New Method for Monitoring Implantation by UsingParticle Inspection

Chia-Hung Sun, Jun-Wei Gong, Ta-Yung Wang, Chung-I Chang, Tings Wang, /ProMOS Technologies Inc.Production Technology Division, ProMOS Technologies Ins., Hsinchu, Taiwan, R.O.C.

Phone:+886-35798308-4562 Fax:+886-35663300-4562* Abstract or junction depth. However, this kind of analysis needs the

The paper provides a new method for monitoring annealing process which made the impurity bond with siliconimplantation conditions. It offers a non-destructive and atoms and is only for blank wafer. Those blank wafers areeconomic method to estimate implantation conditions. A group difficult to reuse for implantation monitoring.of standard wafers are prepared. At first, the photoresist layer TP utilizes a pump laser causing localized heating towas coated on the bare wafer, then implanted with split against the surface of the wafer. The localized heatingconditions of different parameters including of energy, dosage & p Jettilt angle. induces a series of thermal & plasma waves. A probe laser iS

By using particle inspection, the intensity of scattering directed at a portion of the wafer that has been illuminated bylight could be detected before and after implanting. From the pump laser. A detector records the intensity of thecalculating the bias of the intensity of scattering light between reflected probe laser to monitor the surface changes afterbefore and after implanting, calibration curves were plotted with implanting. TP is popular inline monitor for implantation andthe bias of intensity versus the split conditions of the three can be applied on product wafer measurement. But it seems toimplantation parameters. With the new method, the be non-sensitive for implantation energy and angle.implantation condition can be estimated by inspecting the WAT is the most detailed analysis about device electricalintensity of scattering light. As the result of experiment, the new performance. It can monitor single or multiple implantationmethod allows good selectivity in dosage and energy ofimplathdation good selectivity in dosage and energy of

processes, but cannot be tested until the device is completed.For this reason, it can't be used as a real-time monitor for the

* Keywords implantation conditions.Implantation, photoresist, Atomic Force Microscopy (AFM), The implantation process is the key manufacturingsurfscan, scattering light, Metrology process and correlates with device performance strongly. If

the implantation condition does not correspond with the* INTRODUCTION device simulation, all electrical properties will probably fail.

In the semiconductor manufacturing process, How to monitor the implantation process is a topic of interestimplantation is used for modifying electrical properties of to the semiconductor industry at present.devices by bombarding silicon wafer with some specific ions. This paper provides a method for real-time & non-Moreover, by adjusting the implantation conditions of destructive monitoring of implantation conditions. At first, wedifferent parameters such as ion concentration, implantation prepared a group of standard wafers that were coated withenergy, or implantation angle etc., it can manufacture photoresist (PR). These standard wafers were implanted withdesignated devices with particular electrical properties. split conditions of different parameters including energy, dosage &Therefore, how to control the implantation conditions in real- tilt angle. After implantation, the surface property oftime is very critical. photoresist is changed as a function of the impant. In the

The traditional analyses for monitoring implantation viewpoint of optical measurement, the surface propertyconditions include wafer acceptance testing (WAT), secondary variation from split implantation conditions caused differention mass spectroscopy (SIMS), Therma-Probe (TP) and four- level of scattering light intensity. In this paper, we usedpoint probes. SIMS utilizes primary ions with sufficient surfscan to detect intensities of scattering lights. Further, aenergy to bombard the surface of a test sample. Then the calibration curve could be plotted by measuring the physicalsurface atoms or molecules will be scattered which are called property.secondary ions. At last, the compositions and atomic The real-time monitor method not only is a non-distributions in the vertical direction of the testing sample's destructive analysis, but also saves cost, because the standardsurface after implantation can be received by detecting the wafers can be reused by removing the photoresist on thesecondary ions with mass spectrometer and quantitative & wafer after the monitoring is finishing.qualitative analysis. SIMS can be applied on product wafers,but it is time-consumed and expensive. Worst of all, it is * Theorydestructive analysis. The four-point probe method contacts the The photoresist composes of resin, sensitizer andsample and applies electric currents between the outer pins to solvent. It is a kind of material which is deformed easily. Thethe inner pins to measure the voltage between them. From the assumption can be observed from Atomic Force Microscopyformula Rs=k*(V/I), the sheet resistance ofthe sample will be (AFM) scanning. The results are showed as Fig. 1. Fig. 1 (a)obtained. In general, the different implantation conditions will and (b) are the cross-section images of photoresist surfacescause different sheet resistance due to different concentration

978-1-4244-1965-4/08/$25.OO ©B2008 IEEE 162 2008 IEEEISEMI AdvancedSemiconductor Manufacturing Conference

before and after implanting respectively. The profile becomes particle inspection (Surfscan is the main tool for measuringrougher after implanting than before. intensity of scattering light in this study) to obtain a baseline

scattering light intensity. After that, each standard wafer was(a) (b) implanted with split conditions of corresponding parameters

(ion, energy & dosage). In the experiment, the five split.................. mconditions 1.

