Mat. Res. Bull. Vol. i0, pp. 303-312, 1975. Pergamon Press, Inc. Printed in the United States.

CHARACTERISTICS OF YLaTm-GARNET

FILMS FORMAGNETIC BUBBLE APPLICATIONS

A. B. Smith, M. Kestigian, and W. R. Bekebrede Sperry Research Center, Sudbury, Mass. 01776

(Received February 26, 1975; Communicated by R. C. DeVries)

ABSTRACT

Films of YI.6La0.3Tml.l(FeGa)5012 have been grown by liquid-

phase epitaxy on Gd3Gab012 substrates. This is a new film

composition whose high mobility (~i000 cm/sec/0e) and low

temperature coefficient of bubble diameter (~0.1%/~C) make it attractive for bubble domain devices. Films supporting 3 to 8~m-diameter bubbles have been prepared with typical defect densities less than 5/cm 2. Data are presented to show that this material can be grown reproducibly despite a rather large La segregation coefficiento Measurements of the uniaxial anisotropy as a function of lattice mismatch are presented along with the results of ion implantation studies which show that a

dosage of 3 x i018 cm -2 of Ne is sufficient to suppress hard bubbles in this material. Also, bubble velocity measurements are discussed in detail as they reveal a number of interesting effects. These include the existence of a perpendicular velocity component which can be virtually eliminated by apply- ing an in-plane field and the absence of extreme dynamic conversion effects in high gradient fields.

Introduction

F~itaxial garnet films for bubble-domain device applications should have both a high mobility and a low temperature coefficient of bubble diame- ter. In the past, it has not usually been possible to satisfy both of these requirements simultaneously in the same material. For example,(YEu)3(FeGa)5012 has an acceptable temperature coefficient but a poor mobility; the opposite

is true of (YGdTm)s(FeGa)50.~ (i-5) Some of the new garnets containing germanium and calclum are r~orted i6,7) to achieve both requirements; how- ever, it is difficult to reproducibly grow homogeneous low-coercivity films of these materials. Hence, it seems desirable to investigate alternative materials. We report here on YI.6La0.3TmI.I(FeGa)5012 , a new material

303

304 YLaTrn-GAILNET FILMS Vol. I0, No. 4

which has both good mobility and a desirable temperature dependence.

Theoretical Background

In formulating new bubble-domain garnet composit ions, one normally attempts to simultaneously fulfill all of these requirements:

a. Include at least one magnetic rare earth ion in order to provide growth-induced anisotropy.

b. Maximize mobility by choosing the rare earth(s) from those pro- viding the least damping and by using only enough of the rare earth(s) to provide the required anisotropy.

c. Make ~ , the material length parameter, temperature independent by arranging to have the magnetic compensation point occur at a low temperature. ( ~ is related to the compensation temperature

because ~ =~K/M 2 , where K is the uniaxial anisotropy con-

Since~has a negative temperature coefficient, stant. (2,S,9) we wish M 2 to have a similar dependence so that ~ will be

temperature independent. The temperature coefficient of M will be negative if the compensation point is at a low temperature.)

d. Choose the constituent ions so that the resultant film will have a lattice constant that matches the readily available Gd3Ga5012 substrates.

In previous garnet compositions, the necessity for satisfying the essential lattice match condition (d) has severely limited thechoice of ions that could be used for satisfying conditions (a), (b) and (c). This problem would be avoided if one could incorporate either lanthanum or lutetium into the dodecahedral sites. Because lanthanum is so much larger and lutetium so much smaller than the other ions normally present on these sites, only small amounts would be required to satisfy (d). Che could then fill up most of the dodecahedral sites with ionschosen to satisfy (a), (b) and (c). Such an approach has not been followed in the past in making gallium- or aluminum- iron garnets for bubble-domain devices. (Wall mobility experiments using

(YLa)~(FeGa)~019 have been reported (i0); however, the anisotropy in this material must ~ve been primarily strain-induced as no magnetic rare earths were included,) The reluctance of other investigators to employ La or Lu in bubble garnet films is apparently due to fears of large segregation coef- ficients and resultant poor reproducibility. However, we find that (Y, La,Tm)3(PeGa)5012 can be grown with good reproducibility using the growth

procedures described below. This result opens up the possibility of a whole family of garnet compositions containing La and/or Lu.

Film Growth

The melt composition we have used to grow Y1.6La0.3Tml.iFe4.0Gal.0012

films is given in Table I. The films were grown by horizontal dipping using a 60 rpm rotation rate. The melt temperature was N 920°C and the films grew at a rate of N 0.8 ~m/min. Defect densities of less than 5/cm 2 were readily achieved using these growth conditions.

