Kinch, M. Riene and R. Ruth, J. B 10, Bajaj, Phys. Kumar and S. Agarwal, , p. Bahir, Phys. Agarwal , p. Bajaj, W. Tennant, R. Zucca and S. Irvine, Semicond. Bajaj, L. Bubulac P. Newman and W. Tennant, J. A 5 5 , Tennant and P. Newman, J. A 6 4 , Siliquini, J. Dell, C. Musca, E. Smith, L. Faraone and J. Biswajit Behera. Dancen Oo. Mamapaiya Haari. Yuanda Wen. Jayanth Sriranga. Muhammad Shamaim. Muhammad Uzair Arshad. Anshul Gour.
Dan Joshua Espina. Kay White. Aiman Asyraaf. Yousef Bobadilla. More From Navdeep Singh. Navdeep Singh. Aditi Verma. Lex Francis. Shivam Rai. Phase Discriminator- Foster Seeley Discriminator. Popular in Materials. Ivy Joyce. Joon Eeo.
Hence one can make the ratio of formula [3] applied to test sample 1 containing a junction region 5 in the substrate 2, in this embodiment in the form of box-like doping profile, and formula [3] applied to an reference sample containing the the substrate 2 without the junction region 5, i. This ratio R a is then when the laser beam spacing d is large enough.
In formula [4], the excess carrier concentration in the lower layer 4 and its decay length are assumed to be independent from the upper layer 3. This is certainly the case if the inactive doping concentration in the upper layer 3 is not too high.
The formula [4] express that the ratio of the signals measured respectively on the test sample 1 and on reference sample without the junction 5 is proportional to the junction depth X j of the test sample 1, if the pump and probe laser 7,8 are sufficiently spaced. For a large majority of the samples 1, the ratio R a converges to a value between between -1 and 1. This saturation value is then used to obtain the absolute value of the junction of the test sample 1 using formula 4.
Because of the cosine relationship between the ratio R a and the junction depth X j as the expressed in formula [4], the effective ratio of the measured PMOR signals should indeed have a value between -1 and 1.
The fact that the measured ratio obeys this relationship proves that the assumptions made to derive the extraction procedure in section 1 and 2 are indeed correct. One can therefore check the correctness of the method when applied on a particular sample by checking whether the measured ratio R a of the test sample PMOR signal to the reference sample PMOR signal lies between -1 and 1. It is typically for the deepest junctions that the ratio R a did converge for the used range of the laser spacing d, most likely due to the larger impact of inactive dopants in the upper layer 3 on the substrate 4 level thereby impacting the PMOR signal representative of only the substrate 2.
This asymmetry can be explained by the introduction of free carriers into the native oxide which is present on the exposed surface of the Si substrate 2 during the PMOR measurement. This native oxide passivates the exposed surface of the silicon substrate 2. In section 4, it is shown that this charging effect can be advantageously used to reduce the sensitivity of the PMOR measurement to the properties of substrate 2 surface thereby making the above developed model more robust leading to enhanced reproducibility and accuracy of the extraction method.
The correlation is very good for all samples, even for the double boxes of CVD with the deeper junctions. In figure 5 the known SIMS junction depth was therefore used to decide which of both formulas was to be used.
However in both cases, the technique gave unacceptable results. As a result both plasma components remained present in the PMOR signal and formula [4] could not be used.
Hence when selecting the parameters of the PMOR measurement the optical frequency of the probe laser beam can be selected to ensure that the PMOR signal is different from zero. In particular, the extension to annealed implanted profiles is discussed.
Rosseel, et al "Impact of multiple sub-melt laser scans on the activation and diffusion of shallow Boron junctions "IEEE International Conference on Advanced Thermal Processing of Semiconductors, , hereby incorporated by reference in its entirety.
Three series of samples were studied implanted uniform over their surface respectively with B only 0. Samples form all three series were then laser-annealed for different temperatures and different times by scanning for each conditions a laser beam over a part of the surface sample.
For each of the three implant series 14 different annealing conditions were obtained which corresponded to 14 different activated junction profiles.
As each implant condition was done on a single wafer 2, the substrate signal obtained for the different activated junction profiles should indeed be the same.
First, a unique SIMS junction depth was more difficult to define in these implanted profiles due to their finite slope contrary to the box-shaped doping profile of the samples used in sections 1 and 2.
It was therefore proposed to take as measure for the junction depth X j the SIMS junction depth at 10 20 cm Second, as a reference signal without junction 5 was more difficult to obtain, the relative junction extraction method corresponding to formula [6] was preferred above the absolute junction extraction method corresponding to formula [4]. For each series and annealing temperature, the sample which was annealed thrice, i. The relative junction extracted using formula [6] gives a very good agreement with the measured SIMS junction depths.
