Note 42: The Influence of Pump Oil Purity on Roughing Pumps

INTRODUCTION

Pump oils contain a wide variety of volatile semi-volatile organics and the breakdown of these compounds can affect the life of a rough pump. This degradation of pump oils and the resulting effect on the life of vacuum rough pumps is a major concern to the users of mass spectrometer and other vacuum system instruments. Analytical techniques are needed to identify and quantitate volatile organic compounds (VOC's) present in oils. These techniques can be used not only to study the breakdown of vacuum pump oils but can also be incorporated into a quality control program to ensure product purity during manufacture. Sensitive techniques are needed to profile and identify VOC's in pump oils so that they can be used to ensure the life of the pump. Headspace GC analysis, cryofocusing techniques, and high-resolution GC have previously been used for the analysis of commercial oils. Static headspace techniques are limited in their detection and identification of the most volatile organic volatiles and have proven to be very insensitive for the less volatile organics as well as the semi-volatile organics.

The purpose of this investigation is to develop an analytical technique that could detect and identify a wide range of volatile and semi-volatile organic compounds in pump oils over time. For this study, volatile organic compounds are purged from oil samples utilizing a dynamic purge and trap technique (P&T) system followed by trapping on a Tenax® TA adsorbent resin trap. The adsorbent traps are ballistically heated and together with the carrier gas flow through the samples the volatiles are desorbed into the GC injection port and onto the front of the GC column for subsequent analysis by gas chromatography-mass spectrometry (TD-GC-MS). The volatile and semi-volatile organics present in the pump oils are quantified using a matrix spiked deuterated internal standard. The P&T technique permits the analysis of a wider range of both volatile and semi-volatile organic compounds and is more sensitive as compared to the static headspace technique.

Instrumentation

Samples were collected using a Scientific Instrument Services Purge and Trap System. This apparatus consists of a sparge gas inlet connected to a stainless steel purging needle that is inserted through an adaptor fitting into the oil sample in the bottom of the 10 ml purge and trap test tube. A dry purge gas inlet is located at a right angle to the sparge gas inlet at the top of the apparatus. The purpose of the dry purge is to reduce the water vapor condensation on the adsorbent trap. Opposite the dry purge inlet is the connector for the desorption tube. A glass-lined stainless steel (GLT) desorption tube containing the adsorbent resin is attached to this fitting for the trapping of the purged volatiles.

All experiments were conducted using a Scientific Instrument Services model TD-3 Short Path Thermal Desorption System accessory connected to the injection port of an HP 5890 Series II GC interfaced to an HP 5971 Mass Selective Detector (MSD). The mass spectrometer was operated in the electron impact mode (EI) and scanned from 35 to 550 daltons during the GC run for the total ion chromatogram.

The HP 5890 Series II GC contained a short 0.5 meter by 0.53 mm diameter fused silica precolumn attached to the injection port end of a J&W 30 meter x 0.25 mm i.d. DB-5MS capillary column containing a 0.25 micron film thickness. The GC injection port was set to 260 degrees C and a 10:1 split was used. The head of the column was maintained at -70 degrees C using an S.I.S. Cryotrap model 951 during the desorption and extraction process and then ballistically heated to 200 degrees C to release the volatiles after which the GC oven was temperature programmed from 35 degrees C (hold for 5 minutes) to 80 degrees C at 10 degrees C/min, then to 200 degrees C at 4 degrees C/min and finally to 260 degrees C at a rate of 10 degrees C/min.

Experimental

Fresh oil samples of Inland 45, Inland 19 and Invoil 20 pump oils were collected on January 20, 1996, followed by monthly collections of Inland 45 and Inland 19 pump oils from two continuously running Alcatel model UM2012 vacuum pumps. Monthly oil samples were taken for a period of 3 months on February 20, 1996, March 19, 1996 and April 16, 1996 and subsequently analyzed to determine the presence and changes of VOC's over time that would ultimately affect the life of the pump.

Samples from 2 different kinds of pump oil (Inland 45 & Inland 19) were removed from two continuously running Alcatel model UM2012 vacuum pumps over a three month time period and analyzed by a dynamic purge and trap technique to compare and quantitate the volatile organics present. An additional fresh sample of Invoil 20 pump oil was also examined for comparison. These three oils (Inland 19, Invoil 20 and Inland 45) represent the most widely used oils in vacuum pumps on mass spectrometers and related vacuum instruments in the scientific community. For quantification, 100 ng of a deuterated cymene internal standard was spiked into the adsorbent traps after the volatiles from the oil samples had been purged onto the adsorbent resin traps. No correction for extraction efficiency of recovery is achieved using this technique; however, it functions as a useful means to semi-quantitate the volatile and semi-volatile organics purged from the oil samples onto the adsorbent traps.

