UCRL-JC-128293 PREPRINT

Advances in the reduction and compensation of film. stress in high-reflectance multilayer coatings for extreme ultraviolet lithography applications

I?. B. Mirkarimi C. Montcalm

This paper was prepared for and presented at the 23rd Annual International Symposium on Microlithography Santa Clara, California February 22-27,1998

February 20,199s \

This is a preprint of a paper intended for publication in a journal or proceedings. Since changes may be made before publication, this preprint is made available with the understanding that it will not be cited or reproduced without the permission of the author.

DISCLAIMER This document was prepared as an account of work sponsored by an agency of the United States Government Neither the United States Government nor the University of California nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or the University of California. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or the University of California, and shall not be used for advertising or product endorsement purposes.

Advances in the reduction and compensation of film stress in highreflectance multilayer coatings for extreme ultraviolet lithography applications P.B. Mirkarimi a) and C. Montcalm Advanced

Microtechnology

Lawrence

Livermore

Livermore,

Program

National

Laboratory

CA 94550

USA

Abstract Due to the stringent surface figure requirements for the multilayer-coated optics in an extreme ultraviolet (EUV) projection lithography system, it is desirable to minimize deformation due to the multilayer film stress. However, the stress must be reduced or compensated without reducing EUV reflectivity, since the reflectivity has a strong impact on the throughput of a EUV lithography tool. In this work we identify and evaluate several leading techniques for stress reduction and compensation as applied to Mo/Si and MO/Be multilayer films. The measured film stress for Mo/Si films with EUV reflectances near 67.4% at 13.4 nm is approximately - 420 MPa (compressive), while it is approximately +330 MPa (tensile) for MO/Be films with EUV reflectances near 69.4% at 11.4 nm. Varying the Mo-to-Si ratio can be used to reduce the stress to near zero levels, but at a large loss in EUV reflectance (> 20%). The technique of varying the base pressure (impurity level) yielded a 10% decrease in stress with a 2% decrease in reflectance for our multilayers. Post-deposition annealing was performed and it was observed that while the tiost in reflectance is relatively high (3.5%) to bring the stress to near zero levels (i.e., reduce by 1 OO%), the stress can be reduced by 75% with only a 1.3% drop in reflectivity at annealing temperatures near 200 “C. A study of annealing during Mo/Si deposition was also performed; however, no practical advantage was observed by heating during deposition. A new non-thermal (athermal) buffer-layer technique was developed to compensate for the effects of stress. Using this technique with amorphous silicon and MO/Be buffer-layers it was possible to obtain MO/Be and Mo/Si multilayer films with a near zero net film stress and less than a 1% loss in reflectivity. For example a MO/Be film with 68.7% reflectivity at 11.4 nm and a Mo/Si film with 66.5% reflectivity at 13.3 nm were produced with net stress values less than 30 MPa.

a) Electronic mail: Mirkarimil Qllnl.gov or [email protected]

1. Introduction In order to continue increasing transistor density on microprocessor integrated circuits to make them faster and more powerful it will be necessary to reduce feature sizes even further. 1 Extensions of current optical lithography technologies are not expected to be able to produce channel widths below 0.13 pm, and hence there is a need to develop new lithographic technologies. * Extreme ultraviolet (EUV) lithography is one of the technologies being developed for this purpose.3l4 The current designs for an extreme ultraviolet lithography tool entail the use of several optics designed to reflect light in the wavelength range of 11-14 nm at near-normal angles of incidence.3n5 The optics are expected to consist of low thermal expansion glass ceramic substrates coated with reflective Mo/Si or MO/Be multilayer films, and the figure (shape) of some of the substrates needs to be extremely precise (within 0.25 nm over a several inch diameter asphere6) in order to minimize wavefront error and maintain a high resolution. High stress multilayer coatings can deform the precisely figured substrate and increase the wavefront error. Viable methods of stress reduction or compensation need to be identified and evaluated. Previous techniques for reducing stress include (i) varying the composition of Mo/Si by varying the Mo-to-Si ratio,7-10 (ii) varying the base In this paper these methods will be considered, and a new pressure,8 and (iii) post-deposition annealing. ttll* method using buffer-layers to negate multilayer film stress will also be introduced and evaluated.

2.

