfill the required amount of pores and CAP to search for a correlation to the process parameters. Since there is only data from approved samples it can be that, the unqualified data points would exhibit a pattern that is not to be seen in this analysis.
The composition of elements in the powder samples from different sites in the production line varies widely. Table 2 illustrates how the element composition of the five samples differs. Some of the cells in Table 2 are highlighted – these are the ones with minimum and maximum content of the element. The elements with the highest content and thus most impact on the result of the final quality and properties of the coating are iron, aluminum and oxygen (Fe and Al2O3). The three elements are compiled below
together with their largest difference in content. • Iron
Minimum 11.28 at. % (sample 4) Maximum 22.57 at. % (sample 1) Difference of 11.29 at. %.
• Aluminum
Minimum 22.94 at. % (sample 4) Maximum 33.89 at. % (sample 3) Difference of 10.95 at. %.
• Oxygen
Minimum 42.59 at. % (sample 1) Maximum 49.69 at. % (sample 4) Difference of 7.1 at. %.
While the iron content is higher in a sample, the amount of aluminum and oxygen in it is lower and vice versa. The element content of these display large variations, it differs as much as twice the amount in different sites. Moreover, there are smaller variations in some of the contents despite the samples being from the same barrel. Samples 1 and 2 are from the same barrel, however their oxygen content differs 2.21 at. % and the iron content 2.64 at. %. Even though the variations are rather small in comparison to their content, there should not be any at all. A possible cause of this is density variations in the powder. The molecules have different dimensions and atomic weights; naturally they react differently to vibrations that might occur during transport. The powder is fine enough not to layer up, however there might still be variations in composition inside the barrel.
The target value of the electric current is 340A. As seen in Figure 3 there are some differences in current between the Booths and Torches, yet all the data points are close to the target value. There is no correlation between the CAP content of collected data and the electric current within the tolerance range.
The target value of the voltage is 36V. As seen in Figure 4, all of the data points are randomly distributed in between 35V and 39V with slightly higher concentration towards the higher values. There is no correlation between the CAP content of collected data and the voltage within the tolerance range.
The power output has no target value given. However, as the power output depends on the electric current and voltage, there is no correlation between CAP content of collected data and the power output within the tolerance range, as seen in Figure 5.
5 Conclusion
The purpose was to investigate factors affecting the amount of ceramics in the plasma sprayed coating of cylinder linings at Scania. The analysis of data did lead to findings that show the powder is the most critical factor.
• The margins of error in the results using the software AxioVision SE64 are large in the context, which makes them questionable.
• The powder composition varies widely in between batches and sites in the production line, which is a possible cause for irregularities in the coating.
• The powder composition varies slightly inside a barrel, which is another possible cause for irregularities in the coating.
• There is no correlation found between the amount of ceramics in the coating and the chosen process parameters, which makes it difficult to prevent irregularities.
6 Acknowledgements
We would like to thank Scania and the people working there for the support throughout our bachelor thesis work. We would like to especially thank our supervisor Björn Lindbom and his colleagues Anders Kjelledal, Pelle Nilsson, David Björkman, Doctor Jessica Elfsberg for their expertise and support. We are also grateful to WenLi Long and Anders Tilliander at KTH for their contributions. Last but not least, we would like to express our gratitude to Professor Stefan Jonsson at KTH for specially designing the MATLAB program used in this work.
7 References
[1] SCANIA. (2015) Scania Group. [Online] Available from:
http://www.scania.com/scania-‐group. [Accessed: 23rd February 2015]
[2] IGANATOV, A. M. and RUKHADZE, A. A. Plasmas in nature, laboratory and technology, in CAPITELLI, M. and GORSE, C. Plasma Technology Fundamentals and Applications. New York; Plenum Press, 1992. pp.1-‐9.
[3] BOULOS, M. I. (1992) RF Induction Plasma spraying: State-‐of-‐the-‐art Review. Journal
of Thermal Spray Technology. [Online] 1 (3). P.33-‐40. Available from:
http://www.link.springer.com/article/10.1007/BF02657015. [Accessed: 30th January 2015].