1.il() ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~_____iTable. 1: The split conditions of the experiment

' 2 3 A 0 . f h1 .trt tB i M -ItgItem Ion Energy (Kev) Dosage (/cm2)

Fig 1: The AFM section images of photoresist Surface 1 Nitrogen 12 2.5E14(a) before imp. 'b' after imp. 2 BF2 80 4E13(a) before imp. (b) after imp. 3 Arsenic 140 6E12

4 Phosphorous 10 3E13From the above experiment, we can presume that the 5 Boron 80 5E13

surface of photoresist is damaged and roughened by Twist & Tilt angle: 0implantation ions.

Furthermore, the spectra of spectrometer and spectroscopic After implanting, the later scattering light intensity forellipsometry have been collected before and after implanting. each standard wafer would be detected again. The laterThe difference of the spectra is very obvious and the result is scattering light intensity of these five standard wafers allshown in Fig. 2. As the result, the optical property of the increased after implanting. Fig. 3 shows the result byphotoresist would be changed after implanting. The different calculating the bias between later and former scattering lightimplantation condition would cause different level variation intensity. The following experiments are mainly focused onof those physical properties. By quantifying the physical Boron.property, the calibration curve can plotted for monitoringimplantation conditions. --

(a) SEI

eq iv i*.............................

(b) (c)-- 1 ' -__~~~~~-A t~ Fig. 3: The test intensity bias of scattering light

ft~~~~~~~~~~~~~~~~ fe q

(2) A calibration curve of~ implantation energy:First, five standard wafers were implanted with split

Fig. 2: The spectra of test wafer before and after implanting split conditions were 80 Key, 110 Key, 140 Key, 170 Key, and(a) Spectrum of spectrometer 200 Key, and the implantation ion was Boron, the(b) (c) tanX , cos A\of spectroscopic ell1ipsometry implantation concentration was 5E13/cm2& tilt angle was 0°.

A calibration curve was plotted by calculating the bias* EXPERIMENTAL between the later and former intensity. The bias of scattering

(1) The influence of different implantation ions: light intensity was relative to the change of the photoresistThe surface roughness will change the incident angle and physical properties corresponding to different implantation

the different optical properties will vary the optical path conditions.within the film structure. So the intensity of scattering light The calibration curve is shown in Fig. 4 wherein the Xwould be influenced by different levels of one or more axis is implantation energy and Yaxis is the standard intensitylocalized non-uniformities and the optical properties. From bias of scattering light. Fig. 4 shows that the implantationthe above section, the surface roughness and optical properties energy correlates positively to the standard intensity bias.of photoresist will be changed after implanting. Therefore, theintensity of scattering light from photoresist will differ too. (3) A calibration curve of implantation concentration:

In this study, five standard wafers were coated with a The nine standard wafers were implanted with differentphotoresist first, and then were soft baked to remove the concentrations, 5E1 1/cm2, 1E12/cm2, 2E12/cm2, 5E12/cm2,solvent in the photoresist. Next, the scattering light intensity l1E13/cm2, 2E13/cm2, 3E13/cm2, 4E13/cm2, S5E13/cm2 and theof the photoresist from each standard wafer was detected by

163 2008 IEEE/SEMI AdvancedSemiconductor Manufacturing Conference

implantation ion was Boron, the implantation energy was 80 Fig. 6: A calibration curve of implantation tilt angle with theKey & tilt angle was 00. Kev~~~~~ & itage a 'ntensity bias of scattering light (Implantation energy:Following the same process, the calibration curve is inten b ofcating ligt (mp nshown in Fig. 5 wherein X axis is the logarithm of 80 Key & concentration: 5E13/cm2)implantation concentration and Y axis is the standard intensity (5)The design of experiment (Taguchi Method)bias of scattering light. This figure indicates that implantation The main implantation parameters are energy,dosage has a positive correlation to the standard intensity bias. concentration, beam current and angle. In order to study

