A DIFFRACTOMETER FOR X-RAY STUDIES OF BOND COATS BENEATH THERMAL BARRIER COATINGS

K. Vaidyanathan1, D. Pease2, E. Jordan3, H. Canistraro4, M. Gell1, and T. Watkins5

1 Department of Metallurgy & Materials Eng., University of Connecticut

2 Department of Physics, University of Connecticut

3 Department of Mechanical Engineering, University of Connecticut

4 Ward College of Technology, University of Hartford

5 High Temperature Materials Laboratory, Oak Ridge National Laboratory


 
 

ABSTRACT

Commercially important thermal barrier coating (TBCs) assemblies contain a protective vapor deposited yttria stabilized zirconia ceramic layer, over a nickel-aluminum bond coat and a nickel-based superalloy. The thinner high performance coatings typically fail due to mismatch strain at the ceramic/metal interface. We have constructed a laboratory diffractometer with which we have measured diffraction lines of bond coats lying underneath a 125-µm thick thermal barrier coating, without using a rotating anode x-ray source. The diffractometer utilizes a niobium target x-ray tube in Seeman-Bohlin geometry with fixed sample position, where a very large Johannson bent crystal is used for gathering and focusing x-rays to the sample.

1. INTRODUCTION

Thermal barrier coatings (TBC) have been used to protect structural turbine components employed in hot section areas, for the last two decades. There is interest in applying TBCs to even higher temperature environments than is now standard, as well as in developing experimental tests which aid in understanding and predicting failure [1, 2]. The ability to detect alloy phases that form at the bond coat-TBC interface that could affect the coating performance, as a function of thermal exposure would therefore is of use [3]. Although for many purposes the laser fluorescence method has been chosen as the method of choice for stress measurements in TBC assemblies, this method probes only the thermally grown oxide (TGO), which is predominantly a -Al2O3 [4] and does not detect phases or lattice constants. One therefore is motivated to develop an x-ray method for probing stresses and structures within the bond coat at the bond coat - TBC interface.

Bond coat materials consist of various alloy phases. The main elemental ingredients are nickel, other first row transition metals(TM), aluminum, and in some instances platinum or traces of yttrium. The TBC layer deposited on top of the bond coat consists of 7-8 wt. % Y2O3 stabilized ZrO2 (YSZ). The most serious problem these composite materials present for obtaining x-ray diffraction from the bond coat is the x-ray absorption of the TBC. Polishing off the TBC layer is not always practical for tests involving repeated thermal cycling, since thinning the TBC after a particular cycle could drastically influence effects of future thermal cycling on the bond coat. The ability to perform x-ray determinations of stress, phase, and lattice constant of the bond coat itself, underneath the TBC, would therefore be of interest.

TBC protective layers are typically of order 125 m m thick if deposited be electron beam vapor deposition (EB-PVD) and in excess of 200 m m if plasma sprayed. Obtaining meaningful x-ray diffraction patterns from the bond coat underlying these thick layers is a formidable challenge. One approach to this experimental problem would be to use synchrotron based apparatus of some sort. Such an approach precludes development of a general testing method, and we therefore discuss the development of a laboratory diffractometer made especially for diffraction experiments on bond coats underlying TBCs.

2. EXPERIMENTAL PROCEDURE

2. 1. Choice of Incident Radiation

The absorption of x-ray energies by thick TBCs is so great as to render impossible diffraction studies of underlying bond coats, if x-ray radiation from anodes of copper or other first row TMs is utilized. Of the standard x-ray energies used for laboratory diffraction, characteristic x-ray flux from a Mo anode is the only radiation possibly suitable for such an application. In addition to the absorption problem, the TBCs add to the difficulty of bond coat diffraction studies because of interference from YSZ TBC diffraction peaks. The EB-PVD TBC microstructure presents somewhat less of an experimental obstacle than plasma sprayed composites, not only because of the thinner TBC layer (~125 µm versus ~ 200 µm) but because of the pronounced preferred orientation of the EB-PVD TBC columns [2]. This preferred orientation results in less interference between TBC diffraction peaks and underlying weak bond coat peaks. One does not have this advantage with the randomly oriented plasma sprayed TBC overlayers. It has been found that there is a window between the EB-PVD TBC diffraction peaks, which corresponds to bond coat peaks of usable intensity, in the region of reciprocal space corresponding to twice Bragg angle in the 50 to 60 degree range, using a Mo or Nb x-ray anode sources. We therefore now highlight this range of Bragg angle and incident x-ray energy.