[4] FANTASSI, S. et.al. (1993) Proceedig of the ISPC-‐11. Loughboough; International Organising Committee of ISPC 11, 1993. pp.1251.
[5] RILEY, A. et. al. (2008) Aviations Maintenance Technician Handbook. [Online] General. Available from https://www.faa.gov. [Accessed: 30th April 2015]. p.1.23.
[6] SWAPP, S. (2015) Scanning Electron Microscopy. [Online] Available from: http://serc.carleton.edu . [Accessed: 26th April 2015].
8 Appendices
8.1 Appendix 1
Figures 6 – 10 below illustrate the morphology element distributions from the SEM analysis in area A for each respective sample.
Figure 6 -‐ SEM analysis of the powder in Sample 1, Area A
Figure 8 -‐ SEM analysis of the powder in Sample 3, Area A
Figure 10 -‐ SEM analysis of the powder in Sample 5, Area A
8.2 Appendix 2
Tables 3 – 7 below illustrate the atomic element distributions from the EDS analysis for each respective sample over the three areas, A, B and C as well as their mean value.
Table 3 – Atomic distrbution of elements in Sample 1 from EDS analysis
Sample 1 Area
Element
A
[at. %] [at. %] B [at. %] C Mean value [at. %]
Iron 27.39 22.21 18.10 22.57 Aluminum 25.90 27.58 29.07 27.52 Oxygen 38.38 43.03 46.36 42.59 Zirconium 2.68 2.49 2.66 2.61 Chromium 5.09 4.14 3.41 4.21 Molybdenum 0.26 0.26 0.20 0.24 Manganese 0.30 0.28 0.20 0.26 Silicon 0 0 0 0
Table 4 -‐ Atomic distribution of elements in Sample 2 from EDS analysis
Sample 2 Area
Element
A
[at. %] [at. %] B [at. %] C Mean value [at. %]
Iron 21.22 18.52 20.06 19.93 Aluminum 27.84 27.93 28.08 27.95 Oxygen 43.65 46.40 44.36 44.80 Zirconium 2.88 3.22 3.14 3.08 Chromium 3.92 3.51 3.85 3.76 Molybdenum 0.26 0.24 0.28 0.26 Manganese 0.23 0.19 0.23 0.22 Silicon 0 0 0 0
Table 5 -‐ Atomic distribution of elements in Sample 3 from EDS analysis Sample 3 Area Element A
[at. %] [at. %] B [at. %] C Mean value [at. %]
Iron 11.57 11.25 12.24 11.96 Aluminum 34.85 33.66 33.16 33.89 Oxygen 48.74 49.99 49.40 49.38 Zirconium 2.24 2.54 2.52 2.43 Chromium 2.24 2.14 2.27 2.22 Molybdenum 0.13 0.14 0.14 0.14 Manganese 0.13 0.14 0.15 0.14 Silicon 0.09 0.13 0.11 0.11
Table 6 -‐ Atomic distribution of elements in Sample 4 from EDS analysis
Sample 4 Area
Element
A
[at. %] [at. %] B [at. %] C Mean value [at. %]
Iron 12.67 10.24 10.94 11.28 Aluminum 33.87 34.29 33.66 33.94 Oxygen 47.96 50.73 50.37 49.69 Zirconium 2.59 2.40 2.55 2.51 Chromium 2.49 1.98 2.12 2.20 Molybdenum 0.16 0.12 0.12 0.13 Manganese 0.14 0.12 0.14 0.13 Silicon 0.13 0.11 0.11 0.12
Table 7 -‐ Atomic distribution of elements in Sample 5 from EDS analysis Sample 5 Area Element A
[at. %] [at. %] B [at. %] C Mean value [at. %]
Iron 21.24 22.36 19.75 21.12 Aluminum 27.32 26.60 27.41 27.11 Oxygen 44.21 43.37 45.45 44.34 Zirconium 2.73 3.01 3.16 2.97 Chromium 4.01 4.14 3.82 3.99 Molybdenum 0.23 0.25 0.22 0.23 Manganese 0.24 0.27 0.18 0.23 Silicon 0.01 0.01 0 0.01