The reproducibility of the growth process was checked by preparing a series of films using the same solution over two days. The melt was re- equilibrated between dips by raising it to N I120°C. Also, 5.g of Pb0 w e r e

Vol. I0, No. 4 YLaTm-GARNET FILMS 305

added at the end of the first day to make up for evaporation loss. No other changes were made to the melt. The results of measurements on these films are shown in Table 2. X-ray fluorescence data shows no discernable dif- ference in La content among these films. These measurements indicate that the La subscript in the chemical formula should be 0.29 with a measurement uncertainty of ~ 0.015. The variations in 4~M and ~ shown in Table 2 are similar to those observed in growing gallium-iron garnets that do not con- tain lanthanum and are significantly less than the variations often en- countered in growing (YSmCa)3(FeGe)5012 •

TABLE i

Melt Composition Used to Grow Yl.6La0.3Tml.iFe4.0Gal.0012

Y203 0.31 mole %

La203 0.30

Tm203 0.20

Ga203 1.22

Fe203 8.10

PbO 84.77

B203 5.10

TABLE 2 Characteristics of a Series of Yl Sequentially from the Same Melt. "6La0"3Tml'iFe4"0Gal'0012 Films Grown

(davg is the Bubble Diameter at a Field Midway between Runout and Collapse.)

Thickness 4~M £ d Day Film (~m) (G~ ~m) ~)

1 1 5.6 115 0.75 7.2

i 2 6.2 115 0.75 7.i

1 3 6.0 107 0.79 7.4

1 4 5.5 112 0.88 7.6

2 5 5.7 108 0.76 7.4

2 6 5.7 108 0.73 7.7

2 7 6.6 117 0.86 7.4

Hcol

Magnetic Properties

Figure 1 shows the temperature dependence of bubble collapse field and of the demagnetized stripe width W in a typical sample of

306 YLaTm-GARNET FILMS Vol. i0, No. 4

400

300

2oo

100

I I

-o------o

0 I 20

I I I

I I I I I 0 20 40 60 80

TEMPERATURE (°C)

FIG. I

100

Temperature dependence of demagnetized stripe width (W) and bubble collapse field (Hcol) in

YI.6La0.3TmI.IFe4.1Ga0.9012

3

Yi.6La0.3Tml.iFe4.1Ga0.9012. Other important parameters of this film not

shown in the figure are: mismatch between substrate and film lattice con-

stants <0.003A, film thickness = 4.6~m, 4rim = 250G, ~ = 0.39~m, K = 10,570. ergs/cm 3, q = K/2rrM 2 = 4.2, average bubble diameter (i.e., the diameter at a field half way between the collapse and runout fields) = 3.SW2a. If a dif- ferent average bubble diameter is required for a particular application, it can be obtained by changing the Ga203 content of the melt in order to vary

the magnetization. Lower values of 4~Mwill increase bubble size as well as increasing q and the temperature dependence of W; raising 4rim will have the opposite effects. Practical values of bubble size with this material cover the range of 3 to 8~m.

Our original motivation in developing YLaTm-garnet was to create a material having the high mobility of YGdTm-garnet but without its extreme temperature dependence. It is interesting, therefore, to compare the tem- perature dependence of these two materials. Figure 2 shows such a comparison for two samples chosen to have the same 4~M at room temperature. Here we have plotted 4~M and ~ as determined from W and Hco I by the Fowlis-

technique (ii), We see that YLaTm-garnet has a much more negative slope of 4nM.

Therefore, since ~ ~A/K/M 2 and since K decreases with temperature, the £ in YLaTm-garnet varies much less with temperature. This difference in the tem- perature dependence of 4~ is associated with the difference in the compensa- tion temperature of the two materials. By observing the temperature at which the domains disappear, we estimate that the compensation temperature in YGdTm-garnet occurs at ~ -70°C while in YLaTm-garnet it is at N -140°C. One of the main advantages of YLaTm-garnet is its lower compensation point com- pared to any composition containing an appreciable amount of Gd.

Measurements of the uniaxial anisotropy of (YLaTm)3Fe4.0Gal.0012 vs

Vol. i0, No. 4 YLaTrn-GARNET FILMS 307

300 strain are shown in Fig. 3. This data was obtained by the Kurtzig-Hagedorn method(12) on successive films grown while changing the amount of La203 in the melt given in Table I. ~ 200

Growth temperature and growth rate were kept substantially constant for all these films, thus keeping the magnetiza- tion within a range of 124 140 to 176G. (No detailed study of the dependence of K on 10 growth conditions has been made, but we have measured 08

films having K values slightly different from 0.6 those in Fig. 3. One E

zl typical example is the sample of Fig. 1 discussed above. Surprisingly, the magnitude of K seems to be correlated with magnetiza- tion rather than with growth rate or temperature as might be expected.)