(4) A calibration curve of implantation tilt angle: which parameter is the most sensitive, we utilize a kind ofThe five standard wafers were implanted with different design of experiment (Taguchi Method) for analysis. From

tilt angle, 50, 15°, 25°, 350, 45°and the implantation ion was the same process, the relationship between intensity bias andBoron, the implantation energy & concentration were 80 Kev implantation parameter can be found.& SE13/cm2, respectively. The calibration curve is shown in Four control factors were defined in the TaguchiFig. 6 wherein the X axis is implantation tilt angle and Y axis Method. The four factors were energy, concentration, beamis the standard intensity bias of scattering light. According to current and tilt angle respectively.Fig. 6, it is a negative correlation. This indicates that the The experiment selected L9 (34) orthogonal array tosmaller tilt results in stronger impact on the surface of the es,ablish yn experiment time. Thethree levelsphotoresist, because the vertical vector of energy of factors, because it saved experimental time. The three levelsimplantation ions decreases as the tilt angle increases. and four factors were A factor: energy (80, 140& 200 Key), B

I I I I I I I _ ~~~~~~~factor: dosage (2E13, 3E13 and 4E13 atomlcm3), C factor:beam current 200 300 and 400 mA) and D factor: tilt angle

orthogonal array, nine groups models of finite elements wereestablished for simulation, as table. 2 showed.

i

_ ~~~~~~~~~~~~~~~~Table. 2: Plan table of the simulation experiment

IExp. No Energy Dosage I Beam Currenti| Tilt Angle11 80v2E13 200 5_AC0 80 3E13 300 25

Fig. 4: A calibration curve of implantation energy with the 31T 802 4E13o 400 |r45intensity bias of scattering light (Implantation 4 1140 2E13 300 45concentration:5E13/cm2 & tilt angle: 0°) 140 33 0 5l l

- 6 140) 4E13 200 25

7 200 2E13 400 25

8 200 3E13 200 45

_ 05 9_u.u 200 4E13 300 5.....4... Using the bigger-the-best concept of the Taguchi Method

'0.3 we estimated the vau (S/N) and prformed vriance analysis.

too small, so pooled them in error and calculated theconfidence to check which factor were critical. Theconfidences of these four factors were A: 99.95%o, B: 96.33%o,

~~~~~~~~~~~~~~~C: 63.12%o, D: 45.39%o. Table 3.and Fig. 7 are the factorFig. 5: A calibration curve of the logarithm of implantation response tables and factor response graphs of S/N ratio. The

concentration with the intensity bias of scattering light purpose lies in understanding the influence of the quality(Implantation energy:80 Kev& tilt angle:0°) characteristic of each factor and combination of optimum

_ ~~~~~~~~~~~~~~~design.We defined the confidence of critical factor was 9000at least. Through the above analysis result, energy (A) and

difference of standard intensity Sbias of scttring ligh is'_ ~~~~~~~~~~~~obviouswhen implantation energy or dosage increase

=, 0.7 ----.- ~~~~~~~~progressively, but it's not obvious when implantation tiltX W ~~~angle increase progressively

Table. 3: Factor response tableFatr Lev4el 1 Lev4el 2 Level 3 Variance confidence Effect rate

........................nergy -8.44.-4.38.-151.3 .92.9.9...88.9

|Dosage |-5.807 |-4.819 |-3.727 |3.25 |96.33%0| 7.16|IBeam Current -5.231 -4.687 -4.434 0.50 63.12%0o pooled|

164 2008 IEEE/SEMI AdvancedSemiconductor Manufacturing Conference

Tilt Angle -4.702 -4.533 -5.118 0.27 45.39%0 pooled Fig. 9: The curves of implantation tilt angle with TW value &Error - - - 0.38 - 3.85 the intensity bias of scattering light (Implantation

energy: 80 Kev, concentration: 5E 13/cm2)

Fig. 10: The curves of the logarithm of implantationFig 7: Factor response graphs of S/N ratio concentration with TW value & the intensity bias of

scattering light (Implantation energy:80 Kev)(6)Compared with TP500. (7) Different types of photoresists.

Tp500 is a common tool for monitoring implantation This method employed the change of intensity ofconditions by measuring the TW value. First, two groups of scattering light from different photoresists to monitorwafers were prepared. One was coated with photoresist on it implantation conditions, wherein the photoresist can be I-linethen implanted, the other was implanted directly. The split photoresist or deep ultraviolet light photoresist. In thefive conditions of each group were 80 kev, 110 kev, 140 kev, embodiment, three different type of photoresist were used to170kev, 200 kev and all dosage were 5E13/cm2, tilt angle analyze the variation. The curves were plotted on the basis ofwere 00. Fig. 8 shows the trend of TW and the standard the procedures that were_defined above, and the implantationintensity bias of scattering light with implantation energy by energies were 80 Kev, 140 Kev and 200Kev, the implantationmeasuring from TP500 & Surfscan respectively. Next, the concentrations were 5E13& tilt angle was 00. The curves aresplit condition was implantation tilt angle. The five definite shown as Fig. 11. It shows that even if the differentangles were 50, 15°, 25°, 350 & 450 and the implantation photoresist was used, a calibration curve can be establishedenergy and concentration were 80 Kev & 5E13/cm2. The last for monitoring implantation conditions.split condition was dosage. The five definite implantationconditions were lE13/cm2, 2E13/cm2, 3E13/cm2, 4E 3/cm2, DUV5ElI 3/cm2 and the implantation energy were 80 Key. LeA

The result of two groups is showed at Fig. 9 & Fig. 10..