Experimental

2.1 Mo/Si

and

Procedure MO/Be

multilayer

film

synthesis

and

characterization

Mo/Si multilayer films were deposited in a magnetron sputtering system similar to that described in detail elsewhere.13 MO/Be films were deposited in a new magnetron sputtering system that is similar to the system used for Mo/Si films. Two substrate holders are spun during deposition. The substrate holders are affixed to a platter which rotates over the sputter targets. The base pressure was -z 5 x 1 W7 torr, and during deposition argon gas was used and maintained at a constant pressure between 0.8 - 1.8 mtorr. The substrates used in this study were 100 mm diameter, 0.5 mm thick (100) silicon wafers and 25 mm diameter, 0.5 mm thick superpolished zerodur. Cleaning was performed by the manufacturers; however, the wafers were briefly rinsed with isopropyl alcohol while being spun before placement into the deposition systems. Reflectance measurements were performed on beamline 6.3.2. at the Advanced Light Source at Lawrence Berkeley National Laboratory. l4 These measurements were conducted at 5” off-normal incidence over the wavelength range 11 .l-12.1 nm for MO/Be and 12.4-14.2 nm for Mo/Si unless noted otherwise. The MO/Be and Mo/Si multilayer films are typically designed to have their reflectance peaks centered at 11.4 nm and 13.4 nm respectively. Reflectance changes cited in this study are given in absolute terms. For example, if the reflectance changes from 67.4% to 65.4% it is a “2%” reflectance decrease. Stress measurements were conducted on a commercial stress measurement tool that uses a laser to measure the curvature of the wafer before and after coating, and employs a modified Stoney15 equation to calculate stress values. Stress changes cited in this study are given in relative terms. For example, if the stress changes from - 400 MPa to -100 MPa it is a 75% stress decrease. X-ray diffraction (XRD) was performed at near glancing incidence on a commercial x-ray diffractometer. The bi-layer period thickness was determined by fitting small angle XRD peaks using standard analysis methods16l17 using a rotating anode with a Cu target (h = 0.154 nm).

2.2

Annealing

during

deposition

A wafer heater was constructed to enable standard 100 mm diameter silicon wafers to be annealed during deposition to temperatures from ambient to 200 “C. The heating element was a commerical76 mm diameter wafer-shaped heater consisting of thin layers of carbon encased in boron nitride. The heating element was pressed against the backside of a copper block, onto which the silicon wafers were mounted. The temperature was measured via a thermocouple mechanically affixed at the periphery of the back side of the copper block. A calibration run was performed at the beginning and the end of the series to determine the difference in temperature between the back of the copper block and the front of the silicon wafer. In order to use this heater it was not possible to spin the silicon wafers during deposition. Two silicon wafers were coated during each experimental run; the first wafer was heated while the second wafer remained at near-ambient temperatures and served as a control.

2.3

Post-deposition

annealing

Post-deposition annealing was conducted in a commercially available stress measurement tool equipped with an annealing capability. The high EUV reflectance Mo/Si multilayers used were either deposited on 100 mm silicon wafers and cleaved into 25 mm x 25 mm pieces, or deposited directly on 25 mm diameter silicon substrates. The Mo/Si multilayers were heated and cooled at a rate of 5 %/minute under ambient conditions or a nitrogen atmosphere. The stress of the films was periodically monitored during the annealing process. The maximum annealing temperatures were between 100-300 ‘C, and the annealing times were 12,9, 6, 4, and 3 hours for 100, 150, 200, 220, 245, and 300 “C, respectively. The length of time used was primarily governed by the fact that at lower annealing temperatures the stress is reduced more slowly, and takes a longer time to stabilize. Note that the time cited does not include the time to ramp-up and cool-down the samples. A MO/Be multilayer film deposited upon a 100 mm diameter silicon substrate was annealed at 200 “C for several hours, and the heating/cooling rate was similar to that used for Mo/Si annealing. c

2.4

Buffer-layer

deposition

Amorphous silicon (a-Si) buffer-layers were used to compensate the stress of MO/Be multilayers. These a-Si layers were deposited by magnetron sputtering under conditions similiar to those used to deposit a-Si layers in the Mo/Si multilayers. MO/Be buffer-layers were used to compensate the stress of Mo/Si multilayers. They were deposited under similar conditions used to produce high reflectance MO/Be films, except approximately 53 bi-layers were used for the buffer-layers while 70 bi-layers were used for high reflectance MO/Be films.

3.

