The use of a High Current Implanter for 5ElO - E12 Dose Range Implants

N L H Clarke, M T Wauk, G De Cock, Applied Materials Implant Division, Foundry Lane, Horsham, West Sussex, RH13 5PY U.K.

R M Lee, M D S Castle Applied Materials, 9020-1 Capitol of Texas Highway, Austin TX 78759 USA.

The Applied Materials PI9500 xR implanter has been designed to cover the three main areas of implanter operation; low dose, high dose and high energies up to 750 keV, thereby reducing the number of implant tools required in a production environment.

In this paper, low dose data from the extended dose range of the PI9500 xR implanter is presented. Doses down to the 5E10 cm-2 level have been investigated using ThermawaveTM measurements and implant uniformity and repeatability better than 0.4 % have been obtained. An effect due to the wafer surface condition prior to implant has been observed from the thermawave maps. Dose sensitivity data from this and other implants is also reported.

I. INTRODUCTION

The demands of modem process technology require production fabs to have several implanters; at least one high current implanter, a medium current implanter and possibly a high energy implanter for some of the more advanced processes. The ion implant tool is still one of the most costly of wafer processing systems within a wafer fab. This is a combination of comparatively high capital costs, large footprint in the cleanroom, and the necessity for a number of implanters to cover the process range. Obviously a move to reduce any of these factors is desireable.

The Applied Materials PI9500 xR implanter is very competitive in the first of these two factors and has made significant breakthroughs in the last. The system has been designed to cover a broad process envelope of implant operation; high dose, low dose and low and high energy operation (up to 750 keV for P"') in a single system, therefore significantly reducing the equipment expenditure required in a wafer fab.

This paper discusses some of the low dose process issues from the PI 9500 xR, which has traditionally been thought of as a high dose implanter. Dose repeatability at the limit of present day measurement capability, dose sensitivity data and the resulting calibration of TW units into dose units are discussed. Finally, a peculiar effect which has been observed with the lowest doses is discussed, this has been attributed to the starting material quality, this has also been observed and covered in more detail elsewhere [3].

11. EXPERIMENTAL

The majority of the implants completed here were carried out using 200 mm P<lOO> silicon wafers. Most thermawave measurements were taken using a Thermawave model TP 400 XP with some taken using a TP 200 system. In all cases the damage relaxation, oxide compensation and temperature compensation features were used together with standard mode maps.

111. RESULTS AND DISCUSSION

A Low Dose Repeatability

The repeatability in dose of an implanter is a key performance parameter of the tool. This becomes more critical in the low dose regime as implants such as threshold adjusts require dose repeatability to be better than 5%, otherwise unacceptable shifts in parametrics can occur in the fmal product [2].

Fig. 1, shows a run to run repeatability plots from a B+/5.3E11/30 keV implant over a one month period. In the two plots, data from the same set of wafers measured on the 2 different thermawave systems is displayed. The figure shows 2 interesting points. First of all, the repeatability across the data sets is excellent at this dose; being better than 0.5 % run to run at 1 sigma. Although not reported here [6 ] , the within wafer sigma remained better than 0.4 YO TW units throughout this period.

356 I I I I I I I I I I I I I 1 1 1 I I I I I I

1 / 1 1 1 / 1 1 1 1 1 1 1 / / 1 1 1 / I / I 1

I ; I 1 1 I I

mb , ,.07. 07.;. U b i k k $. d. i. i. 2* 22.n . 2 4 . 2 5 - z B - 2 ~ - 2 $ . , a b Mar Mar Mar ai Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar Mar

IMRANTWTE

Fig.1. Implant repeatability over 1 month as measured on 2 Themawave systems for a B+/5.3E11/30 keV implant.

0-7803-3289-X/97$10.00 0 1 997 IEEE 494

This repeatability is the “raw sigma”, i.e. the value including errors associated with the measurement equipment: This “gauge repeatability” can have a significant impact on the measurement of implanter performance even over a relatively short time period of 8 hours [4]. Taking this into account, the actual implant repeatability can be expected to be somewhat better than this raw value.

Secondly, continuing on the issue of metrology, there is an apparent shift in TW units between the 2 systems (around 2.7 %), alerting one to the dangers in comparing data between two measurement tools even from the same manufacturer.

