EVALUATION OF ATOM-PROBE SPECTRA FROM TITANIUM CARBONITRIDE

HAL Id: jpa-00229961

https://hal.archives-ouvertes.fr/jpa-00229961

Submitted on 1 Jan 1989

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

HAL, est

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

EVALUATION OF ATOM-PROBE SPECTRA FROM TITANIUM CARBONITRIDE

U. Rolander, H.-O. Andrén

To cite this version:

U. Rolander, H.-O. Andrén. EVALUATION OF ATOM-PROBE SPECTRA FROM TITA- NIUM CARBONITRIDE. Journal de Physique Colloques, 1989, 50 (C8), pp.C8-371-C8-376.

�10.1051/jphyscol:1989863�. �jpa-00229961�

COLLOQUE DE PHYSIQUE

Colloque C8, Supplkment au n 0 l l , Tome 50, novembre 1989

EVALUATION OF ATOM-PROBE SPECTRA FROM TITANIUM CARBONITRIDE

U. ROLANDER and H

.

^-0. A N D R ~ N

Department o f Physics Chalmers U n i v e r s i t y o f Technology, S - 4 1 2 96 Gdteborg, Sweden

Abstract

-

Four (Ti, W)(C, N)-Ni alloys with known composition have been used to investigate how the detected composition varies with crystallography, pulse ratio, nitrogen content and temperature. The interpretation of mass spectra is discussed and conditions of analysis with which correct results can be obtained are given.

Introduction

Atom-probe analyses of the cubic carbide phase in cemented carbide materials have proven to be difficult to evaluate. This is due to the rather complex and irregular field evaporation behavior of this phase [I]. In general, ions tend to evaporate in pairs or clusters to a much higher extent in carbides and carbonitrides than in metallic systems. As a consequence a significant amount of the most abundant ions is lost because of pile-up in the detector. In the case of titanium compounds, the evaluation is further complicated by overlaps between the 0+ and 48Ti3+ peaks and between the C2+ and 48Ti2+ peaks. These overlaps cannot be deconvoluted di- rectly because of the systematic errors in the detected isotope distribution of titanium. However, if a tilted detector is used to increase the multi-hit detectivity, a statistical method can be used to correct spectra for the pile-up effect [21. In this investigation four Ti(C,N) - 2 vol% Ni alloys containing 0.22 to 22.5 at% nitrogen were used to investigate the conditions under which correct analyses can be obtained, and how spectra should be interpreted.

Experimental

The chemical compositions of the four materials used are shown in Table 1. The nickel content (2 vol% in all four materials) is excluded in order to estimate the composition of the carbonitride phase. The materials were liquid phase sintered at 1520°C. The homogeneity of the materials was checked using wavelength dispersive x-ray spectrometry (WDS).

Specimens were prepared by electropolishing using 5 % H2S04 in methanol as electrolyte. The preparation of needle shaped specimens from cemented carbides and similar materials is well documented in the literature [3,4,5].

Analyses where made using a pulse ratio of 15%, 20% and 25% and a specimen temperature of 50K, 70K, and 90K. Since the field evaporation field of cubic carbides is highly anisotropic over the specimen surface, [I], analyses were made at the edge of a <001> pole (minimum field evaporation field), at the edge of a < I l l > pole (maximum field evaporation field) and at a high indexed location close to the center of the triangle formed by the <001>, <011> and < I l l > poles.

All analyses were made at a pressure of less than 3 x 10-10 torr.

The atom-probe used in this investigation has been described in detail previously [6, 71. Flight times are measured over a flight length of 1.6 meters using four parallel counters. A Poschenrieder type energy compensator is used to increase the mass resolution of the system.

The dead time of the detection system has been measured previously [I]. It was shown that ions arriving with a time difference of less than 9 ns are detected as one ion.

For the analyses made at 50K and 70K, a 30' vertically tilted detector was used in order to improve the multi-hit detectivity. The ions were focused into a 2 cm vertical line over the detector surface. At 10 kV and a mass to charge ratio of 24 u this introduces an artificial spread in flight time of 35 ns, corresponding to a spread in mass to charge ratio of 0.3 u. In addition, all analyses were corrected for losses due to detector pile up, using a statistical correction procedure.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1989863

Both the performance of a tilted detector and the statistical correction procedure are described in detail elsewhere [2].

Results and Discussion Interpretation of spectra.

Figure 1 shows a spectrum obtained from material C (9.19 at% nitrogen). Except for the usual ions from C, N, Ti and W several types of molecular ions were detected. At m/q=18 a small peak, which could be water from the residual gas in the vacuum system, was seen. However, by merging several analyses a small peak at 18.5 which fits well with the isotope distribution for carbon (3.4 % of the 18 peak) was found. Thus, the peak at 18 must be interpreted as C32+ ions.