~~~~~~~~~~~~11Fig.11: The curves of implantation energy withTWvle&Hwvr htrss utb rcse ythein otenit

]T500 bias~~~~~~~~~.of.... scatterng.ligh.betwee.differnt.photresist

Fig.~ ~~ ~~~~~ ~~ ~~~rcs 8:Tecreffipattoneeg ihT au owevr photorestimusatigbe poesse by stheyphtoethe intensity bias of scattering light (ImplantationprcsflwboeimanngIn rdrtsuythinfluence of photo process flow on this study, the steps ofconcentration: 5E13/cm2 &tilt angle: 0°) exposure, post-exposure bake, and development were added

after photoresist coating. Finally, the implantation and the_ _ ~~~~~~~~~~~~biasof scattering light intensity were be performed &

|_ ~~~ ~ ~~~~ ~~~~~ifany additional process is used, a new calibration curve canF~~~~~~~~~~~~~~be plotted and used as a new monitoring standard.

jjjjjj .-iiiiiiiii-l- ..............................................................

165 2008 IEEE/SEMI AdvancedSemiconductor Manufacturing Conference

-4--lid! ptoocess

--PR coatiAgo.nly

Fig. 12: The curves of implantation energy with the intensitybias of scattering light between non-development &development (Implantation concentration: 5E1 3/cm2.)

* CONCLUSIONSThe three calibration curves can be used to monitor the

implantation tools & conditions. For example, the bias ofscattering light intensity of the test wafer with photoresistcoating was calculated by the later and former scattering lightintensity. We can estimate the implantation condition bycorresponding to the intensity bias of scattering light fromrespective calibration curve. If the deviation between thepredetermined value and estimative value is larger, thecondition of implantation will be judged as abnormal. On thecontrary, if the deviation is small, the implantation conditioncan be matched to setup and the product wafer can beperformed with this kind of condition. The new method takesadvantage of reusing test wafer because the coated photoresistcan be removed easily and silicon substrate under thephotoresist will not be damaged. Therefore, the test wafer canbe reuse again by just coating a photoresist layer.

TP seems to be non-sensitive for implantation energy andangle. But it has same tread and sensitivity in dosage likesurfscan to put in use for the method. However, the laser spotof TP is small enough to measure test key on product wafer.

Finally, the different type photoresists performed withthe same split implantation conditions can obtain differentcalibration curves where the trends are similar. And if extrastep of exposure, post-exposure bake and development afterthe step of soft bake are performed, then other calibrationcurves can be established too. Whenever a new photoresist orany additional process is used, a new calibration curve has tobe plotted and used as a new monitoring standard.

References:[1] D.R. Rohner "Method for monitoring the performance of an ion implanterusing reusable wafers" US Patent 5861632 (1997).[2] L. Li, "Use of Fourier series in the analysis of discontinuous structure," J.Opt, Soc. Am. A 13, 1870-1876 (1996).

166 2008 IEEE/SEMI AdvancedSemiconductor Manufacturing Conference