Effect of fretting on electroplated Pd-Ni contacts

Effect of fretting on electroplated Pd-Ni contacts

339 Wear, 162-164 (1993) 339-346 Effect of fretting on electroplated Hideaki Pd-Ni contacts Murata Yazaki Corporation, 1500 Msyuk.u, Susono, Shiz...

814KB Sizes 3 Downloads 43 Views

339

Wear, 162-164 (1993) 339-346

Effect of fretting on electroplated Hideaki

Pd-Ni contacts

Murata

Yazaki Corporation, 1500 Msyuk.u, Susono, Shizuoka 410-11 (Japan)

Yasuo

Imada,

Fumihiro

Honda

and Koichi

Nakajima

Toyota Technological Institute, 2 Hisakata, Tempaku, Nagoya 448 (Japan)

Abstract Surface observations of the composition and chemical state were carried out to understand the fretting behaviour of an electroplated Pd-2Chvt.%Ni contact compared with that of an electroplated Pd contact. The experiments were conducted in ambient room air and in the relevant organic vapour. By fretting in ambient room air, the Pd-Ni contact acquired a high contact resistance because of oxidation, but with an Au-flash plating 0.1 pm thick

on it, it remained stable and with a low contact resistance. This is attributed to the solid lubricant effect because of the soft, thin layer of gold on the much harder Pd-Ni. By fretting in the environment containing toluene vapour, organic deposits (frictional polymers) were found on the Pd-Ni contact surface as well as on the Pd contact. Investigation of a possible correlation between the quantities of C and 0 on the contact surface detected by electron probe microanalysis revealed the existence of a small amount of organic deposits. At a low toluene concentration, some metal oxides were produced during the formation of organic products, and the correlation coefficient of C and 0 was reduced. By Fourier transform IR spectroscopy, the organic products were observed to include aryl or unsaturated esters. Organic deposits on the Pd-Ni contact functioned as a lubricant and inhibited the oxidation of the metal surface. This inhibitory effect was confirmed by characteristic X-ray spectrum analysis. In contrast, the deposits on the Pd contact did not act sufficiently as a lubricant, resulting in an increased contact resistance. Therefore, the Pd-Ni contact has some advantages over the Pd contact in the environment containing toluene vapour.

1. Introduction Low power separable electronic connectors which require high reliability have numerous practical applications in industry. For example, an automobile incorporates some reliable connectors between systems that are important to improve facility functions, control procedures and emergency responses, e.g. electronically controlled fuel injection, air bag systems, etc. Most of those connectors have stationary contact. Fretting, however, can occur in separable electronic connectors owing to differential thermal expansion and contraction of members to which mating contacts are attached and/or owing to external vibration. Fretting is a phenomenon which in tribology refers to a situation where two surfaces in contact are subjected to an oscillatory relative motion of low amplitude [l]. By fretting in air, corrosion proceeds at a high rate, i.e. fretting corrodes at the contact surface. The corrosion layers formed often insulate and eventually isolate the surface electrically with an increase in the contact

0043-1648/93/$6.00

resistance. Contact materials which rapidly form oxides or other corrosion products are most prone to cause this phenomenon. The preferred contact material for separable connectors is considered to be Au, because it does not produce an insulating film in virtually any environment and lasts as a contact for many years with high reliability. However, Au is rather expensive and the cost fluctuates. This is one reason for much research into lower cost materials. Pt-group metals are known to be a substitute for Au [2], and Pd has been used for some connector contacts that require high reliability. However, Pd causes some problems. By fretting in environments containing organic vapours, organic products (frictional polymers) are formed on the Pd contact surface. The organic product deposits in and around the contact area, and leads to a high contact resistance and eventually to contact failure. Hermance and Egan [3] elucidated this phenomenon in detail and Crossland et al. [4] found that plastic

0 1993 - Elsevier Sequoia. All rights reserved

H. Murata et al. I Fretting effects on electroplated contacts

340

materials used around connectors were a major source of organic vapours. Numerous reports have been published on frictional polymers [2-6], but some do not distinguish frictional polymer formation from fretting corrosion as a cause of high contact resistance, in which the analysis of wear scars and wear debris is not sufficient [5]. In the case of an electroplated Pd contact, Pd absorbs much H during plating, and sometimes has many cracks in the plated layer because of hydrogen embrittlement. The corrosion of underlayers or a substrate is often attributed to these cracks. In this study, electroplated Pd-Ni alloy was chosen as a substitute for Au. Surface observations of the composition and chemical state were carried out to understand the fretting behaviour of an electroplated Pd-2Owt.%Ni contact in comparison with that of an electroplated Pd contact. Experiments were conducted both in ambient room air and in environments containing relatively high concentrations of organic vapours, in order to distinguish the fretting corrosion and the frictional polymer formation as causes of high contact resistance.