Results

and Discussion

3.1 Background:

stress

and

reflectance

The synthesis of Mo/Si multilayers for use as EUV reflectors dates back many years in the literature.18 Our Mo/Si multilayer films have been optimized for high reflectance in the extreme ultraviolet spectral region. With a bi-layer period thickness of 6.9 nm, a MO fraction of 0.4, and 40.5 bi-layer periods, near-normal reflectances of approximately 67.4% at 13.4 nm are now routinely achievable. Under these conditions the measured stress of the Mo/Si multilayers is approximately - 420 MPa (compressive).lg The MO layers have a tensile stress, the Si layers have a compressive stress, and the MoSi, inter-facial layers are believed to have a compressive stresst2 We have measured the stress of 100 nm thick MO and Si films and obtained values of approximately +300 MPa (tensile) and -1300 MPa (compressive) respectively. Of course the absolute stress values for the much thinner layers within the multilayer are expected to differ from these values, particularly for the Mo.~* MO/Be multilayers have only recently been considered for use as EUV reflectors21 and the stress of the multilayers has not been reported. Our MO/Be films optimized for a high EUV reflectance have a bi-Iaye! period of 5.8 nm, a MO fraction of 0.4, and 70 bi-layer periods. This yields near-normal reflectances that are approximately 69.4%. Under these conditions the measured stress of the MO/Be multilayers is approximately +330 MPa (tensile). The MO layers are in tension and the Be layers are in compression. The measured stress of 100 nm thick MO and Be films was approximately +300 MPa (tensile) and -350 MPa (compressive) respectively. There are two points to make with respect to the difference between Mo/Si and MO/Be stresses: (i) Mo/Si and MO/Be have opposite signs for the stress (compressive vs. tensile) and (ii) even though the MO/Be stress is lower, the larger film thickness entails that it will apply a greater moment on the optic/substrate, resulting in a greater distortion of the figure.

3.2

Stress

3.2.1 Varying

Reduction/Compensation the Mo-to-Si

Techniques

ratio.

As noted previously by Nguyen et a/.719 and later by Windt et a/. 8 and Tinone et a/.‘*, the stress in Mo/Si multilayers can be tuned to zero by an appropriate adjustment of the MO fraction. However, these studies did not systematically investigate the effect that such a modification would have on the EUV reflectance which, as mentioned previously, can have a significant effect on the throughput of a EUV lithography tool. We have investigated the effect of varying the MO fraction, I-, at a constant bi-layer thickness, A, on both the Mo/Si film stress and EUV reflectance, as shown in Figure 1. Note that the MO fraction, I-, is defined as tJ ($,,,,+ tsi)p where t,.,Oand tsi are the thicknesses of individual MO and Si layers, respectively. The bi-layer period thickness, A, is defined as (tM, + tsi). It is observed that the stress can be tuned to zero at a bi-layer period thickness of 6.9 nm at a MO fraction of approximately 0.68. However, the reflectance near 13.4 nm is reduced from above 66% to 44%, which would be an unacceptable loss for a EUV lithography tool. A loss in reflectance with increasing MO fraction is expected,22 since the amount of the absorber (MO) is being increased above the optimal value. Also, the MO roughness is expected to increase with increasing thickness.

600

0.8 0.7

0.5 0.4 i5j -200 3 2 -400

0.3

g 22 8 5 z

0.2

-600

0.1 (3.23

6.4

d.5

6.6

0’.7

d.8

0.13

I? (i.e., MO fraction) Figure 1. Stress and EUV reflectance of Mo/Si multilayer films as a function of the fraction of molybdenum, a fixed bi-layer thickness of approximately 6.9 nm.

r, at

3.2.2 Varying the base pressure Windt et aL8 have discovered that the stress in Mo/Si can be significantly reduced if base pressure (i.e., background impurity level) is increased during deposition. They observed a decrease in stress from -440 MPa to -280 MPa with no measurable degradation in the EUV reflectance as the base pressure was increased into the 10m5torr range. It is unknown whether these findings are directly applicable to our efforts since the maximum reflectance of the Mo/Si multilayers used in their study was approximately 62%. The Mo/Si reflectance may be less susceptible to impurities in this reflectance regime than for our 67% reflectance multilayers. To investigate this we performed an experiment : Mo/Si was deposited in two successive experimental runs under identical conditions except for an order of magnitude difference in base pressure (5 x 10” torr vs. 5 x 10.’ torr). Figure 2 shows a plot of reflectance vs. wavelength for Mo/Si multilayers deposited at two different base pressures. Mo/Si deposited at a higher base pressure showed a lower stress of - 380 MPa (vs. - 420 MPa for the control sample) , generally consistent with Windt et a/. 8 However, the reflectance decreased from 67.4% to 65.4%, which is undesirable, particularly for such a relatively small decrease in film stress. It should be noted that a more exhaustive study is needed before it can be determined with confidence whether this technique is or is not viable for use with our high reflectance (67%-range) Mo/Si multilayers. For example, at higher base pressures the composition of the background gas may differ between our deposition system and the system used in the Windt et al. study.

Phase = 5 x 10 -6 torr Reflectance = 65.4%

12.6

12.8

13.0

13.2

13.4

Wavelength

13.6

13.8

14.0

14.2

(nm)

Figure 2. Reflectance as a function of wavelength for Mo/Si multilayer films deposited under nearly identical conditions except for the base pressure.