The peak at 19 increases with increasing nitrogen content and originates from C2N2+ ions. So far, no evidence has been found for the existence of CNz2+ or N32+ ions. The peaks at 20 and 22 are Ne+ ions. Neon was used for imaging the specimens. The prominent peak at m/q=26 increases with increasing nitrogen content and originates from CN+ ions. This suggests that C2+

ions may also exist and overlap with the main Ti2+ peak at m/q=24. After the statistical correction procedure these were found and could be deconvoluted. An insignificant amount of ions was usually detected at 28. Since these could not be correlated to the nitrogen content they were interpreted as CO+ rather than N2+. CO+ ions may well come from adsorbed residual carbon monoxide in the vacuum system so they were left out of the analyses.Between m/q=30 and 33 TiN2+ and Ti02+ ions were detected. A small amount of C3+ and C2N+ ions was sometimes found at 36 and 38 respectively. Table 2 shows the relative abundance of each type of complex ion in terms of its contribution to the total C or N content. The data is taken from the analysis of alloy D done at 70 K and using a pulse ratio of 20 ^%.

As expected the amount of complex ions decreases with increasing field evaporation field, that is with decreasing temperature, nitrogen content or distance to the < I l l > pole. An increased field evaporation field also shifts the relative abundances of the Ti, W, C and N peaks towards higher charge.

Oxygen would arrive mainly as O+ ions and overlap with the main Ti3+ peak at m/q=16.

These were also found. However, the oxygen content was found to vary in a rather unreproducible manner between values that agreed well with the bulk values and values that were significantly to low. The result can be explained since, during the time period when the analyses where done, the residual gas in the atom-probe contained a significant amount of hydrocarbons. These show up as a slight noise between mass to charge ratios 15 and 18. This noise is not significant for the total analyses, but will make deconvolution of the 0+ peaks unreliable. The small ~ i 3 + peaks which are used in the deconvolution will be slightly to high leading to a rather large underestimation of the 160+ peak. For this reason, we have in the present case chosen not to deconvolute the 0+ peaks but instead simply subtracted the bulk oxygen content of table 1 from the measured titanium content.

Effects of crystallography

Atom-probe analyses made close to < I l l > poles were always Ti deficient, irrespective of pulse ratio and temperature, due to preferential field evaporation. The analyses made at the <001>

poles where instead often, but not always, C deficient. This could be explained by surface diffusion of C. However, these analyses were all made with a conventional 0' tilted detector, which means that very few pairs were detected in the carbon peaks. This makes statistical correction of the carbon peaks difficult. Analyses made in the high indexed direction and with low mass resolution showed best reproducibility.

Judging from FIM images, the crystallographic anisotropy of the field evaporation field decreases with increasing nitrogen content and with increasing temperature.

Effect of pulse ratio

Table 3 shows three atom-probe analyses of alloy C , made at a specimen temperature of 70 K and with different pulse ratios. As can be seen, at this temperature, a pulse ratio of 15% is not sufficient to avoid preferential field evaporation of titanium between pulses. This analysis is substantially titanium deficient. The other two analyses, however, agree well with the bulk composition.

The errors given are maximum errors, that is the the sum of the statistical error , the error for the dead time correction and the error for the C2+ deconvolution. Clearly, this is an overestimation of the actual errors.

In figure 2 the normalized metal content, that is the experimental content divided by the bulk content, is plotted as a function of the pulse ratio. Clearly, at pulse ratios of 20% or more, correct results are obtained.

Effect of nitrogen content

Table 4 shows atom-probe analyses of all four materials, made at 70 K and using a pulse ratio of 20%. The analysis of alloy A, which does not contain nitrogen, is substantially titanium deficient.

For this alloy, where the crystallographic anisotropy is most pronounced, we have obviously not managed to avoid preferential field evaporation. The situation is not improved by increasing the pulse ratio to 25%.

Figure 3 shows the normalized metal content as a function of nitrogen content. It is seen that for alloys B, C and D, the experimental compositions agree well with the bulk compositions. The analyses of alloys B and D are somewhat low in nitrogen content. However, since the measured stoichiometries are correct, we believe that this discrepancy is due to slightly inhomogeneous carbonitride grains.

Effect of specimen temperature

Table 5 shows analyses of alloy A made at SOK, 70K and 90K. The analysis at 90K was done using 15% pulse ratio and the other two were done using a pulse ratio of 20%. All analyses where done in the high indexed direction. The analysis done at 90K is somewhat uncertain since it was done without the tilted detector. The data show that the preferential field evaporation of Ti decreases with increasing temperature. There is no reason why this should not be true also for the nitrogen containing alloys but this has not been investigated.

In figure 4 the normalized metal content is plotted against temperature. A correct stoichiometry is obtained at 90K.

The results suggest that analyses of titanium carbonitride should be made at a high enough temperature so that the crystallographic anisotropy is kept sufficiently low. At this temperature the pulse ratio should be adjusted so that preferential field evaporation of titanium is avoided.

Conclusions

Four (Ti, W)(C, N)

-

Ni alloys have been used, to investigate the conditions under which atom-probe analysis of titanium carbonitrides gives correct results.

All alloys show crystallographic anisotropy. The effects of this decreases with increasing nitrogen content and temperature, and can be minimized by performing the analyses in high index directions.