For present purposes, a Mo anode is the most advantageous of the standard x-ray tube anodes. However, there is an advantage of using a Nb anode instead. The Nb Ka energy is just less than the K shell binding energy of the Y dopant, whereas the Mo Ka energy is maximally efficient at exciting Y K shell electrons. Although Y is only 7 to 8 % of the TBC composition, the reduced absorption of Nb relative to Mo radiation becomes quite significant for the large TBC thickness. In addition, for reasons discussed below, we prefer not to use a graphite monochromator as part of the detector assembly but use instead a large Johannson bent crystal to condition the beam impinging on our samples. We then obtain the expected large Y fluorescence signal if Mo Ka radiation is used, but not if Nb Ka radiation is used. Any fluorescence excited from the TM elements in the bond coat is completely absorbed in the TBC. We have obtained a specially manufactured Nb anode sealed diffraction tube that is compatible with a standard tube tower, for less expense than a peltier cooled energy dispersive detector. We operate this Nb tube at approximately 55% of the power rating of a corresponding Mo diffraction tube. Therefore, the intensity of the diffuse scattering from the TBC which interferes with the weak bond coat peaks, is diminished along with the bond coat peak intensity by about 55%. Because of the enhanced bond coat diffraction relative to TBC diffuse scatter intensity, the tube with Nb anode is advantageous over a tube with a Mo anode. We use a standard, low noise scintillation counter to detect the diffracted x-rays.

For the standard Bragg-Brentano diffraction condition, in which the incident and exit angle are equal to each other and to the Bragg angle, one has the shortest total path through the TBC to the bond coat and back to the detector. In Table 1 below are tabulated the calculated attenuation through the TBC corresponding to incident x-rays, for various incident angles (chi), thickness of TBC, and x-ray anode materials. For simplicity we assume equal angles for both incident and diffracted beams. The 20 µm TBC at 2 degree chi case would correspond to a glancing angle study of bond coat diffraction under a partially polished off but structurally intact TBC layer. One concludes that such studies are impossible using Cu radiation, and that Nb is preferable to Mo.

Table 1: Attenuation factor for different x-ray sources.

TBC Thickness (m m)
Chi Angle
X-ray

Anode Type

Attenuation

Factor

100
30°
Cu
3.52 ´ 109
Mo
31
Nb
19
100
15°
Cu
3.90 ´ 109
Mo
93
Nb
49
200
30°
Cu
1.24 ´ 1019
Mo
967
Nb
370
20
2°
Cu
1.70 ´ 1014
Mo
167
Nb
82

 

2. 2 Seeman-Bohlin Geometry: Advantages for the Present Application

Although the usual Bragg-Brentano (BB) configuration has many advantages of convenience and relative ease of alignment, for measurements restricted to the back reflection region, where one wishes to vary the incident angle, there are advantages to the Seeman-Bohlin (SB) arrangement. The schematic of the SB geometry is shown in Figure 1. An excellent review of the relative merits of these two diffractometer configurations is given by King et al [5]. At the outset we point out that the optics developed by Schuster and Gobel [6], whereby a parabolic multilayer yields a parallel x-ray beam from a line source [6], is not presently commercially available for short wavelengths such as is obtained from a Mo or Nb anode. Were a parabolic focusing mirror available for Mo radiation, one would not need to depend on the parafocusing condition. However, in the absence of such a device, one can make use of the fact that in the Seemann-Bohlin arrangement parafocusing can be satisfied for incident and exit angles, measured relative to the sample surface, that are unequal. This feature enables stress measurements to be performed without sacrificing either intensity or resolution. In addition, King et al [5] have pointed out that the source aperture of the SB diffractometer is significantly smaller than that of the BB diffractometer in the back reflection region, so that for the same intensity the SB diffractometer is capable of greater angular resolutions [5]. We are unable to utilize here another potential advantage of the SB geometry, which is that one can replace parafocusing with perfect focusing in the diffraction plane if the sample can be formed to the SB focusing circle [7]. Unfortunately, the TBC samples we test are inherently flat. The effect of divergence out of the diffraction plane is lessened in the back reflection region, and therefore we use a "loose" soller slit arrangement with a spacing to plate thickness ratio of about five to one. We use two rotation stages of different rotational radii to maintain the focus of the silicon crystal, the sample, and the detector axis on the same circle.