The data in Fig. 3 indicate that the uniaxial anis0tropy constant asso- ciated with the growth- induced anisotropy is 7100 ergs/cm 3 while the strain- induced portion is 4.4 x 105 ergs/cm3/A. Since x-ray

I I I I I

I I I I I

0.4 - -

02 - 4 0

O

1 I I _

I I I I I - 2 0 0 2 0 4 0 6 0 8 0

TEMPERATURE(°C)

FIG. 2

Temperature dependence of magnetization (4~M) and characteristic length (~) in

Yi.6La0.3Tml.iFe4.1Ga0.9012 and

Yi.2Gd0.6Tml.2Fe4.4Ga0.6 012

fluorescence measurements show that the Tm content remained relatively con- stant (! 9% for all except the one sample having af - as = .013), we can

apply the theory of Heinz et al (13) to these data to estimate kll I. We obtain kll I = 1.3 x i0 ~6. This value is more than sufficient to permit hard bubble suppression by ion implantation (14). After implanting a series of samples at

different dosages, we find an Ne dosage of 3 x i013cm-2 at 100KeV to be sufficient.

Velocity data on an ion implanted YI.6La0.sTmI.IFe4.0Gal.0012 film

is shown in curve (a) of Fig. 4. These data were obtained by the bubble-shift technique using the same type of current conductors described by Vella- Coleiro and Tabor (15) except that all dimensions were increased bya factor of two. All measurements were made using a compensating pulse (15). The bars associated with each data point in Fig. 4 represent the spread in data obtained by repeating the measurement ten times; the point represents the average of these ten measurements. There is no evidence of dynamic conver-

sion (16) in this curve except perhaps in the data taken with the largest drive field. However, the situation is not clear because of spurious domains (17) nucleated during the pulse. These domains are located in front of the bubble and hence could exert a repelling force on it. Of course, the

308 YLaTrn-GARNET FILMS Vol. i0, No. 4

K (erg/cm 3)

I I t -0.010

10,000

0 0.010

af - a s (.~) FIG. 3

Uniaxial anisotropy constant (K) as a function of substrate/lattice mismatch (af -a s) in (YLaTm)3Fe4.0GaI.O012

exac t l o c a t i o n of t h e s e domains dur ing the pu l se i s not known. However, i t can be r e a d i l y c a l c u l a t e d t h a t a t the beginning of the pu l se the net f i e l d (b ias + pulse) i s zero on ly 26.Wn from the l ead ing edge of the bubble, hence n u c l e a t i o n could occur q u i t e c lose to the bubble under o b s e r v a t i o n . In obtaining the ~H = 14.0e point in Fig. 4, a pulse width of i00 nsec was em- ployed and the bubble moved N 5.~m. Longer pulses and their associated greater bubble displacements did give lower apparent velocities, which is consistent with the above model of interactions with spurious domains. Shorter pulses were not employed because of the inaccuracies associated with pulse width measurements caused by the - i0 nsec rise and fall times of virtually all available high-power pulse generators.

Another impor tan t aspec t of bubble p ropaga t ion i s the angle a t which the bubbles t r a v e l . In the YLaTm-garnet f i lm of F ig . 4, we observe propaga- t i o n a t ~ 50 ° t o the app l i ed g r a d i e n t a t low d r ives (AH < ~ 2) . As the drive field is increased, this angle decreases until the bubble travels vir- tually along the gradient for AH ~ 7. (The velocities plotted in Fig. 4 are for the component of motion along the gradient.) Angular propagation is not unique to this material; similar behavior has been observed in other im- planted garnet films (17,18,19). It is interesting to note that the angular propagation in YLaTm-garnet does not have any obvious deleterious effect on device performance (20).

As has been observed in other implanted garnet materials(18), bubble propagation in this material can be changed drastically by the application of in-plane fields. Curve (b) in Fig. 4 shows the Velocity vs drive character- istic obtained with an in-plane field perpendicular to the field gradient. Similar results are obtained with a field applied along the gradient. In both cases there is a definite threshold value of in-plane field (-- I00 Oe) at which the bubble begins to travel substantially parallel to the gradient and an increase in speed is observed. (It should be noted that the absolute

Vol. I0, No. 4 YLaTrn-GARNET FILMS 309

10,0001 ' I ' I ' ] '

L - _

8,000

E v >. I.- o 0 _1 LU >

6 ,000

4 ,000

= 3 6 0 0 cm/s Oe ®

2,000~- r "r/4'-'-'-/~=980cm/sOe

0 ~ "0 4 8 12 16

AH (Oe) FIG. 4

Bubble velocity vs the field difference across the bubble (AH) as measured by the bubble-shift technique: a. without any in-plane field

b. with an in-plane field of 150.0e perpendicular to the gradient direction.