In order to obtain a metric of the absolute value of this extracted junction, the SIMS junction depth is used as reference junction depth in formula [6]. In particular, all curves converge towards the SIMS junction depth measured on the sample which was annealed seven times.
Notice the non-monotonic behaviour for the low temperature annealing. This does not seem to be physical and could be attributed to the fact that the reference sample having 3 laser anneal scans was not sufficiently activated.
These defects have a negative impact on the formation of the plasma components. Just like for box-like active doping profiles, it can determine the relative junction depth or, if one has an equivalent substrate available for reference, even the absolute junction depth. The accuracy of the methods depends on how well the two major assumptions made when developing the formulas [1] to [6] are verified. When using TP on a silicon sample, the thermal component should cause no problem if there are not too many defects in the substrate 2.
This actually implies two additional requirements. This is the case if a sufficient number of doping atoms is activated. The extraction methods disclosed should therefore be preferentially used on well annealed structures. Second it requires that all substrates 2 need to be the same from TP point of view.
This behaviour was unexpected from theoretical point of view since the only relevant parameter is the high-injection ambipolar diffusivity which should be independent from the doping type. This is observed experimentally as a saturation of the signal. In conclusion, the technique will be more accurate if the surface is passivated.
One would indeed expect that saturation the PMOR signal should be the same on all type of substrates whatever the doping type. In conclusion, the technique should be more accurate if the surface is passivated in whatever way. This passivation technique is sometimes referred to as field-induced passivation. The electrical field reduces the concentration of one carrier type at or near the surface, hence reducing the recombination at the surface where the PMOR measurement is performed.
Uncontrolled surface recombination may strongly interfere with the carrier recombination, and hence with the excess carrier concentration, in the underlying layers such as the doped layer 3 and the layer 4 or bulk 2.
One can use an external voltage or one can precharge the surface. One can form one or more dielectric layer 19 on this surface which dielectric layers then have an inherent charge density sufficient to prevent recombination near that charged dielectric layer. An example of such charge is AlON which is known to have high negative charge density. One can also reduce the surface recombination. If a third laser is provided in the PMOR apparatus this third laser beam can prescan the surface to which the probe laser beam 7 will be focused such that charging of the surface to be measured is done prior to the step of performing a PMOR measurement on this surface.
In addition one perform the PMOR measurement in an ambient which dissociates into molecules such that upon providing energy by the third laser beam these molecules will bind to the surface to be scanned by the probe laser beam 7. United States patent application US , hereby incorporated by reference teaches the principle of corona charging. By depositing ions generated by corona discharging, a small electrical field is built over the native oxide which prevents the mobile charge carriers to reach the surface.
The relative determination of the junction depth with respect to e. Hence fast high resolution maps showing the relative variation of the junction depth over an area are therefore feasible. The ratio of the signals should indeed saturate at a value between -1 and 1 when the laser spacing d is sufficiently large.
The first derivative of formula [3] with respect to junction depth gives. Notice that these reproducibility values include both the tool reproducibility and wafer-to-wafer reproducibility. If one wants to determine the absolute value of this junction depth 5, the reference PMOR signal is either only representative of the lower layer 4, i. If one wants to determine the relative value of this junction depth 5, the reference PMOR signal is representative of the lower layer 4 and the junction 5.
The reference PMOR signal can be obtained in different ways. This substrate can comprise additional doped regions 11' 11 having the junction 5' with junction depth X ' j. These regions 11', 11, 12 can be spaced apart by an isolation region When manufacturing semiconductor devices this isolation region 10 is typically the field oxide or shallow trench insulation region.
Due to process variations the activated doping profile between these doped regions 11, 11' might vary resulting in a variation of the junction depth X j , X' j from one region to another. By performing the PMOR measurement on one of this doped regions 11, 11' at an sufficient offset d between the probe 8 and pump 7 laser beam, one obtains a PMOR signal in which the layer 3 plasma wave component and the thermal wave component are essentially absent and which substantially depends only on the junction depth X j and the excess carrier concentration in the underlying layer 4.
This substrate further comprises at least one undoped region By repeating the PMOR measurement on one of this undoped regions 12 at an sufficient offset d between the probe 8 and pump 7 laser beam, one obtains a PMOR signal in which the layer 3 plasma wave component and the thermal wave component are essentially absent and which substantially depends only on the excess carrier concentration in the underlying layer 4.
The ratio R a of both PMOR signals will yield the absolute value of the junction depth X j of the measured doped region In this approach the optical and physical parameters of the substrate 2 determining the PMOR signal are the same for both the reference sample and the test sample as the doped 11 and undoped 12 regions are formed on the same substrate 2.
This substrate can comprise additional doped regions 11', 11" having the junction 5', 5" with junction depth X' j , X'' j.
0コメント