2. Experimental methods 2.1. Apparatus Figure l(a) shows schematically the experimental apparatus. The rotary motion of the stepping motor was transformed to the reciprocal movement of low amplitude through a highly accurate micrometre-scale /Balancing

Arm

1 ,,Load

of-

(a) Fretting

apparatus

Flow Meter with Needle Valve

+ Fretting Apparatus

I Dry ‘Air Toluene,

,/ Plate

Specimen

k

(b) Specimens

and holders

Fig. 1. Schematic representations of (a) the fretting and (b) the specimens and holders.

apparatus

etc.

Fig. 2. Schematic diagram of the system to control the concentration of organic vapour.

screw, and the fretting was reproduced between a stationary pin specimen and a plate specimen fixed to a sliding table. The same material was used for the pin and the mating plate, and fretted throughout this work. Dead weight was loaded on the pin specimen. Both specimens and holders are shown schematically in Fig. l(b). The pin specimen had a hemispherical head with a radius of curvature of 1.4 mm. The stepping motor was controlled by bipolar integrated circuits. The slider table and the plate specimen moved 5 pm by one pulsed signal, and the pulse rate was convertible from 50 to 50 000 s-l. The contact resistance of the two mating contacts was monitored with a commercial digital micro-ohm meter, using a four-wire measurement method and applying an a.c. current to reduce the contact resistance owing to the connection of the measurement wires and the effect of the thermoelectromotive force. The currents for the measurement were less than 5 mA and the open circuit potential was less than 20 mV. During the measurement, the fretting motion was not interrupted in order to understand the exact effects of the fretting on the contact resistance. Contact resistance data were collected by a 16-bit personal computer through GP-IB and stored on a floppy disk for subsequent statistical analysis. The sampling frequency could be set as desired by software. The simple system to control the concentration of organic vapour is shown in Fig. 2. The total flow rate of dry air was 0.002 m3 min-‘. For the case of toluene, the concentration could be changed from 0.2% to 9.0% (by volume) by this system. 2.2.

_- -. f.s??q

-\

Materials and testing

The basic specimen for testing (i.e. electroplated Pd-2Owt;%Ni on a brass substrate 2.0 pm thick) displayed a high hardness, low friction coefficient and no cracks in the plated layer. The main components of the plating bath were chlorides of Pd, Ni and NH,. An electroplated Pd-Ni specimen with a layer 0.1 pm thick of electroplated Au (Au-flashed Pd-Ni) was also prepared, to compare with Pd-Ni, Pd, Ni, Au and Auflashed Pd. Each specimen was electroplated on the

341

H. Murata et al. I Fretting effects on electmplated contacts TABLE

1. Hardnesses

of the specimens

Plating

Pd-Ni

Pd

Ni

AU

=OSN

464

238

321

152

brass substrate. The thickness of the Pd, Ni and Au plating was 2.0 pm and that of the Au-flash plating was 0.1 pm. The Knoop hardnesses of these specimens are indicated in Table 1. The specimens were washed ultrasonically in a reagent-grade acetone for 1 min. The fretting amplitudes ranged from 10 to 200 pm and the fretting frequencies were adjusted depending on the amplitude (i.e. the span of relative displacement in fretting) owing to the characte~stics of the motor and its controller. When the amplitude was 20 pm, the frequencies were in the range of 7-20 Hz. The loads applied ranged from 0.25 to 5.0 N. The results of the 0.5 N test are presented in this report as representative, because the dependence on the applied load was virtually implicit in this load range. Testing was iterated in the range of 104-105 cycles. Experiments were conducted in ambient room air and in environments containing relatively high concentrations of toluene. The fretted contact surfaces were analysed using electron probe X-ray microanalysis (EPMA) and microbeam Fourier transform IR (FTIR) spectrometry.