2. 3. Johannson Focusing Monochromator

Bent focusing crystals have been placed between the x-ray source and diffractometer for several reasons. Asymmetric focusing crystals can have the advantage of maintaining a short distance and enhanced acceptance angle relative to the x-ray source, while the distance from the center of the monochromator to the focus may be lengthened to accommodate experimental apparatus of large radius. Such monochromators are often designed to have sufficient resolution so that only the Ka1 line of the Ka1,2 doublet pair is transmitted. This type of monochromator has been used in combination with a SB diffractometer in a commercial device for studying surfaces by glancing-angle diffraction. Our focusing monochromator is a much larger variant of the commercial device. We use a symmetric Johannson ground and bent silicon (111) monochromator that is 12 cm in length and 4 cm high. The crystal grinding radius of 1 meter is chosen so as to produce enough mosaicity to completely accept the entire Nb Ka1 peak and remove most, but not all of the Ka2 peak. The relationship between brightness and bending radius was based on theoretical results of Suortti et al [8], which show that a reflectivity of between 50% and 70% can be obtained over the angular width corresponding to the Nb Ka1 line. A previous application of this crystal is discussed by Canistraro et al [9], and made use of the fact that the focused image is quite sharp (a line source ~ 400 m m in width). The Si(111) monochromator removes the Nb Kb line and also Bremstrahlung except for the small band of continuum radiation lying in the energy range encompassed by the Nb Ka1 peak.

3. EXPERIMENTAL RESULTS

Figure 2 illustrates on top a diffraction scan taken with a state of art commercial diffractometer on a flat TBC specimen consisting of ~ 125 m m TBC, overlying a bond coat which largely consists of a b phase Ni(Pt,Al) alloy. The Bragg-Brentano mode is chosen, and a molybdenum anode x-ray tube is used. The tube is not a rotating anode tube and is operated at settings of 45 volts and 40 mA beam current. The diffractometer is operated with a slit of 0.25 mm between the sample and the detector. Below in Figure 2, is illustrated a diffraction scan taken under the same conditions, on a sample having a similar history, except that now the TBC layer is polished down to a thickness of 10 m m. The doublet peaks indicated by the symbol T come from the TBC and are reduced in intensity in the polished specimen. The doublet peaks indicated by B come from the bond coat and are clearly visible in the polished specimen but completely invisible in the scan of the unpolished TBC specimen. We attempted to observe the bond coat peak doublet through the unpolished TBC by step scanning this angular region over a time period of hours, but were unable to detect the bond coat peak.

Figure 3 shows a scan on a similar unpolished specimen using our spectrometer. We used a chi angle of 28° so as to approach the condition that the angle of incidence equals the angle of reflection for the bond coat peak we are attempting to observe. The Nb tube was operated at 50 kV and 20 mA. The slit positioned on the focusing circle, between the sample and the detector, had a width of 0.3 mm. In Figure 3 the central member of a "triplet" TBC peak (labeled TT) is centered on twice Bragg angle 61° for the specialized diffractometer data, whereas the triplet maximum is centered on about 59.5° for the data taken with a commercial diffractometer. The displacement to higher angle for the data shown on Figure 3 is due to the use of Nb as opposed to Mo radiation for the anode material. The intensity of peak TT is about ten times as much using our diffractometer as obtained using the commercial instrument (peak T). The full width at half maximum (FWHM) of this peak is about 1.5 times the FWHM for the corresponding peak observed using Bragg-Brentano instrument as shown in Fig. 1. The peak to the right of the central TT maximum is the Ka 2 portion of the K doublet. Its intensity is diminished relative to the central peak maximum more in Figure 3 than in Figure 2. This observation is consistent with diminution of the Ka 2 peak by the silicon monochromator for the Figure 3 data.