3.2.3 Post-deposition

annealing

(a) Mo/Si While there are a number of studies in which the effect of post-deposition annealing on Mo/Si multilayers was investigated,11*12*23-28 there are only two studiesttlt* ’In which the effect of annealing on stress was consideredill’*, and only one study in which both stress and reflectance were considered.’ t Kola et a/. 11 observed that post-deposition annnealing at 200 “C could reduce the stress from -350 MPa to -150 MPa, a reduction of 57%, with a small decrease in reflectance (from 58.5% to 57.8%). Since the maximum EUV reflectance of the Mo/Si multilayers in the Kola et al. study was significantly lower than the reflectance of the Mo/Si multilayers used in this study, it cannot be assumed that the same functional dependence of stress and reflectance on annealing will be observed. We therefore conducted an investigation of the effects of postdeposition annealing on stress and reflectance, as described below. The stress and EUV reflectance is plotted as a function the annealing temperature in Figure 3. As the temperature is increased to about 200 “C there is a 75% reduction in stress with a relatively small decrease of 1.3% in EUV reflectance. One can go to 220 “C and obtain an 85% reduction in stress with only a 1.5% reflectance decrease. Above 220 “C the reflectance loss increases, and although the stress can be reduced to zero with annealing at 275 OC , it is at a cost of 3.5% in the EUV reflectance. Thus the cost in reflectance for this technique is quite reasonable for stress reduction in the O-85% range, and becomes expensive if zero stress is required.

100

Post-deposition

Annealing

(Ex-situ)

0.69

T

0

0.67

2 z

0.66

Reflectance

-100

z

0.65

g -200 cn

0.64

F

0.63

-300

P

0.62 0.61 -50

0

50

100

Annealing Figure 3. Stress and EUV reflectance post-deposition (ex-situ) annealing.

200

250

Temperature

(“C)

150

300

350

of Mo/Si multilayer films as a function of the annealing

temperature

for

At 200 “C the results indicate a larger stress reduction (75% vs. 57%) and a larger reflectance loss (1.3% vs. 0.7%) than that observed by Kola et al. However, at 165 “C we observe a very similar reflectance reduction (0.7%) and stress reduction (56 %) to what Kola et a/. observed at 200 ‘C. The cause of the difference is unknown but could be due to: (i) a difference in the abs.olute temperatures as measured by the annealing equipment in the two studies, (ii) the shorter annealing times used in the Kola et al. study, or (iii) the dissimilarities between the Mo/Si multilayers used in the two studies. Our analysis indicates that much of this difference is likely due to a difference in annealing times between the two studies. Regardless of the source of this difference, both studies indicate that post-deposition annealing can be used to significantly reduce Mo/Si film stress at a relatively small cost in EUV reflectance.

(b) MO/Be Annealing can reduce the defect density in a material, which generally serves to reduce compressive stress. 2g Unlike Mo/Si, MO/Be multilayers have a net tensile stress and it is expected that annealing would serve to increase the (tensile) stress of the multilayer. We annealed a MO/Be multilayer film on Si(100) at 200 “C and observed that the stress increased from approximately +330 MPa (tensile) to +470 MPa (tensile). Therefore this does not appear to be a viable technique for stress reduction in the MO/Be system.

3.2.4 Annealing

of Mo/Si During

Deposition

It is not known if the figure of a precisely figured optic will be affected by annealing at temperatures approaching or exceeding 200 ‘C, and hence there is some uncertainty regarding the viability of an annealing procedure for stress reduction. There is the possibility that by annealing during the deposition process, a lower temperature could be used to achieve the same decrease in stress, and it may do so at a lower cost in reflectance than post-deposition annealing. Also, Kloidt et aL30 conducted a study of annealing during deposition for Mo/Si, and while they did not measure film stress they did observe a significant increase in EUV reflectivity by annealing at temperatures near 150 “C. However, it should be noted that the maximum Mo/Si EUV reflectivity obtained was only 46% in this study. We investigated the effect of annealing during deposition on our high reflectance (>67%) Mo/Si multilayer films. Figure 4 (a) shows the stress and reflectance for Mo/Si films as a function of the substrate temperature during deposition. Initially the stress is a relatively weak function of the deposition temperature; however, as the temperature is increased above approximately 150 “C the stress reduction becomes a strong function of temperature. Near this same temperature the reflectance also starts to become a strong function of the substrate temperature. At the highest temperature sampled, 200 ‘C, the stress is reduced by 70 %. Unfortunately the reflectance is also reduced by 3.9% relative to an unannealed Mo/Si multilayer. Figure 4(b) combines the data from Figures 3 and 4(a) and shows the stress and reflectance results for annealing during and after deposition. Post-deposition annealing yields a similar reduction in stress (75% vs. 70% for annealing during deposition) at 200 “C and does so at a lower cost in the EUV reflectance (1.4% vs. 3.9% for annealing during deposition). This coupled with the technical challenges in uniformly heating a several centimeter thick spinning substrate during the deposition process suggests that if annealing were to be used, post-deposition annealing is the preferred option.