Analyses made at 70 K and with a pulse ratio of 20% give correct results except for almost pure titanium carbide. In that case a substantial amount of titanium is lost, probably due to

preferential field evaporation of titanium between pulses.

At 90K correct results could be obtained also for the least nitrogen containing alloy.

Acknowledgements

This work has been financially supported by AB Sandvik Coromant and the National Swedish Board for Technical Development (STU). AB Sandvik Coromant also provided the materials and made the WDS measurements.

References

[I] U. Rolander and H. 0. Andren, J. Phys. (Paris), Colloq. C6, 49 (1988) 299 [2] U. Rolander and H.O. AndrQn, this conference

[3] A. Henjered and H. Norden: J. Phys. E: Sci. Instrum. 8,617-619 (1983) [4] U. Ro1ander:J. Phys. (Paris), Colloq. C7,47 (1986) 449

[5] U. Rolander and H.O. Andren,

ICSHM3 -Nassau, Bahamas, 1987, Mater. Sci. Eng., A105/106 (1988) 283 [6] H. 0. Andren and H. Norden, Scand. J. Metall., 8 (1979) 147

H. 0. Andren, J. Phys. (Paris), Colloq. C7,47 (1986) 483

Tablel. Chemical composition of the carbonitride phase in the four alloys studied. All materials also contain 2 vol% nickel. All values are in at%.

Allov: A B C D

Ti 50.14 49.91 50.36 50.22

Table 3. Atom-probe analyses of alloy C done with three different pulse amplitude to DC voltage ratios.

Table 2. Contribution of the different species to the total C or N content. Data is taken from the analysis of alloy D in table 4 All values are in percent of total C or N content respectively.

pulse ratio: 15% 20 % 25 %

Element

C N

Ti 44.97 f 2.22 50.16

+

^{3.32 50.54 f 2.32}

W 0.50

+

^{0.11 0.28}

+

^{0.11 0.53 f 0.1 1}

C 44.49

+

^{2.56 39.03 f 3.89 38.65}

+

^2.68

N 8.48

+

^{0.42 8.96 f 0.62 8.71}

+

^0.43

0 (bulk value) 1.57 1.57 1.57

C2+ C+ C2+ CN+ C32+ C3+ C2N+ N2+ N+ TiN2+

39.1 32.9 17.6 6.2 1.6 1.3 1.1 0.2

-

7.9

-

0.7 0.1 3.6 64.2 23.5

Table 4. Atom-probe analyses of all four alloys made at 70 K, using a pulse ratio of 20%. Errors are given as the sum of the statistical error, the error in the dead time correction and the error in the deconvolution of C2+.

Allov: A B C D

Ti 46.89 +. 2.08 50.82 f 4.35 50.16 f 3.32 50.60 f 2.39

W 0.79

+

^{0.12 0.59}

+

^{0.21 0.28 f 0.11} 0.40 -1.0.11

C 50.92 f 2.44 44.07 f 5.21 39.03 f 3.89 26.72 f 2.60 N 0.16 f 0.05 2.95 f 0.46 8.96 f 0.62 20.34 f 0.68

0 (bulk value) 1.34 1.57 1.57 1.94

Table 5. Atom-probe analyses of alloy A done at three different temperatures

Ti 44.05 f 1.46 46.89 f 2.08 49.36 f 2.76

W 0.94 f 0.28 0.79 f 0.12 0.61 +_ 0.17

C 53.32

+

^{1.47 50.92 f 2.44 48.23 f 4.86}

N 0.35 f 0.17 0.16 f 0.05 0.47 f 0.15

0 (bulk value) 1.34 1.34 1.34

mass/charge ratio (amu)

Figure 1. Atom-probe spectrum obtained from alloy C. The bar width is 0.1 atomic mass units.

Y 12

u

8

fi 8 1.1

3 _Y m Figure 2. Normalized detected metal content

2

^1,o

for alloy C, (Ti

+

^WIexp / (Ti

+

^W)bulk , as a

z

function of pulse ratio. Correct results are

.*

^N

obtained for pulse ratios of 20 % or higher. Data "m 0 9 is taken from Table 3. Error bars are maximum

6

errors.

2

_0,s

10 15 20 25 30

pulse (%I

Y 12

Y

8

fi $ 1,l

3 m

Y

Figure 3. Normalized detected metal content,

i2

^1,0

(Ti

+

W),xp / (Ti + W)bulk, as a function of bulk

a, .-

_N

nitrogen content. Correct results are obtained

except for alloy A which contains almost no

s

00,9 nitrogen. Data is taken from Table 4. Error bars

g

are maximum errors.

2

u 12

Y $

F;

8

1,l

3 m

0,s;

-

^{I I}

10 20 30

nitrogen content [at%]

P

-

Y

-

Figure 4. Normalized detected metal content,

z _.*

^N

^-

(Ti

+

W)exp / (Ti

+

W)bulk , as a function of "m 0,9 specimen temperature. . Data is taken from

Table 5. Error bars are maximum errors.

I

^fi _OpS40

'A

^'

'

A

^*

^{s b .}

^{90 '100}

temperature (K) .---

-

7

-

,AL*)

-

- -

* I

-