Two ml aliquots of oil were removed monthly from each vacuum pump and transferred into a 10 ml test tube and heated to 60 degrees C for 30 minutes. Samples were sparged with high purity helium at 20 ml/min (600 ml) with an additional 25 ml/min (750 ml) dry purge using an S.I.S. Purge and Trap System. Volatile analytes were gas extracted and carried to a preconditioned 4.0 mm i.d. glass-lined stainless steel desorption tube packed with 200 mg of Tenax TA. Once the samples were collected, they were spiked with 100 ng of d-cymene internal standard by injecting 1 ul of a 100 ng/ul of a d-14 cymene stock solution in methanol by syringe injection into the Tenax matrix, and then purged for an additional 5 minutes at 50 ml/min (250 ml) to remove the methanol. The desorption tube with sample and internal standard was then attached to the Short Path Thermal Desorption System and a syringe needle attached. The desorption tube was injected into the GC injection port at desorption block temperatures of 220 degrees C for 5 minutes at a flow rate of 12 ml per minute. The GC Injection port was maintained at 260 degrees and the GC Cryo-Trap at the front of the GC column was maintained at -70 degrees C during this desorption process. A 10:1 split was used during desorption to avoid overloading the GC capillary column. After desorption was complete, the GC Cryo-Trap was rapidly heated to 220 degrees C to release the volatiles and the GC was temperature programmed as described above. The GC peaks were detected via the HP MSD and identified using the Wiley NBS reference library to confirm the identification of each compound eluted.

Results and Discussion

The three pump oils Inland 19, Invoil 20 and Inland 45 are the most widely used oils for vacuum pumps for mass spectrometers and other scientific vacuum equipment. Inland 19 was the standard pump oil until about 8 years ago, when Invoil 20 was recommended by SIS for this purpose due to is lower vapor pressure and better vacuum capacities for the vacuum rough pumps. More recently Inland 45 has been recommended by both SIS and HP for use in vacuum pumps on mass spectrometers. This oil has an even better vacuum rating than the other two oils and will produce the optimum vacuum levels for mass spectrometers. These three oils are used by the majority of vacuum pump manufacturers (including Alcatel and Edwards) as well as by the majority of mass spectrometer manufacturers.

The first part of this study is to compare the three pump oils to verify why the Inland 45 has proven superior to the Invoil 20 which has proven superior to the Inland 19 vacuum pump oils. The comparison of these oils is shown in Figures 1, 2 and 3. The second part of this study demonstrates the degradation of the Inland 19 and Inland 45 pump oils after 1, 2 and 3 months of continuous use in an Alcatel vacuum pump. The inlet port of the vacuum pump is blanked off to prevent air or other contaminants from entering the vacuum pump. This mimics the continuous use to which these oils would be exposed to in a mass spectrometer system. However, it does not truly demonstrate the actual use in a mass spectrometer during which a large vacuum chamber is pumped and the injection of samples into the vacuum system which eventually are purged into the vacuum pump oil. These conditions would further degrade the pump oils at various rates depending on the instrument environment and the types of analytes injected into the mass spectrometer.

Over 100 volatile organics were identified in each of the oil samples analyzed (Fig 1, 2 and 3). The oil samples studied produced 50 or more volatile organics which were identified in addition to many more that were either too weak to identify or in which a good NBS library match was not achievable. The compounds identified are listed in Table I which also serves as an index to the total ion chromatograms for each of the pump oils analyzed.

The analysis of fresh Inland 19 pump oil is shown in Fig. 1. It was found to contain several straight chain hydrocarbons from decane through dodecane (peaks 53, 67 and 75) as well as numerous substituted benzenes (peaks 19, 29, 31, 36, 44, 47, 51, 56 and 61), aldehydes from pentanal through decanal (peaks 15, 24, 39, 54, 68 and 76) and ketones (peaks 5, 11, 21, 34, 72, and 82). A high concentration of the aromatic compound toluene (peak 19) was also present. Several siloxanes (peaks 26, 48 and 69) were also detected. Hydrocarbon chains shorter than C10 were not detected. The internal standard d14-cymene is peak 55 in all the chromatograms.