100

Annealing

During Deposition

(In-situ)

1 0.69

$0

0.68

I2

0.67

--loo

z

0.66

Reflectance

g -200 cn

0.65

G -300 -2 2 -400

z C n

0.64

z E G

0.63

$

0.62

-500 0

I

I

I

I

50

100

150

200

Substrate Temperature

0.61 250

(oC)

(4

100 2.

0

%

-100

$i!

-200

iTj

-300

7s 2

-400 -500

1 0.69 Reflectance

-

0.65 if

I -I-I

!S)

0.64 ?iis 0.63 z CD 0.62

-/ 0.61

I,,.,,,,,,,,,

0

50

100

150

Temperature

200

250

300

(oC)

(b) Figure 4. (a) Stress and EUV reflectance of Mo/Si as a function of the silicon substrate temperature for annealing during deposition (in-situ annealing); (b) Stress and EUV reflectance as a function of temperature both post-deposition annealing and annealing during deposition.

for

3.2.5

Buffer-layer

technique

For many materials of interest for optical applications, stress can affect the optical properties. One example would be semiconductor multilayers used for light emission. 31 For the case of multilayer-coated optics for EUVL applications, stress by itself has a very minor effect on the EUV reflective properties. The main risk that stress poses is a distortion of the figure of the optic. Therefore, if one could find a way to counteract the distortion of figure caused by the EUV multilayer stress one would have a potential solution. This can be achieved by using a buffer-layer with a stress of sufficient magnitude and opposite sign to counteract the distortion due to the reflective multilayer, as shown in Figure 5. One important criteria is thatthe buffed-layer be nearly as smooth as the substrate, or it will introduce additional roughness into the EUV reflective multilayer on top, and degrade the reflectivity. 32-34 Note that this technique addresses the multilayer stress problem via a compensation of film stress, unlike the techniques described above which worked by reducing film stress.

Net film stress tuned to near zero. l-

r

Multilayer film Stress = + 0

Buffer-layer Stress = - 0

Substrate Figure 5. Schematic diagram illustrating the buffer-layer concept.

(a) Buffer-layer

technique

applied

to the

MO/Be

system

.

Since MO/Be multilayer films are under a tensile stress, a buffer-layer designed for this system would need to be compressively stressed. One promising candidate is amorphous silicon, although there are other possible candidates including amorphous carbon. We deposited a 110 nm thick amorphous silicon (a-si) film and conducted stress and atomic force microscopy (AFM) measurements on the film. The a-Si film stress was -1300 MPa and the surface roughness was approximately 0.18 nm rms (note: our typical 100 mm diameter silicon wafers have an rms roughness near 0.09 nm). This a-Si film therefore fills two important conditions: (i) it has a stress opposite in sign and of sufficient magnitude to counteract the MO/Be stress, and (ii) the surface roughness is sufficiently low, and is not expected to introduce a significant amount of roughness in the MO/Be multlayer deposited on top of it. Also, it should be noted that the AFM measurements were done on a-Si films after several days of air exposure whereas the a-Si films used as buffer-layers (for subsequent MO/Be deposition) were exposed to atmosphere for only several minutes. Hence the actual a-Si roughness prior to coating might be a little lower than 0.18 nm rms. A 400 nm thick MO/Be multilayer film was deposited upon a 110 nm thick amorphous silicon buffer-layer on a Si(100) substrate. Figure 6 shows a cross-sectional transmission electron microscopy micrograph of one of these samples. Close examination of the multilayer showed that it looked similar to MO/Be films deposited directly on Si(100) substrates, i.e., if any roughness was introduced by the buffer-layer it was relatively small. This was confirmed with EUV reflectance measurements shown in Figure 7. The reflectance and stress for

MO/Be on an a-Si buffer-layer on Si(100) is compared to a sample deposited in the same run in which MO/Be was deposited directly on Si(100). In using the buffer-layer one is able to decrease the net stress from +330 MPa (tensile) to -22 MPa (compressive) with only a 0.7% decrease in the peak reflectance.

