1. Define Problem Select Appropriate Methods Obtain and store sample Pre-treat sample Perform required measurements Compare results with standards Apply necessary statistical methods Present results in a format understandable to analyst Present results to customer

2. How do you decide the best method for your problem?

3. Factors of the method which will influence method choice 1. Precision: Provides a measure of random error in a measurement Precision vs. accuracy

4. Accuracy and Precision  Accuracy determined by factors including calibration, lab technique, sample storage, collection, etc.  Precision limited by noise alone  Measured by:  Std deviation  Variance = s 2  Coefficient of variance CV = s/x * 100  Relative std. dev.  RSD = s/x

5. Factors of the method which will influence method choice 2. Bias: systematic error How do you know when you have bias? Can come from: Instrumental errors Personal errors Method errors

6. 3. Sensitivity: Which metal method is most sensitive?

7. 4. Detection limit Determines whether or not a signal has been detected and the confidence such a decision can be made

8. Detection limits and noise  Noise often dictates how low a detection limit will be  Detection limit at least 3x std deviation of blank can be as high as 100X Which one has the lower detection limit?

9. 5. Concentration range  Working range where calibration curve is linear

10. 6. Selectivity  Extent to which method is free from interferences Which method is better for drug testing, Mass Spec or TLC?

11. More about Noise 1.Noise is everywhere 2.Noise is usually constant in magnitude 3.Signal to noise ratio is like GPA (figure or merit) See transparency

12. What causes noise? 1. Chemical noise: Due to undetected (to the scientist) changes in the chemistry of the reaction. Temp, glassware, fumes, etc.

13. 2. Instrumental Noise a. Johnson or thermal noise: Due to movement of electrons or other charge carriers Also called white noise b/c occurs at all frequencies Where: rm s = root mean square noise k = Boltzman constant T = absolute temp R = resistance  f = instrument’s bandwidth How would you minimize?

14. b. Shot noise (much smaller than Thermal noise)  Due to individual charge carriers moving across n  p junction or arriving at electrode surface i av = shot noise I = intensity of signal e = charge on e -  f = band width

15. c. Flicker noise Observed at low frequencies (< 100Hz)

16. d. Environmental Noise  See transparency  Most common are electric and magnetic fields @ 60 Hz and harmonics, 120, 180, 240, etc.  Proper grounding of instruments and transmission will lower noise

17. How can you enhance your signal relative to your noise?  Hardware 1. Low pass filters