The effects of the parameters tested for Pd-Ni were compared with those for some other contact materials at the conditions 15 pm, 9.6 Hz and 0.5 N. Figure 4 shows the contact resistance of some contacts as a function of the number of cycles. Though, the Ni content was only 20 wt.%, the Pd-Ni contact acquired a high contact resistance, similar to the Ni contact, owing to oxidation, However, the Pd-Ni contact with an Au layer 0.1 pm thick (Au-flashed Pd-Ni) remained stable and evidently produced a low contact resistance, similar to the Au contact. Wear losses of these contacts are shown in Fig. 5. Wear loss was calculated from the width of the contact area and the radius of curvature of a pin specimen. The wear loss of the Au-flashed Pd-Ni contact was revealed to be the smallest of all the materials tested. The low contact resistance and the small wear loss of the Au-flashed Pd-Ni contact are attributed to the solid lubricant effect because of the softer, thin Au layer on the much harder Pd-Ni. The fundamentals of thin metallic film lubrication were elucidated by Bowden and Tabor [7] who studied soft Fb and In on a hard steel. Antler and Feder [S] later elucidated the same effect obtained for a thin Au layer on a Pd contact, showing that the harder substrate bore the load, while 3

Pi’

Pd-Ni

1

3. Results and discussion 3.1. Fretting in ambient room air Initially, experiments were conducted in order to understand the effects of test parameters that included fretting amplitude, fretting frequency, load and number of cycles. Figure 3 presents the effect of the fretting amplitude, in which the results of quantitative analysis made by EPMA are shown. The chemical composition in weight per cent of 0 and C was measured at 50 points on a sample. Fifty sets of data were averaged and plotted as a function of the fretting amplitude. In Fig. 3, values of abscissa show the fretting amplitude set up by the motor controller. In the stroke range from about 15 to 40 pm, large amounts of 0 and C were detected from worn surfaces. This indicates that fretting corrosion and frictional polymer formation are considered to occur actively in this stroke range. When the stroke is less than 15 grn, the relative displa~ment between the two mated contacts may be insufficient for fretting corrosion. However, when the stroke is more than 40 pm, wear debris and corrosion products formed at the interface may be easily removed because of the longer relative displacement.

> 0

i

0

‘a,

i

01iJ

I

20

40 Stroke

60

/ 80

100

(!Jm)

Fig. 3. Quantities of 0 and C detected on the worn surfaces by EPMA vs. the fretting stroke (in ambient air).

Number

of Cycles

Fig. 4. Variations of the contact resistances of some contacts with the number of cycles in ambient air. 15 Nrn; 9.6 Hz; 0.5 N; Au-flash plating 0.1 pm thick.

H. h&rata et ai. i Fretting effects on ele~~plate~ contacts

342

r Pd-Mi Pd Ni

0+

O(wt%) la1 original surface Fig. 5. Wear loss of each contact after 10’ cycles. 15 pm; 9.6 Hz, 0.5 N; ambient room air; Au-flash plating 0.1 pm thick.

0

0 .s

I.0

I.5

2.8

25

O(wtXl

(b)

Number of Cycles Fig. 6. Variations of the contact resistances of some contacts with the number of cycles in toluene-saturated air. 15 pm; 9.6 Hz; 0.5 N, Au-flash plating 0.1 pm thick.

shear occurred within the interfacial film. From the fact that electroplated Pd-Ni is harder than electroplated Pd, the thin Au layer on the Pd-Ni is quite likely to provide a more suitable solid lubricant effect than that on Pd. 3.2. Frettirzgin the e~vi~on~e~~ contusing trifle vapour 3.2.1. Detection of the frictional polymer Similar experiments were undertaken in toluenesaturated air to confirm the formation of a frictional polymer, structure and composition. The toluene concentration in the en~ronment was about 3%. Figure 6 shows the contact resistances of some contacts plotted VS. the number of cycles. The contact resistances of Pd and especially Pd-Ni were lower and more stable than those in the ambient atmosphere of Fig. 4. To reveal the cause of this phenomenon, the worn surface of the Pd-Ni contact was analysed by EPMA. The weight percentages of C and 0 were determined and the results are shown in Fig. ‘7. As compared with the original surface, extremely large amounts of C were detected on the worn surface. Though the weight percentages of C and 0 detected on the original surface have no correlation with one another, there is a correlation on the worn surface. From the fact that the

worn surface

Fig. 7. Correlation between the quantities of C and 0 on (a) the original and (b) the worn Pd-Ni contact surface detected by EPMA. 15 pm 9.6 Hz; 0.5 N, lo4 cycles; toluene-saturated air.