The results from several repeatability measurements is displayed in Table. I below. The result for the implant discussed above in Fig. 1 shows a measurement repeatability value of 0.226 YO (1 sigma) after 25 sequential probings of the same wafer. The table shows the 1 sigma measurement repeatability values for a variety of implant conditions and displays quite considerable variation in repeatability for a number of implants. Many of the lowest values occur in the 500 - 700 TW

TABLE I THERMAWAVE MEASUREMENT REPEATABILITY

Implant Condition Waf Thermawave data No of size TW TW maps

[mm] Mean [“/I B12E13132 keV

B14E12115 keV P+++Il E l 31600 keV

P+++Il E l 31600 keV

P+13E13/10 keV B+15E11/25 keV Asl5E10130 keV

811 e l 4/80 keV ASl3El1180 keV

As/5E10130 keV Bl4E12l15 keV B12E1515 keV AS11 E14180 keV B l l E14180 keV B15E15180 keV

AS15E15180 keV

150

200 200

200

200 200 200 150

200 200

200 200 200 200

200 200

716.3

51 5.3 1017.0

1015.8

684.0 281.8 151.1

1167.1

496.4 152.9

518.5 1104.0

11919.0 1 153.3 3960.9

25973.3

0.19 50

0.09 50 0.176 50

0.179 25

0.115 25 0.226 25 0.595 25 0.164 25 0.183 25

0.433 25 0.147 50 0.046 25 0.206 25 0.086 25

0.159 25 0.104 25

band, which is a region of poor sensitivity to dose variation, as will be seen in section 111-B.

Fig. 2, shows a similar repeatability plot to Fig 1. at the very extreme of the dose envelope for the PI 9500xR at 5E10 cm-’. This shows the TW results from 8 implants each with a single wafer. Beams of boron were tuned in-between

each of these implants to simulate a production type scenario with different beam set-ups. The beam current used for this series of implants was - 1.1 pA. The system is capable of beam currents in the range from 1pA to 22 mA with a less than 5 minute tune time.

149 ~ 0.3

10.25 - ~~ 0.2 8

.~ 0.05 5

.- b

-0 .~ 0.1 5 .. 0.15 E

Fig. 2. As/5E10/30 keV TW run to run repeatability plot over a series of 8 implants

The result is an implant repeatability of 0.38 % and an average within wafer unifomiity better than 0.3 % in TW units, again this includes metrology contributions.

In this particular data set, the wafers were pre-annealed before implant (1050 “C, 120 s). Section 111-C covers the reasons for this in more detail; sufficient to say we have observed effects from the Sti3rt material of the virgin test wafers which can have an imlpact on the results of the post- implanted wafer.

B Dose Sensitivity

One of the main drawbacks with the thermawave technique is the semi-quantitative nature of the TW measurement. The output data of the tool (TW units) have no physical relation to device parametrics, which is in contrast to sheet resistance. If data on dose is required, a calibration exercise has to be carried out. This task will have to be carried out for each species, dose, energy, different oxide thicknesses and beam currents, as the calibration curve is sensitive to many effects [ 5 ] .

A very useful parameter 1.0 consider when making dose judgement for an implant is the so called thermawave sensitivity. This is defined as:

( 1 ) YO Change in TW value

% Change in dose S =

This parameter is arrived at by completing a series of implants at doses f 10 YO and f 20 YO of the dose at which

495

the sensitivity data is required, and then plotting the change in measured TW units against the as-dialled dose in the implant recipe.

An example of this type of plot for an As+/5E10/30 keV implant is displayed in Fig. 3. It shows the measured TW value versus the selected dose in the implant recipe. A least- squares fit of the dataset is overlaid and a value of sensitivity to dose of 0.55 is arrived at by using equation 1.

0.8 -

0.7 .

Sensitivity factor = 0.55

160

A

A

.- Y) 150

140

130

0.1 -

I . .

120 3.50E+10 4.WE+10 4,50E+10 5.00E+10 5.50E+10 6.00E+10 6.50E+10

Measured Dose (cm-2)

A A Van’an Data PM4T Data AAA

Fig. 3. As/5E10/30 TW Sensitivity plot.

Fig. 4. shows a plot of TW sensitivity data gathered from 2 sources; Varian data [l], plotted as small triangles, and Applied Materials data as open squares.

Fig 4. Themawave sensitivity data from Varian and Applied Materials

This plot is a large collection of TW sensitivity values arrived at in the manner as described for Fig. 3; each point culminating from a curve of at least 3 implants. It is important to note that there is a great diversity in the implant conditions which provided these data points.