The total ion chromatogram of Inland 45 is shown in Fig. 2. It is quite evident that this sample contains significantly less volatiles than the Inland 19 pump oil shown in Fig 1. This explains the lower vapor pressure for Inland 45 and also the reason that better vacuum levels can be achieved with this oil. A number of hydrocarbons from hexane through dodecane (peaks 6, 13, 23, 38, 53, 67 and 75) were detected as well as a large number of aromatics (peaks 19, 31, 36, 44, 47, 51, 59, 60 and 65). The siloxanes hexamethyl-cyclotrisiloxane, octamethyl-cyclotetrasiloxane and decamethyl-cyclopentasiloxane which are related to cleaning and lubricating compounds were detected in both the fresh Inland 19 and Inland 45 pump oils (Figs. 1&2). Absent from the Inland 45 oil are the aldehydes and ketones which were so prevalent in the Inland 19 oil.

The total ion chromatogram of Invoil 20 is shown in Fig. 3. It is quite evident that this sample contains significantly less volatiles than the Inland 19 pump oil shown in Fig 1. This explains the lower vapor pressure for Inland 45 and also the reason that better vacuum levels can be achieved with this oil. However, the Invoil 20 contained several low boiling aldehydes (peaks 5, 11 and 15) which were not detected in the Inland 45. This demonstrates the advantage of Inland 45 as the optimum pump oil for vacuum systems. The straight chain hydrocarbons were absent from the Invoil 20 oil with the exception of dodecane (peak 75). Only two aromatics compounds toluene and xylene (peaks 19 and 31) were detected; however, a large number of aldehydes from pentanal through decanal (peaks 15, 24, 39, 54, 68 and 76) and ketones (peaks 5, 11, 18, 21, 22, 34, 37 and 42) were identified. Of the siloxanes only hexamethyl-cyclotrisiloxane was detected in Invoil 20.

After 1 month, the concentrations of VOC's in the Inland 19 pump oil slightly increased with the exception of the aromatic compound toluene which was significantly reduced with time (Fig. 4). The compounds 5-ethyldihydro-2(3)-furanone (peak 62), dihydro-5-pentyl-2(3H)-furanone (peak 79) and 5-hexyldihydro-2(3H)-furanone (peak 88) were also identified at this time. Generally, the concentrations of VOC's in the Inland 19 pump oil remained approximately the same after 2 months with slight increases in hydrocarbons and siloxanes and continued decrease in the aromatic compound toluene (Fig. 5). After 3 months, there were significant increases in the levels of the aldehydes butanal, pentanal, hexanal, heptanal, octanal, nonanal and decanal (peaks 4, 15, 24, 39, 54, 68 and 76) (Fig. 6) as compared to the 2 month sample of Inland 19 pump oil. The increased concentration of these aldehydes is significant after three months of continued use. These aldehydes can be expected to be converted to acids with continued use of the vacuum pumps. This decrease of pH will contribute to increased vacuum pump wear. This demonstrates the necessity of changing vacuum pump oils after three months of continued use in order to maximize vacuum pump life.

The volatiles in Inland 45 pump oil increased dramatically after only 1 month of continued use (Fig. 7). Inland 45 pump oil exhibited significant increases in the propanoic, butanoic, pentanoic, hexanoic and heptanoic acids (peaks 17, 27, 40, 57 and 64) as well as aldehydes, ketones and furanone derivatives after the initial month. Significant amounts of dihydro-5-methyl-2(3H)-furanone (peak 43), dihydro-5-propyl-2-(3H)-furanone (peak 70) and 5-butyldihydro-2(3H)-furanone (peak 78) were also detected at this time (Fig. 7). After 2 months of continued use, the concentrations of VOC's in the Inland 45 pump oil were significantly reduced with the exception of the siloxanes (Fig. 8). The same decreasing trend continued in the Inland 45 pump oil at 3 months with continued reduction in the concentrations of volatile organics (Fig. 9). It appears that the most significant changes in pump oil occurs after its initial use with gradual changes in pump oil purity after continued use.

Conclusion

The Short Path Thermal Desorption System used in conjunction with a dynamic purge and trap technique permits the identification and quantification of trace levels of volatile organics in pump oils. The method previously described permits both the qualitative and quantitative analysis of pump oils to identify the volatile organics, contaminants and off-odor compounds present. This technique has proven effective in detecting and identifying a larger number of organic compounds at concentrations lower than was previously obtainable via other analysis techniques. It also permits the analysis of a wider range of both volatile and semi-volatile organic compounds and is more sensitive as compared to the static headspace technique. It represents a tremendous improvement over the time-consuming solvent extraction techniques normally used in the laboratory. This technique can be easily incorporated into a troubleshooting technique to detect problems in a wide variety of commercial products, to compare competing manufacturers products, as well as a quality control program.