weight percentage of C is nearly proportional to that of 0, the hydrocarbon containing 0 with a certain ratio should be expected to exist on the worn surface. According to the same analysis of other contact materials, it was revealed that the Pd contact and the Au-flashed Pd-Ni contact also have organic compounds on their worn surfaces, whereas the Ni contact and the Au contact do not. For further details on the organic compounds, they were analysed by FT-IR spectroscopy. Untouched organic products on the worn surface were analysed using an IR-microscope equipped with a high sensitivity, liquid N,-cooled detector. Since the spectra obtained exhibited non-linear baselines, they were corrected by means of computer-aided procedures. These techniques are extremely useful for IR analysis of micr~uantities of crude, solid organic compounds. An FT-IR spectrum obtained from the worn surface of the Pd-Ni contact is shown in Fig. 8. The spectra from the Pd contact and the Au-flashed Pd-Ni contact were essentially the same as this. A complete elucidation of the structure of the organic compounds is not possible from the spectral consideration alone but, from some characteristic bands, it is thought that the organic products existing on the Pd-Ni contact include aromatic or unsaturated aliphatic ester linkages. A strong band in the 1720-1710 cm-’ region is assigned to the C=O stretching band of aromatic or unsaturated aliphatic esters, and a strong C-O stretch-

343

H. Murata et al. / Fretting effects on electroplated contacts

0.008-

,” “,

&km-

_n

2

1.401

0.004-

1.424

v3

s4

0.002-

1.447 Wavelength,nm Ni L spectra

1.470

1.493

0.000-

Worn Surface with Organic Deposits (Spec. 1)*O. 89 (Spec. 2ItO. 15

---

-

-0.0044000

3000

2500

Yavenumber

2000

1500

,000

(cm-‘)

Fig. 8. FT-IR spectrum obtained from the worn surface of the Pd-NI contact fretted in toluene-saturated air.

ing band of these esters also appears in the 1270-1250 cm-l region. The enhanced band in the 1600 cm-’ region corresponds to low frequency shifted oletin vibrations (aliphatic) and in-plane skeletal vibrations of aromatic rings (aromatic) [6]. Furthermore, bands in the 3000-2800 cm-’ region probably exhibit original toluene molecules. Hence, it is likely that the organic product consists of aromatic or unsaturated esters and toluene molecules. To understand the effects of the organic deposits on the Pd-Ni contact, a characteristic X-ray spectrum analysis of Ni was undertaken by EPMA. The diameter of the electron beam was 50 pm. By altering the wavelength, the characteristic X-ray spectrum was detected in the range from 1.401 to 1.493 nm. The Ni L spectra shown as a broken line and a solid line in Fig. 9(a), were obtained, respectively, from an original surface of the Pd-Ni contact and its worn surface which was fretted in ambient room air and had nickel oxides on it. The chemical state of Ni on the worn surface in toluene-saturated air was estimated by means of the curve fitting technique [9]. The results are shown in Fig. 9(b). The broken line is the Ni L spectrum obtained from the worn surface of the Pd-Ni contact fretted in toluene-saturated air, and the solid line is a spectrum synthesized by mixing 85% of the spectrum from the original surface (broken line in Fig. 9(a)) and 15% of the spectrum from the worn surface with Ni oxides (solid line in Fig. 9(b)). From the good agreement of both spectra shown in Fig. 9(b), it is reasonable to estimate that the worn surface of the Pd-Ni contact fretted in toluene-saturated air contains about 15% of the Ni oxides formed in the ambient air. Hence, the organic compound generated on the Pd-Ni contact can inhibit the oxidation reactions of contact materials. Furthermore, it functions as a good

Wavelength,nm Ni L spectra

Fig. 9. Ni L characteristic contacts.

X-ray spectra detected

for the Pd-Ni

leg

252

Toluene

Concentration

(%I

Fig. 10. Quantities of C and 0 detected on the worn surface by EPMA vs. the toluene concentration. Pd-Ni; 20 pm; 7.0 Hz, 0.5 N; 20 000 cycles.

lubricant and leads to a low contact resistance as shown in Fig. 6.