MO/Be Multilayer

Amorphous Si Buffer-Layer (110 nm thick) Si( 100) Substrate

Figure 6. Cross-sectional transmission electron microscopy micrograph showing a MO/Be multilayer on an amorphous silicon film on a silicon substrate. The MO/Be multilayer bi-layer period thickness was 5.8 nm.

c 0.80 I

MO/Be on Si( 100). R max= 69.4%. Stress = + 333 MPa.

\ \ 0.00

I 11.2

i J/

\

MolBe on a-Si on Si( 100). Rmar = 68.7%. Stress = - 22 MPa.

I 11.4

11.6

Wavelength

11.8

12.0

12.2

(nm)

Figure 7. Reflectance as a function of wavelength for MO/Be multilayer deposited on: (i) an amorphous silicon buffer-layer which was deposited on a silicon substrate, and (ii) a bare silicon substrate. Both MO/Be multilayer depositions were performed during the same experimental run.

I

It is anticipated that a low thermal expansion coefficient material such as zerodur will be used as substrates for the EUV reflective multilayer coatings. We used the same buffer-layer scheme described above on zerodur substrates, and a reflectance of 68.9% was obtained, as shown in Figure 8. During the same run identical films were deposited on a silicon substrate, and a reflectance of 68.6%, and a net stress of -23 MPa was obtained. Therefore the net stress for the films on the zerodur substrate is expected to be near -23 MPa. The slightly higher reflectance obtained for the coated zerodur is likely due to a smoother finish for this particular zerodur substrate relative to the silicon substrate. AFM measurements yielded a surface roughness of 0.07 nm for this zerodur substrate, where our silicon wafers typically have a measured roughness closer to 0.09 nm.

0.60 0.70

MO/Be on a-S on Zerodur.

-

w 11.4

11.6

11.8

Wavelength

12.0

(nm)

Figure 8. Reflectance as a function of wavelength for MO/Be multilayer deposited amorphous silicon buffer-layer which was deposited on a zerodur substrate. (b) Buffer-layer

technique

applied

to the Mo/Si

system

on an c

It is more difficult to apply the buffer-layer technique to the Mo/Si system than the MO/Be system, because it is easier to identify ultrasmooth films which are compressively stressed than it is to identify ultrasmooth I films with a tensile stress. Also, potential candidates (i.e., smooth films with a tensile stress) are usually deposited at elevated temperatures by CVD processes, which is not a practical process for use with EUVL optics. MO/Be multilayers are a potential buffer-layer candidate for use in the Mo/Si system. MO/Be multilayers have a tensile stress, and their high reflectance indicates that they are reasonably smooth. We deposited Mo/Si on a MO/Be buffer-layer on Si(lO0) as shown schematically in Figure 9. The net stress was reduced by over 93%, to -28 MPa, when a 300 nm thick MO/Be buffer-layer was employed. The reflectance drop was only 0.9% relative to a Mo/Si on Si(lO0) sample deposited during the same experimental run, but in a different sample holder. The reflectance curves are shown in Figure 10. A small difference in the peak wavelengths are observed; however, most of the difference can be accounted for by a slight difference in the substrate holders for the Mo/Si deposition system, and not due to the buffer-layer.

Mo/Si (280 nm MO/Be (300 nm thick) Substrate Figure 9. Schematic

diagram showing a Mo/Si multilayer film on a MO/Be buffer-layer

0.80

-

0.70

-

on a substrate.

Stress = - 470 MPa.

I

13.0

13.5

Wavelength

14.0

14.5

(nm)

Figure 10. Reflectance as a function of wavelength for Mo/Si multilayer deposited on: (i) a MO/Be multilayer buffer-layer which was deposited on a silicon substrate, and (ii) on a bare silicon substrate.

c)

Practical

issues

For both the MO/Be and Mo/Si multilayer systems we have demonstrated that it is possible to reduce the stress to near zero levels using an athermal buffer-layer technique, and that the drop in EUV reflectance is very small (less than 1%). One issue not discussed thus far is the anticipated need to to control the multilayer film thickness uniformity across a several inch diameter optic to within several tenths of a percent,35 which is needed to minimize wavefront error.36 Thicker films do not preclude one from meeting these specifications, but they may make it more challenging. The amorphous silicon buffer-layer used in the MO/Be case adds only 110 nm to the total film thickness, or an increase of 27%. However, the MO/Be buffer-layer adds 300 nm to the total film thickness, or an increase of 107%. It should be noted that neither of these buffer-layers has been fully optimized, and it is expected that their thicknesses can be reduced further.