The TBC doublet peak centered at about 53° is not as enhanced in intensity for the data in Figure 3, relative to the corresponding peak at slightly greater than 52° in Figure 2, as is peak TT relative to peak T (a factor of ~ 2.6 instead of ~ 10). There can be differences in preferred orientation of the TBC between specimens, and in addition there is the complication that a bond coat doublet appears in the region of the TBC doublet for the polished specimen in Figure 2. The diffuse background scattering intensity for Figure 3 is about five times that observed in Figure 2. The most significant observation, however, is that in Figure 3 one observes a weak doublet at somewhat less than 56° corresponding to the polished specimen bond coat doublet observed in Figure 2 at about 54.5°. We refer to this doublet as peak BB. We are clearly observing a bond coat diffraction peak through the unpolished TBC barrier, without using either a synchrotron source or rotating anode x-ray tube.

4. SUMMARY AND CONCLUSIONS

The diffuse scatter intensity with our SB diffractometer is five times that of the BB diffractometer, despite the fact that our Nb x-ray tube is operated at about half the power of the Mo tube used with the BB diffractometer. This implies that an order of magnitude intensity gain is possible using our arrangement relative to the BB diffractometer. If large area powder samples were used, which could be bent to coincide with the focusing circle, the intensity and resolution could be further enhanced. One could monochromatize the Bremstrahlung with our large bent crystal, use a rotating anode tube and perform diffraction studies using continuum radiation. The resulting intensity would be only an order of magnitude less than that obtained with a stationary anode tube and BB geometry using Ka characteristic radiation [10]. We therefore suggest the possibility of a laboratory facility for diffraction studies on concentrated samples using anomalous scattering, making use of x-ray energies tunable between elemental absorption edges.

References

  1. A.G. Evans, D.R. Mumm, J.W. Hutchinson, G.H. Meier, and F.S. Pettit, Prog. Mat. Sci., 46, (2001), 505
  2. M. Gell, K. Vaidyanathan, B. Barber, J. Cheng, and E.H. Jordan, Met. Trans., 30A, (1999), 427.
  3. M. Gell, E. Jordan, K. Vaidyanathan, K. McCarron, B. Barber, Y.H. Sohn, and V.K. Tolpygo, Surf. Coat. Tech., 120-121, (1999), 53.
  4. Q. Ma and D. R. Clarke, J. Am. Ceram. Soc. 78, 1994, 298
  5. H.W. King, C. J. Gillman, and F.G. Huggins, Adv. X-ray Anal., 13, (1970), 550
  6. M. Schuster and H. Gobel, Adv. X-ray Anal., 39 , (1997), 57.
  7. H.P. Klug and L.E. Alexander, X-ray Diffraction Procedures, John Wiley and Sons, New York, 1954
  8. P. Suortti, P. Pattison, and W. Weyrich, J. Appl. Cryst., 19, (1986), 336.
  9. H. Canistraro, E. Jordan, and D. Pease, Rev. Sci. Instr., 69, (1998), 452.
  10. K. Kim, in X-ray Data Booklet, edited by A. Thompson and D. Vaughan, Lawrence Berkley Laboratory, University of California, Berkley, California, 2001, pp. 2-15.

Figure 1: Schematic of the diffraction geometry in the Seeman-Bohlin diffractomer. The sample may be placed closer to the focus of the bent crystal than to the focus of the x-ray tube, therefore increasing the solid angle for accepting x-rays. There is no divergence slit between the crystal focus and the sample.
 
Figure 2: Results from Bragg-Brentano diffractometer, polished versus full-thickness TBC specimen, using the Mo Ka radiation.

 
 
Figure 3: Results from Seeman-Bohlin diffractometer, full-thickness TBC specimen, using the Nb Ka radiation. The bond coat peak (BB) can be seen through the full thickness Notice also, how the diffraction peaks from Ka2 can be suppressed relative to Ka1.