3.2.2.

Effect of toluene concentration

Using the system shown in Fig. 2, the concentration of toluene vapour in the atmosphere was controlled in the range from 0.2% to 9.0%. The experimental conditions were as follows: load of 0.5 N, amplitude of 20 pm and frequency of 7.0 Hz. After 20 000 cycles of sliding, the worn surfaces were analysed by EPMA. Figure 10 shows the relationship between the toluene concentration in the atmosphere and the quantities of C and 0 detected on the worn surfaces. At a concentration of over 3% toluene, the quantities of C and

344

H. Murata et al. I Fretting effects on electroplated contacts

0 did not increase any more, and the contact resistances were kept low and stable. Hence, even at a rather high concentration, the organic compound formed on the Pd-Ni contact does not accumulate above a certain quantity at the interface and does not cause an increase in the contact resistance. The formation and exclusion by fretting actions likely reached equilibrium. At low toluene concentrations the quantity of C decreased and that of 0 increased. As shown in Fig. 11, the correlation coefficient of the distributions of C and 0 contents was reduced. One of the reasons for these phenomena may be that more metal oxides are produced during the formation of organic compounds at low concentrations. If almost all the detected C and 0 are components of the organic compounds, the correlation between the two elements will become linear, as shown in Fig. 7(b), and the correlation coefficient will approach unity. Another reason may be that, at low toluene concentrations, the ester linkage content of the organic compound may well increase because sufficient 0 exists for the formation of ester linkages. Experimental conditions where the formation of organic compounds and the remarkable oxidation occur at the same time are not appropriate to understand the behaviour of the former alone. An Au-flash plating

on the Pd-Ni contact has already been known to inhibit oxidation (fretting corrosion) efficiently, remain stable and assure low contact resistance.

3.2.3. Effect of the number of cycles Organic compounds formed on the Pd contact sometimes lead to high contact resistance [2]. To confirm the behaviour of the organic compound, experiments with long test iterations were undertaken for the Pd-Ni contact. A pair of Pd-Ni contacts with a load of 0.5 N were fretted in toluene-saturated air at an amplitude of 20 pm and frequency of 7.0 Hz. As a result, the organic compound formed on the Pd-Ni contact does not accumulate over a certain quantity at the interface, even after a relatively large number of cycles, i.e. 10’ cycles. Moreover, it is confirmed that the correlation between C and 0, and also the FT-IR spectra obtained from the worn surfaces, do not change with the number of cycles. Hence, the organic compound formed on the Pd-Ni contact does not vary with the number of cycles from both the quantitative and qualitative points of view. Variations in the contact resistances of the Pd-Ni contact and the Pd contact with the number of cycles are shown in Fig. 12. The Pd-Ni contact exhibited remarkably low and stable contact resistance during the test iteration of 10’ cycles. It is noted that the organic compound formed on the Pd-Ni contact acts as good lubricant and does not cause an increase in the contact resistance.

3.2.4. Compa%on

with Pd and Au-flashed Pd

contacts

Toluene

Concentration

Fig. 11. Correlation coefficient vs. toluene concentration.

0

20000

between

40000 60000 Number of Cycles

(X1 quantities

80000

of C and 0

100000

Fig. 12. Variations of contact resistances of the Pd-Ni contacts with the number of cycles in toluene-saturated pm; 7.0 Hz; 0.5 N.

and Pd air. 20

By fretting in an environment containing toluene vapour, the Pd contact acquired a higher and less stable contact resistance than the Pd-Ni contact, as shown in Fig. 12. Furthermore, the Au-flashed Pd contact exhibited the same tendency against the Au-flashed Pd-Ni contact. By FT-IR spectroscopy, the organic compounds formed on the Pd-Ni, Pd, Au-flashed Pd-Ni and Auflashed Pd contacts similarly consist of aromatic or unsaturated esters and toluene molecules. Furthermore, by EPMA, it was confirmed that the correlation coefficients of C and 0, and the C:O atomic ratios were all the same on the surfaces of these contact materials. Figure 13 shows contact regions and the characteristic X-ray images of C. The worn surface of the Pd contact is very rough. Less C (organic compounds) is in the contact region and a large amount is around it. The organic compound is considered to be easily excluded outside the interface and does not act sufficiently as a lubricant. This is why the Pd contact acquires a greater

H. Murata et al. / Fretting effects on electroplated contacts

345

contact can be superior to the Au-flashed Pd contact as well.

