課程名稱︰分析化學三
課程性質︰化學系五選三必修
課程教師︰何國榮
開課學院:理學院
開課系所︰化學系
考試日期(年月日)︰2013.11.08
考試時限(分鐘):120mins
是否需發放獎勵金:是
(如未明確表示,則不予發放)
試題 :
下面灰色註記的字體是依最後發批改完的考卷個人認為
&助教解釋重要的答題關鍵,不保證完全正確,僅供參考
1. Explain the standard deviation of a calibration curve can be calculated
although only one analysis is performed for each standard solution.
(6 points)
假設各個conc 的soln 其standard deviation相同,所以可以只測各個
soln一次來做calibration curve.
signal│ ╱ (真正考卷上我一共畫了5個€,代表distribution)
│ €
│ ╱
│ €
│ ╱
│ €
│╱
└────────conc
2. Explain that error is an idealized concept and errors can not be known
exactly.(6 points)
因為實際上我們並不知道"真值",所以無真正求得實驗result與真值
之間的差異
3. Describe the conditoin that accuracy can not be used to estimate systemic
error of a method. (6 points)
accuracy 包含random/ systemic error, 測量次數不多時,random error
仍暫依定的比例,此時無法以accuracy來估計systemetic error
_ _
4. Describe the condition of using the equation μ=X±zσ/√n (6 points)
表示再測量次數有限下,但該方法的標準差是貞前經許多次量測所知道的,
故可在公式中使用σ&z
5. Explain that the between sample mean square can not be used to estimate
σ(下標factor)^2 directly. (6 points)
SSB除包含controled factor所造成的σ(factoe)外,也包含了random error的因素,
所以無法直接以SSB估計σ(factor)^2
6. Define the limit of detection with 5% α error (n=7) (6 points)
D.F.=6時,某方法可襙到SIGNAL的最低濃度,且有1-5%的信心水準,(5%可能是偽陽)
7. For a atest of M.D.L. of 10ppb, explain that not detected can not be
interpreted as less than 10 ppb. (6 points)
講義上有圖,在M.D.L.左邊會被判定為not detected,此時11ppm的
sample仍有分布,所以不可以用"not detected"來宣稱conc<10
8. List and describe the two major components of quality assuarance.
(6 points)
QA (quality assessment):用以評估數據是否在合理範圍內的方法
QC (quality control):用以確保釋數據在合理範圍內的步驟(S.O.P)
9. Describe the relation between accuracy, truness, precision and type
of error. (6 points)
Accuracy┌random error →precision
│
└systemetic error →truness/bias
Accuracy: 準確度,測得result與reference value間clossness的程度,包和
random/systemetic error兩部分(precision與truness)
Precision:測得results間彼此clossness的程度,表示random error
Truness: 許多次測量results與reference value間clossness的程度,表示
systemietic error
當high precision, high truness時,則有好的accuravy
10. Explain the contribution of Fourier Transform to 13(上標)C NMR.
(6 points)
13C在天然物中含量不高,NMR訊號不強,若以FT在短時間內測量許多次,再將
結果相加轉為frequency domain,即可得到較好的S/N ratio
11. Explain that the sample is often dissolved in CDCl3 or D2O in NMR
experiment. (6 points)
a.若是H而不是D,則會造成1H-NMR爆表的現象
b.D solvent有locking的作用,因為其δ已知,當所施加磁場改變時,我們會發現
D solvent的δ也會shift,可做及時的校正以lock祝磁場
12. Explain that NMR of higher magnetic field provide better sensitivity
and better resolution. (6 points)
a.較高B中能階的splitting較大
b.H間的coupling 與B無關
13. Explain the Hubaux-Vos procedure is often not a good method to estimate
L.O.D. (6 points)
因為standard deviation通常和conc相關,所以如果假設兩者獨立,以較高conc
時得到的σ來估計LOD則會造成誤差(LOD的σ會被高估)
14. Explain the purpose of performing F test in the estimation of M.D.L
(6 points)
這題如果只把流程寫出來大概只會得一分,助教說眾覑是要寫出假設在偵測極限
附近standard deviaiotn相同,所以才用F test來看影無差異
15. Explain that the triangular instead of rectangular distribution is
used to estimate the standard uncertainty of volumetric glassware.
(6 points)
因為glassware volume的真正體積應該在其所標示的值得附近,亦即愈接近
x bar,其分布應愈高
16. List the three major factors of data quality. (6 points)
representability:代表性
indentity:定性
accuracy:確定性→定量
17. Explain that the variance can be used to compare means more than two
population. (6 points)
如果多個populations 的variance相近,其背後意義是他們可能來自同一個
更大的集合,意即他們可能有相近的mean,因此可用variance來比較多個
population mean是否相近
--
推 jane820113:ji3
噓 edda20047:你是餓到不會切換輸入法了齁林呱呱
→ jane820113:XDDDDDD 謝化咦好兇
噓 ruby40861:還錯字,版主塊水桶她
→ ruby40861:啊我也錯字QQQQQQ
--
※ 發信站: 批踢踢實業坊(ptt.cc)
◆ From: 180.177.66.148
※ 編輯: ruby40861 來自: 180.177.66.148 (01/11 13:52)