4. Future

Work

While EUV reflectance is an important criteria for evaluating the viability of various stress reduction and compensation strategies, there are other factors that are also expected to play a role. Issues currently being investigated which were not addressed in this paper include: (i) the stability of Mo/Si and MO/Be multilayers, both before and after application of the stress reduction/compensation techniques described above; (ii) the compatibility of stress reduction or compensation techniques with precisely figured low thermal expansion glass ceramics; and (iii) the stress uniformity and isotropy before and after stress reduction or compensation. Except near the edges of the coating, if the multilayer stress is isotropic and uniform the deformation of the optic will be spherical, and the most of the spherical component of the deformation should be compensable by an appropriate adjustment of the optics during the alignment process. z7 However, such methods are not expected to work for the non-spherical component of the deformation, and therefore the stress uniformity and isotropy needs to be characterized. Work is also in progress with Stanford Universit98 to better understand the evolution of stress during Mo/Si film growth, as well as the effects of deposition parameters such as the sputter gas pressure and composition. Some of these findings are expected to be discussed in future publications.

5. Summary The measured film stress for Mo/Si films with EUV reflectances of approximately 67.4% is approximately - 420 MPa (compressive), while it is approximately +330 MPa (tensile) for MO/Be multilayers with EUV reflectances of approximately 69.4%. In this work we identify and evaluate several leading techniques for stress reduction or compensation as applied to Mo/Si and MO/Be multilayer films. The results are summarized in Table 1. For the greatest stress reduction or compensation with the lowest cost in EUV reflectance, the athermal buffer-layer technique and post-deposition annealing are the leading candidates. Work is in progress which will consider other criteria such as stress/reflectance stability, isotropy, and compatibility with precisely figured substrates.

Table 1. The stress reduction and reflectance techniques discussed in this study.

loss for each of the stress reduction and compensation

Technique Athermal Buffer-Layer Post-deposition

Post-deposition Annealing

(Mo/Si & MO/Be)

Annealing (Mo/Si): At 275oC: At 2OOoC: Annealing

(MO/Be; 200 “C)

During Deposition

Varying Base Pressure Varying Composition

(MO/S; 200°C )

(Mo/Si) (Mo/Si)

% Stress Reduction

% Reflectance Loss (absolute”)

> 90%

c 1%

100% 75%

3.5% 1.3%

(42% stress increase) 70%

3.9%

16%

2.0%

22% 106% 6% 46% a) By “absolute” it is meant for example that the % reflectance loss in going from 67.4% to 65.4% is 2.0%.

Acknowledgements The authors wish to thank their colleagues in the multilayer group in the Advanced Microtechnology Program of Lawrence-Livermore National Laboratory (E. Spiller, S. Bajt, F. J. Weber, F.R. Grabner, R.D. Behymer, and J.A. Folta) for assistance and useful discussions pertaining to this study. They also wish to thank M.A. Wall for the transmission electron microscopy analysis and S. Baker for the AFM work. This work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under contract No. W-7405-ENG-48. Funding was provided by the Extreme Ultraviolet Lithography Limited Liability Corporation under a Cooperative Research and Developmen< Agreement.