4. Conclusions

P d -N

i

contact

5-G

P d

contact

Fig. 13. Scanning electron microscopy @EM) photographs and C characteristic X-ray images of contact areas. 20 pm; 7.0 Hz; 0.5 N; 20000 cycles; toluene-saturated air.

but less stable contact resistance. In contrast, the worn surface of the Pd-Ni contact is very smooth, and has a relatively large amount of organic compounds which function as a good lubricant in the contact area and only a small amount around it. In short, the organic compound formed on the Pd-Ni contact is considered to adhere to metal, covering the surface, and hence serves as a superior lubricant. Since esters are derivatives of carboxylic acids, they can form metallic soaps with some active metals [7]. Pd-Ni contains relatively active Ni, so that the Ni and the esters of the organic compound can form adsorption bonds, similar to some metallic soaps, which can be adsorbed strongly to the metal surface. This is why the organic compound formed on the Pd-Ni contact inhibits the oxidation, functions as a good lubricant and leads to a lower and more stable contact resistance than occurs with the Pd contact. Therefore, the Pd-Ni contact is superior to the Pd contact when used for environments containing toluene vapour, and the Au-flashed Pd-Ni

The following conclusions could be drawn. (1) By analysing the worn surfaces, a larger amount of 0 was detected in the stroke range from 15 to 40 pm. Fretting corrosion occurred actively in this stroke range. (2) By fretting in ambient room air, the Pd-Ni contact acquired a high contact resistance owing to oxidation, but a thin Au-flash plating 0.1 pm thick on it remained stable and showed a low contact resistance similar to an Au contact. Furthermore, wear loss of the Auflashed Pd-Ni contact was less than that of the Au contact because of the solid lubricant effect. (3) Organic compounds formed on the Pd-Ni surface were revealed to include aromatic or unsaturated esters and toluene molecules. (4) From investigation of the correlation between the quantities of C and 0 detected from the worn surface by EPMA, a thin layer of organic compounds can be formed in restricted conditions. (5) Even at rather high concentrations and after a relatively large number of cycles, the organic compound formed on the Pd-Ni contact did not accumulate significantly at the interface and did not increase the contact resistance. (6) The organic deposits on the Pd-Ni contact are expected to be adsorbed strongly to the metal surface because of adsorption bonds, similar to metallic soaps. Hence, they function as a lubricant and inhibit the oxidation of the metal surfaces. These are the reasons for the low and stable contact resistances of the Pd-Ni contact. (7) The Pd contact acquired a higher and less stable contact resistance than that of the Pd-Ni contact. Furthermore, the Au-flashed Pd contact was inferior to the Au-flashed Pd-Ni contact in the same way, because the organic deposits on the Pd contact are excluded easily outside the interface during the fretting motion and do not act as a lubricant, resulting in an increased contact resistance.

References 1 A. Kassmann, T. Imrell, S. Hogmark and S. Jacobson, Proc. 36th IEEE Holm Con5 on Electrical Contacts, Monttial, August, 1990, p. 395. 2 M. Antler and E. S. Sproles, IEEE Trans., Components, Hybrids and Manufacturing Technology Society, 5 (1) (1982) 158. 3 H. W. Hermance and T. F. Egan, Bell Syst. Tech. J., 37 (1958) 739.

346

H. Murata et al. ! Fretting effects on electroplated contacts

4 W. A. Crossland, P. M. K. Murphy and I. A. Murdoch, IEEE Trans., Parts, Hybrids, Packag., 10 (1974) 60. 5 M. Hasegawa and K. Sawa, IEEE Trans., Components, Hybrids and Manufacturing Technology Society (1) (1990) 33. 6 B. T. Reagor and L. Seibles, Proc. IEEE Holm Conf on Electrical Contacts, Chicago, IL, September, 1980, p. 95.

7 F. P. Bowden and D. Tabor, The Friction and Lubrication of Solids, Vol. 1, Clarendon Press, Oxford, 1950, p. 111. 8 M. Antler and M. Feder, Pmt. Electron. Components 36th Co& May 1986, Seattle, p. 244. 9 Y. Imada, F. Honda and K. Nakajima, X-ray Anal., 21 (1990) 111.