Inland 45 has been shown to be the optimum pump oil for rough vacuum pumps used in mass spectrometers and other vacuum systems in scientific instruments. It contains the lowest level of volatile organics than either of the two popular vacuum pump oils studied. It also contained low levels of aldehydes and ketones.

Manufactures of vacuum pumps recommend changing pump oils after every 3 months of pump use. The large chain hydrocarbons and other large synthetics present in the oils break down with time producing a large number of lower boiling compounds. These compounds all raise the vapor pressure of the oil thereby reducing its capacity to achieve low vacuum levels. As shown above, aldehydes and ketones are generated from the degradation of vacuum pump oils with continued use. With time, these aldehydes can be converted to acids which will rapidly enhance the wear inside the vacuum pump thereby reducing its life. This confirms the need to change vacuum pump oils on a regular basis.

This study is being continued. The two pumps with the Inland 19 and Invoil 45 are continuing to operate and additional oil samples will be removed and tested to further complete this study.

Index for Figures 1 through 9

1.

2-propanone

2.

2-methyl-1,3-butadiene

3.

1-hexene

4.

butanal

5.

2-butanone

6.

hexane

7.

acetic acid

8.

1,3-hexadien-5-yne

9.

1,5-hexadien-3-yne

10.

benzene

11.

2-pentanone

12.

1,2-dimethyl-cyclopentane

13.

heptane

14.

1-heptene

15.

pentanal

16.

methyl-cyclohexane

17.

propanoic acid

18.

4-methyl-2-pentanone

19.

toluene

20.

1-octene

21.

2-hexanone

22.

cyclopentanone

23.

octane

24.

hexanal

25.

tetrachloro-ethene

26.

hexamethyl-cyclotrisiloxane

27.

butanoic acid

28.

3-methyl-cyclopentanone

29.

1,3-dimethyl-benzene

30.

ethyl-benzene

31.

xylene

32.

1,2-dimethyl-benzene

33.

4-methyl-hexanal

34.

2-heptanone

35.

1,3-dimethyl-benzene

36.

styrene

37.

cyclohexanone

38.

nonane

39.

heptanal

40.

pentanoic acid

41.

alpha-pinene

42.

6-methyl-2-heptanone

43.

dihydro-5-methyl-2(3H)-furanone

44.

1-ethyl-2-methyl-benzene

45.

1-ethyl-4-methyl-benzene

46.

benzaldehyde

47.

1,3,5-trimethyl-benzene

48.

octamethyl-cyclotetrasiloxane

49.

2-octano

50.

1-decene

51.

1,2,4-trimethyl-benzene

52.

hexanoic acid

53.

decane

54.

octanal

55.

d-cymene

56.

1,2,3-trimethyl-benzene

57.

hexanoic acid

58.

limonene

59.

1-methyl-3-propyl-benzene

60.

1-ethyl-3,5-dimethyl-benzene

61.

1,2,4,5-tetramethyl-benzene

62.

5-ethyldihydro-2(3H)-furanone

63.

1-phenyl-ethanone

64.

heptanoic acid

65.

1-ethyl-2,3-dimethyl-benzene

66.

2-nonanone

67.

undecane

68.

nonanal

69.

decamethyl-cyclopentasiloxane

70.

dihydro-5-propyl-2(3H)-furanone

71.

6-dodecanone

72.

2-decanone

73.

1-tetradecanol

74.

naphthalene

75.

dodecane

76.

decanal

77.

1,2-benzisothiazole

78.

5-butyldihydro-2(3H)-furanone

79.

dihydro-5-pentyl-2(3H)-furanone

80.

tetrahydro-6-propyl-2H-pyran-2-one

81.

3-undecanone

82.

2-undecanone

83.

undecane

84.

heneicosane

85.

tridecane

86.

1-octadecanol

87.

dodecanal

88.

5-hexyldihydro-2(3H)-furanone

89.

2-ethenyl-naphthalene

90.

2-dodecanone

91.

tetradecane

92.

6,10-dimethyl-2-undecanone

93.

5-heptyldihydro-2(3H)-furanone

94.

Pentadecane

95.

diphenyl-methanone