References 1 P. Gargini. J. Glaze, and 0. Williams, Solid State Technology 41, 73 (1998). 2 K. Derbyshire, Solid State Technology 41, 78 (1998). 3 G. D. Kubiak and D. R. Kania, eds,OSA Trends in Optics and Photonics Vol. 4, Extreme Ultraviolet Lithography (Optical Society of America, Washington DC, 1996). 4 T. Beardsley, Scientific American 277, 15 (1997). 5 0. W. Sweeney, D. Ft. Shafer, H. N. Chapman, and R. M. Hudyma, paper in this proceedings volume. 6 D. R. Kania, D. P. Gaines, D. S. Sweeney, G. E. Sommargren, B. LaFontaine, S. P. Vernon, D. A. Tichenor, J. E. Bjorkholm, F. Zernicke, and R. N. Kestner, J. Vat. Sci. Technol. B 14,3706 (1996). 7 T. D. Nguyen, X. Lu, and J. H. Underwood, in Physics of X-Ray Multilayer Structures; Vol. 6 (Optical Society of America, Washington, D.C., 1994), p. 103-105. 8 D. L. Windt, W. L. Brown, C. A. Volkert, and W. K. Waskiewicz, J. Appl. Phys. 78, 2423-2430 (1995). 9 T. D. Nguyen, C. Khan-Malek, and J. H. Underwood, in OSA Proceedings on Extreme Ultraviolet Lithography; Vol. 23, edited by F. Zernike and D. T. Atwood (Optical Society of America, Washington, D.C., 1994), p. 56-59. lo M. C. K. Tinone, T. Haga, and H. Kinoshita, J. Electron Spectrosc. Relat. Phenom. 80,461 (1996). 11 R. Ft. Kola, D. L. Windt, W. K. Waskiewicz, B. E. Weir, R. Hull, G. K. Celler, and C. A, Volkert, Appl. Phys. Lett. 60, 3120 (1992). l2 M. E. Kassner, F. J. Weber, J. Koike, and R. S. Rosen, J. Mat. Sci. 31, 2291 (1996). I3 D. G. Stearns, R. S. Rosen, and S. P. Vernon, J. Vat. Sci. Technol. A 9,2662 (1991). 14 J. H. Underwood, E. M. Gullikson, M. Koike. P. J. Eatson, P. E. Denham, K. D. Franck, R. E. Tackaberry, and W. F. Steele, Rev. Sci. Instrum. 67 (1996); available on CD only. I5 G. G. Stoney, Proc. Roy. Sot. London A82,172 (1909). 1 6 B. L. Henke. E. M. Gullikson, J. Kerner, A. L. Oren, and R. L. Blake, J. X-Ray Sci. Technol. 2, 17-79 (1990). I7 E. Spiller, Revue Phys. Appl. 23, 1687-1700 (1988). 18 T. W. Barbee, Jr., S. Mrowka, and M. C. Hettrick, Appl. Opt. 24, 883-886 (1985). 19 The stress of the Mo/Si multilayers has been observed to decay up to 10% over the first few months, making the assignment of a generalized stress value more difficult. This will be discussed in detail in a future publication. 20 T. D. Nguyen, in Intrinsic stress and microstructural evolution in sputtered nanometer single and multilayered films, San Francisco, 1994 (Mat. Res. Sot. Symp. Proc.), p. 579-590. 2l K. M. Skulina, C. S. Alford, R. M. Bionta, D. M. Makowiecki, E. M. Gullikson, R. Soufli, J. B. Kortright, and J. H. Underwood, Appl. Opt. 34, 3727-3730 (1995). 22 E. Spiller, Soft X-Ray Optics (SPIE Optical Engineering Press, Bellingham, 1994). 23 K. Holloway, K. B. Do, and R. Sinclair, J. Appl. Phys. 65, 474 (1989). 24 D. G. Stearns, M. B. Stearns, Y. Cheng, J. H. Stith, and N. M. Ceglio, J. Appl. Phys. 67, 2415 (1990).

25 26 27 28

Z. Jiang, X. Jiang, W. Liu, and Z. Wu, J. Appl. Phys. 65, 196 (1989). Ft. S. Rosen, 0. G. Stearns, M. E. Kassner, J.-l. Koike, Y. Cheng, and S. P. Vernon, Nanostruc. Mater. 3, 195 (1993). J. M. Liang and L. J. Chen, J, Appl. Phys. 79, 4072 (1996). U. Kleineberg, H.-J. Stock, A. Kloidt, B. Schmiederskamp, U. Heinzmann, S. Hopfe, and R. Scholz, Phys. Stat. Sol. 145,539

(1994).

2g H. Windischmann, Crit. Rev. Sol. State 17, 547 (1992). 30 A. Kloidt, K. Nolting, U. Kleineberg, B. Scmiedeskamp, U. Heinzman, P. Muller, and M. Kuhne, Appl. Phys. Lett. 58, 2601 (1991). 31 E. H. C. Parker, ed., The Technology and Physics of Molecular Beam Epitaxy (Plenum, New York, 1985). 32 D. G. Steams, J. Appl. Phys. 71, 4286 (1992). 33 E. Spiller, D. Stearns, and M. Krumrey, J. Appl. Phys. 74, 107 (1993). 34 D. L. Windt, W. K. Waskiewict, and J. E. Griffith, Appl. Optics 33, 2025 (1994). 35 C. Montcalm, S. Bajt, P.B. Mirkarimi, F.J. Weber, E. Spiller, and J.A. Folta, paper in this proceedings volume. 36 A. K. Ray-Chaudhuri, K. D. Krenz, R. P. Nissen, S. J. Haney, C. H. Fields, W. C. Sweatt, and A. A. MacDowell, in Etireme Ultraviolet Lithography, edited by G. D. Kubiak and 0. R. Kania (Optical Society of America, Washington DC, 1996), p. 128-l 32. 37 A. K. Ray-Chaudhuri, S. Gianoulakis, P. Spence, M. Kanouff, C. Moen, J. Jortner, J. Friesen, and J. Schwegel, paper in this proceedings volume. 38 Work with James Freitag, Bruce Clemens, and Vidya Ramaswamy in the Department of Materials Science and Engineering at Stanford University.

Technical Information Department • Lawrence Livermore National Laboratory University of California • Livermore, California 94551