It can be seen that the plot has a characteristic asymmetrical “V” shape with a minima around the 600 TW unit area. This is an area in which the TW system is quite insensitive to dose. It is well known that the sensitivity to dose fluctuation for an implant is dependent on the implant conditions themselves [ 5 ] . In one regime of TW values an apparently good result in repeatability and uniformity can be quite insensitive to changes in dose and would therefore be a poor choice as an implant monitoring device. This is particularly important to bear in mind when deciding implant conditions for statistical process control (SPC) trending of an implanter as the aim of such an implant is to pick up true dose changes associated with the implanter.

C Wafer Start Material Effects

During the course of this work it became increasingly obvious that the TW value of the virgin test wafer had a significant effect on the post implant measurements. This is dramatically demonstrated in Figs 5a and 5b.

Before implant: 54.1 TW units +/- 5.78 %

After implant: 165.9 TW units+/- 1.38 %

Figs. 5a and 5b. Shows TW measurements of the same wafer before and after an As/5E10/30 keV implant

496

Here the TW measurement was made on the same wafer before and after the As/BE10/30 keV implant. The pattern from the virgin test wafer and indeed the non-uniformity (wafer (3 = 1.38 %) can be seen to be clearly carried forward to the post implant map. This has been observed elsewhere [3] and is thought to be attributed to the wafer polishing procedures of the silicon manufacturer.

This effect can be avoided by giving the wafers a short RTA anneal before the implant (1050 “C, 120 s). Fig. 6 shows the effect of pre-annealing wafers at a higher dose across a series of implants. Each implant consists of 2 virgin prime wafers; 1 of which was annealed and 1 un-annealed.

525 ,

520

515

510

505 3 500

E ::: 485

480

475 456-01 457-01 458-01 459-01 460-01

LOT NUMBER

Fig. 6. B/4E12/15 keV run to run repeatability as a function of wafer pre-implant anneal condition.

The effect of annealing the wafers before implant has the effect of a 5 % shift in TW value; even at this moderate dose and TW level. The calculated repeatability is also seen to improve slightly after the wafers have been pre-annealed.

Other data not presented here [6] has shown that the cross-wafer uniformity of the previously annealed wafers is also significantly improved after implantation as compared to un-annealed wafers after implantation. This can be seen to some extent in comparing the post implant sigma of Fig. 5b. and the wafer uniformity plot of Fig. 2. Here the post implant sigma of pre-annealed wafers is around 0.2 %, whereas if no pre-annealing is carried out on the wafers as in the case of Fig. 5b, the sigma is 1.38 YO. The TW shift effect of pre-annealing the wafers at the 5E10 dose level is now 15 % as compared with the 5 % value of Fig. 6 . for the B+/4E12/15 keV implant condition.

Iv . CONCLUSIONS

We have seen good low dose repeatability data on this traditionally high current machine. This is has been shown

at some of the lowest doses attainable even for medium current implanters. As one nears the lowest doses available in a production environment, we have observed that the measurement ability also begins to impart significant levels of error in the implant repeatability.

Thermawave sensitivity data was discussed, and the importance of the choice of operating regime for low dose SPC type monitoring raised as an important consideration. This has quite important ramifications on the detection sensitivity to dose variation filom the ion implanter.

As the lowest doses are approached the residual polishing damage of the wafer substrate has been observed to have an impact on TW repeatability of these implants. We have seen that a pre-anneal of the wafers prior to implant can help to reduce complications due to sitart material quality.

ACKNOWL,EDGMENTS

I would like to thank Gerard O’Halloran for useful discussions with some of the TP 200 data. I would also like to thank Ron Bellchambers and Andy Allen for all the extensive implantion work they did that made this possible.

REFEFLENCES

[l] R. Callahan, Vuriun Product Bulletin #1163, Varian Implant Systems, 35 Dory Rd, Gloucester MA. Jan 22 1992.

[2] M.A. Wedman, W.L Smth, “Thermal-wave measurements of ion implanted silicon”, Nucl. Znstr. and Meth. B21, (1987), 559 - 562.

[3] R. M. Lee et al, “Thermawave and substrate considerations for mid E10 implants”, these proceedings.

[4] D.E. Kamenitsa, R. B. Simonton, “Sources of variation in Thermawave measurements of ion implanted wafers”, Nucl. Znstr. and Meth. B74 (1993) 234 - 237.

[5] B. J. Kirby, L.A. Larson, R. Y. L.iang. “Thermal-wave ,measurements of ion implanted silicon”, Nucl. h t r . and Mefh. BZI, (1987) 550 - 553.

[6] N. Clarke. Unpublished

497