(The YI.6La0.3TmI.IFe4.0Gal.0012 film used for these

measurements had the following characteristics: bubble diameter = 5.6~m,

film thickness = 5.7~m, substrate/film lattice mismatch < 0.003A, K = 6440 ergs/cm 3, 4r~ = 150.G, q K/2 M 2 = 7.2,

ion implantation: 1014cm -2 Ne ions at i00 KeV)

speed increases substantially, not just the projection of the velocity along the gradient direction.) This threshold does not appear to be directly related to effects involving saturation of the implanted layer since the speed for normal bubbles before implantation was the same as that shown in curve (a). Therefore, the in-plane field appears to modify bubble behavior by modifying the structure of the cylindrical wall of the bubble. It appears that, under the influence of the in-plane field, the bubble switches to a two-Bloch-line configuration allowing it to travel parallel to the gradient (21), If this is so, then in the absence of the in-plane field, it must have a different con-

310 YLaTrn-GAPd~ET FILMS Vol. I0, No. 4

figuration. This conclusion contradicts the simple theory of the effect of

the implanted layer (22) but could be consistent with a more refined theory that included the effect of the domain walls in this layer (19,23).

Conclusions

The inclusion of La in bubble-domain garnet films can provide the means for achieving a desirable combination of bubble-domain properties. An example is the material YI.6Lao.3TmI.I(FeGa)5012 which has a mobility of

lO00cm/sec/Oe. A temperature coefficient of bubble diameter of ~ 0.1%/~C is observed for 3.3~m-diameter bubbles. Good quality films of this material can be grown reproducibly by liquld-phase epitaxy in spite of the large difference in atomic size between lanthanum and the other cations that occupy the dodecahedral sites.

Acknowledgments

The authors would like to acknowledge useful conversations with W. D. Doyle, R. M. Josephs, and B. F. Stein and the expert technical assis- tance of A. Doppler, F. Garabedian, R. Giacalone, W. Goller, and C. Ward.

References

I. J. W. Moody, R. M. Sandfort, and R. W. Shaw, Semi-Annual Technical Report, ARPA Contract No. DAAHOI-72-C-I098 (February ii, 1973).

2. R. M. Sandfort, R. W. Shaw, and J. W. Moody, AIP Conf. Proc. 18, 237 (1973).

3. R. M. Josephs, Appl. Phys. Le t t . 25, 244 (1974).

4. G. P. Ve l la-Cole i ro , AlP Conf. Proc 10, 424 (1972).

5. F. C. Rossol, AlP Conf. Proc. 10, 359 (1972).

6. W. A. Bonner, J. E. Geusic, D. H. Smith, L. G. Van U i t e r t , and G. P. Ve l la -Cole i ro , Mat. Res. Bu l l . ~, 1223 (1973).

7. J .W. Nielsen, S. L. Blank, D. H. Smith, G. P. Ve l la -Cole i ro , F. B. Hagedorn, R. L. Barns, and W. A. B io l s i , J. Electron. Mater. ~, 693 (1974).

8. P. W. Shumate, J r . , D. H. Smith, and F. B. Hagedorn, J. Appl. Phys. 44, 449 (1973).

9. R. C. LeCraw and R. D. Pierce, AIP Conf. Proc. ~, 200 (1971).

I0. R. J. Rijnierse and F. H. de Leeuw, AIP Conf. Proc. 18, 199 (1973).

ii. D. C. Fowlis and J. A. Copeland, AIP Conf. Proc. ~, 240 (1971).

12. A. J. Kurtzig and F. B. Hagedorn, IEEE Trans. MAG-7, 473 (1971)! W. F. DTuyvesteyn, J. W. F. Dorleijn, and P. J. Rijnierse, J. Appl. Phys. 44, 2397 (1973).

13. D. M. Heinz, P. J. Besser, J. M. Owens, J. E. Mee, and G. R. Pulliam, J. Appl. Phys. 42, 1243 (1971).

14. R. Wolfe, J. C. North, and Y. P. Lai, Appl. Phys. Lett. 22, 683 (1973).

Volo I0, No. 4 YLaTm-GARNET FILMS 311

15. G. P. Vella-Coleiro and W. J. Tabor, Appl. Phys. Lett 21, 7 (1972).

16. G. P. Vella-Coleiro, AIP Conf. Proc 18, 217 (1973).

17. R. M. Josephs and B. F. Stein, paper No. 6D-5, Magnetism and Magnetic Materials Conference, San Francisco, 1974.

18. D. C. Bullock, AIP Conf. Proc. 18, 232 (1973).

19. T. L. Hsu, paper No. 7D-4, Magnetism and Magnetic Materials Conference, San Francisco, 1974.

20. W. D. Doyle, private communication.

21. J. C. Slonczewski and A. P. Malozemoff, AlP Conf. Proc. I0, 458 (1972).

22. A. Rosencwaig, Bell Syst. Tech. J. 51, 1440 (1972).

23. R. Wolfe and J. C. North, Appl. Phys. Lett. 25, 122 (1974).